/*
* Allocate a new page (either by recycling, or by extending the index file).
*
* The returned buffer is already pinned and exclusive-locked.
* Caller is responsible for initializing the page by calling SpGistInitBuffer.
*/
Buffer
SpGistNewBuffer(Relation index)
{
Buffer buffer;
bool needLock;
/* First, try to get a page from FSM */
for (;;)
{
BlockNumber blkno = GetFreeIndexPage(index);
if (blkno == InvalidBlockNumber)
break; /* nothing known to FSM */
/*
* The fixed pages shouldn't ever be listed in FSM, but just in case
* one is, ignore it.
*/
if (SpGistBlockIsFixed(blkno))
continue;
buffer = ReadBuffer(index, blkno);
/*
* We have to guard against the possibility that someone else already
* recycled this page; the buffer may be locked if so.
*/
if (ConditionalLockBuffer(buffer))
{
Page page = BufferGetPage(buffer);
if (PageIsNew(page))
return buffer; /* OK to use, if never initialized */
if (SpGistPageIsDeleted(page) || PageIsEmpty(page))
return buffer; /* OK to use */
LockBuffer(buffer, BUFFER_LOCK_UNLOCK);
}
/* Can't use it, so release buffer and try again */
ReleaseBuffer(buffer);
}
/* Must extend the file */
needLock = !RELATION_IS_LOCAL(index);
if (needLock)
LockRelationForExtension(index, ExclusiveLock);
buffer = ReadBuffer(index, P_NEW);
LockBuffer(buffer, BUFFER_LOCK_EXCLUSIVE);
if (needLock)
UnlockRelationForExtension(index, ExclusiveLock);
return buffer;
}
/*
* Allocate a new page (either by recycling, or by extending the index file)
*
* The returned buffer is already pinned and exclusive-locked
*
* Caller is responsible for initializing the page by calling GISTInitBuffer
*/
Buffer
gistNewBuffer(Relation r)
{
Buffer buffer;
bool needLock;
MIRROREDLOCK_BUFMGR_MUST_ALREADY_BE_HELD;
/* First, try to get a page from FSM */
for (;;)
{
BlockNumber blkno = GetFreeIndexPage(&r->rd_node);
if (blkno == InvalidBlockNumber)
break; /* nothing left in FSM */
buffer = ReadBuffer(r, blkno);
/*
* We have to guard against the possibility that someone else already
* recycled this page; the buffer may be locked if so.
*/
if (ConditionalLockBuffer(buffer))
{
Page page = BufferGetPage(buffer);
if (PageIsNew(page))
return buffer; /* OK to use, if never initialized */
gistcheckpage(r, buffer);
if (GistPageIsDeleted(page))
return buffer; /* OK to use */
LockBuffer(buffer, GIST_UNLOCK);
}
/* Can't use it, so release buffer and try again */
ReleaseBuffer(buffer);
}
/* Must extend the file */
needLock = !RELATION_IS_LOCAL(r);
if (needLock)
LockRelationForExtension(r, ExclusiveLock);
buffer = ReadBuffer(r, P_NEW);
LockBuffer(buffer, GIST_EXCLUSIVE);
if (needLock)
UnlockRelationForExtension(r, ExclusiveLock);
return buffer;
}
Buffer
GinNewBuffer(Relation index)
{
Buffer buffer;
bool needLock;
/* First, try to get a page from FSM */
for (;;)
{
BlockNumber blkno = GetFreeIndexPage(index);
if (blkno == InvalidBlockNumber)
break;
buffer = ReadBuffer(index, blkno);
/*
* We have to guard against the possibility that someone else already
* recycled this page; the buffer may be locked if so.
*/
if (ConditionalLockBuffer(buffer))
{
Page page = BufferGetPage(buffer);
if (PageIsNew(page))
return buffer; /* OK to use, if never initialized */
if (GinPageIsDeleted(page))
return buffer; /* OK to use */
LockBuffer(buffer, GIN_UNLOCK);
}
/* Can't use it, so release buffer and try again */
ReleaseBuffer(buffer);
}
/* Must extend the file */
needLock = !RELATION_IS_LOCAL(index);
if (needLock)
LockRelationForExtension(index, ExclusiveLock);
buffer = ReadBuffer(index, P_NEW);
LockBuffer(buffer, GIN_EXCLUSIVE);
if (needLock)
UnlockRelationForExtension(index, ExclusiveLock);
return buffer;
}
/*
* _bitmap_getbuf() -- return the buffer for the given block number and
* the access method.
*/
Buffer
_bitmap_getbuf(Relation rel, BlockNumber blkno, int access)
{
Buffer buf;
MIRROREDLOCK_BUFMGR_MUST_ALREADY_BE_HELD;
if (blkno != P_NEW)
{
buf = ReadBuffer(rel, blkno);
if (access != BM_NOLOCK)
LockBuffer(buf, access);
}
else
{
bool needLock;
Assert(access == BM_WRITE);
/*
* Extend the relation by one page.
*
* We have to use a lock to ensure no one else is extending the rel at
* the same time, else we will both try to initialize the same new
* page. We can skip locking for new or temp relations, however,
* since no one else could be accessing them.
*/
needLock = !RELATION_IS_LOCAL(rel);
if (needLock)
LockRelationForExtension(rel, ExclusiveLock);
buf = ReadBuffer(rel, P_NEW);
/* Acquire buffer lock on new page */
LockBuffer(buf, BM_WRITE);
/*
* Release the file-extension lock; it's now OK for someone else to
* extend the relation some more.
*/
if (needLock)
UnlockRelationForExtension(rel, ExclusiveLock);
}
return buf;
}
/*
* _bt_mergeload - Merge two streams of index tuples into new index files.
*/
static void
_bt_mergeload(Spooler *self, BTWriteState *wstate, BTSpool *btspool, BTReader *btspool2, Relation heapRel)
{
BTPageState *state = NULL;
IndexTuple itup,
itup2;
bool should_free = false;
TupleDesc tupdes = RelationGetDescr(wstate->index);
int keysz = RelationGetNumberOfAttributes(wstate->index);
ScanKey indexScanKey;
ON_DUPLICATE on_duplicate = self->on_duplicate;
Assert(btspool != NULL);
/* the preparation of merge */
itup = BTSpoolGetNextItem(btspool, NULL, &should_free);
itup2 = BTReaderGetNextItem(btspool2);
indexScanKey = _bt_mkscankey_nodata(wstate->index);
for (;;)
{
bool load1 = true; /* load BTSpool next ? */
bool hasnull;
int32 compare;
if (self->dup_old + self->dup_new > self->max_dup_errors)
ereport(ERROR,
(errcode(ERRCODE_INTERNAL_ERROR),
errmsg("Maximum duplicate error count exceeded")));
if (itup2 == NULL)
{
if (itup == NULL)
break;
}
else if (itup != NULL)
{
compare = compare_indextuple(itup, itup2, indexScanKey,
keysz, tupdes, &hasnull);
if (compare == 0 && !hasnull && btspool->isunique)
{
ItemPointerData t_tid2;
/*
* t_tid is update by heap_is_visible(), because use it for an
* index, t_tid backup
*/
ItemPointerCopy(&itup2->t_tid, &t_tid2);
/* The tuple pointed by the old index should not be visible. */
if (!heap_is_visible(heapRel, &itup->t_tid))
{
itup = BTSpoolGetNextItem(btspool, itup, &should_free);
}
else if (!heap_is_visible(heapRel, &itup2->t_tid))
{
itup2 = BTReaderGetNextItem(btspool2);
}
else
{
if (on_duplicate == ON_DUPLICATE_KEEP_NEW)
{
self->dup_old++;
remove_duplicate(self, heapRel, itup2,
RelationGetRelationName(wstate->index));
itup2 = BTReaderGetNextItem(btspool2);
}
else
{
ItemPointerCopy(&t_tid2, &itup2->t_tid);
self->dup_new++;
remove_duplicate(self, heapRel, itup,
RelationGetRelationName(wstate->index));
itup = BTSpoolGetNextItem(btspool, itup, &should_free);
}
}
continue;
}
else if (compare > 0)
load1 = false;
}
else
load1 = false;
BULKLOAD_PROFILE(&prof_merge_unique);
/* When we see first tuple, create first index page */
if (state == NULL)
state = _bt_pagestate(wstate, 0);
if (load1)
{
IndexTuple next_itup = NULL;
bool next_should_free = false;
for (;;)
{
/* get next item */
next_itup = BTSpoolGetNextItem(btspool, next_itup,
&next_should_free);
if (!btspool->isunique || next_itup == NULL)
break;
compare = compare_indextuple(itup, next_itup, indexScanKey,
keysz, tupdes, &hasnull);
if (compare < 0 || hasnull)
break;
if (compare > 0)
{
/* shouldn't happen */
elog(ERROR, "faild in tuplesort_performsort");
}
/*
* If tupple is deleted by other unique indexes, not visible
*/
if (!heap_is_visible(heapRel, &next_itup->t_tid))
{
continue;
}
if (!heap_is_visible(heapRel, &itup->t_tid))
{
if (should_free)
pfree(itup);
itup = next_itup;
should_free = next_should_free;
next_should_free = false;
continue;
}
/* not unique between input files */
self->dup_new++;
remove_duplicate(self, heapRel, next_itup,
RelationGetRelationName(wstate->index));
if (self->dup_old + self->dup_new > self->max_dup_errors)
ereport(ERROR,
(errcode(ERRCODE_INTERNAL_ERROR),
errmsg("Maximum duplicate error count exceeded")));
}
_bt_buildadd(wstate, state, itup);
if (should_free)
pfree(itup);
itup = next_itup;
should_free = next_should_free;
}
else
{
_bt_buildadd(wstate, state, itup2);
itup2 = BTReaderGetNextItem(btspool2);
}
BULKLOAD_PROFILE(&prof_merge_insert);
}
_bt_freeskey(indexScanKey);
/* Close down final pages and write the metapage */
_bt_uppershutdown(wstate, state);
/*
* If the index isn't temp, we must fsync it down to disk before it's safe
* to commit the transaction. (For a temp index we don't care since the
* index will be uninteresting after a crash anyway.)
*
* It's obvious that we must do this when not WAL-logging the build. It's
* less obvious that we have to do it even if we did WAL-log the index
* pages. The reason is that since we're building outside shared buffers,
* a CHECKPOINT occurring during the build has no way to flush the
* previously written data to disk (indeed it won't know the index even
* exists). A crash later on would replay WAL from the checkpoint,
* therefore it wouldn't replay our earlier WAL entries. If we do not
* fsync those pages here, they might still not be on disk when the crash
* occurs.
*/
if (!RELATION_IS_LOCAL(wstate->index))
{
RelationOpenSmgr(wstate->index);
smgrimmedsync(wstate->index->rd_smgr, MAIN_FORKNUM);
}
BULKLOAD_PROFILE(&prof_merge_term);
}
static void
_bt_mergebuild(Spooler *self, BTSpool *btspool)
{
Relation heapRel = self->relinfo->ri_RelationDesc;
BTWriteState wstate;
BTReader reader;
bool merge;
Assert(btspool->index->rd_index->indisvalid);
tuplesort_performsort(btspool->sortstate);
wstate.index = btspool->index;
/*
* We need to log index creation in WAL iff WAL archiving is enabled AND
* it's not a temp index.
*/
wstate.btws_use_wal = self->use_wal &&
XLogArchivingActive() && !RELATION_IS_LOCAL(wstate.index);
/* reserve the metapage */
wstate.btws_pages_alloced = BTREE_METAPAGE + 1;
wstate.btws_pages_written = 0;
wstate.btws_zeropage = NULL; /* until needed */
/*
* Flush dirty buffers so that we will read the index files directly
* in order to get pre-existing data. We must acquire AccessExclusiveLock
* for the target table for calling FlushRelationBuffer().
*/
LockRelation(wstate.index, AccessExclusiveLock);
FlushRelationBuffers(wstate.index);
BULKLOAD_PROFILE(&prof_flush);
merge = BTReaderInit(&reader, wstate.index);
elog(DEBUG1, "pg_bulkload: build \"%s\" %s merge (%s wal)",
RelationGetRelationName(wstate.index),
merge ? "with" : "without",
wstate.btws_use_wal ? "with" : "without");
/* Assign a new file node. */
RelationSetNewRelfilenode(wstate.index, InvalidTransactionId);
if (merge || (btspool->isunique && self->max_dup_errors > 0))
{
/* Merge two streams into the new file node that we assigned. */
BULKLOAD_PROFILE_PUSH();
_bt_mergeload(self, &wstate, btspool, &reader, heapRel);
BULKLOAD_PROFILE_POP();
BULKLOAD_PROFILE(&prof_merge);
}
else
{
/* Fast path for newly created index. */
_bt_load(&wstate, btspool, NULL);
BULKLOAD_PROFILE(&prof_index);
}
BTReaderTerm(&reader);
}
/*
* VACUUM cleanup: update FSM
*/
IndexBulkDeleteResult *
gistvacuumcleanup(IndexVacuumInfo *info, IndexBulkDeleteResult *stats)
{
Relation rel = info->index;
BlockNumber npages,
blkno;
BlockNumber totFreePages;
bool needLock;
/* No-op in ANALYZE ONLY mode */
if (info->analyze_only)
return stats;
/* Set up all-zero stats if gistbulkdelete wasn't called */
if (stats == NULL)
{
stats = (IndexBulkDeleteResult *) palloc0(sizeof(IndexBulkDeleteResult));
/* use heap's tuple count */
stats->num_index_tuples = info->num_heap_tuples;
stats->estimated_count = info->estimated_count;
/*
* XXX the above is wrong if index is partial. Would it be OK to just
* return NULL, or is there work we must do below?
*/
}
/*
* Need lock unless it's local to this backend.
*/
needLock = !RELATION_IS_LOCAL(rel);
/* try to find deleted pages */
if (needLock)
LockRelationForExtension(rel, ExclusiveLock);
npages = RelationGetNumberOfBlocks(rel);
if (needLock)
UnlockRelationForExtension(rel, ExclusiveLock);
totFreePages = 0;
for (blkno = GIST_ROOT_BLKNO + 1; blkno < npages; blkno++)
{
Buffer buffer;
Page page;
vacuum_delay_point();
buffer = ReadBufferExtended(rel, MAIN_FORKNUM, blkno, RBM_NORMAL,
info->strategy);
LockBuffer(buffer, GIST_SHARE);
page = (Page) BufferGetPage(buffer);
if (PageIsNew(page) || GistPageIsDeleted(page))
{
totFreePages++;
RecordFreeIndexPage(rel, blkno);
}
UnlockReleaseBuffer(buffer);
}
/* Finally, vacuum the FSM */
IndexFreeSpaceMapVacuum(info->index);
/* return statistics */
stats->pages_free = totFreePages;
if (needLock)
LockRelationForExtension(rel, ExclusiveLock);
stats->num_pages = RelationGetNumberOfBlocks(rel);
if (needLock)
UnlockRelationForExtension(rel, ExclusiveLock);
return stats;
}
/*
* Try to extend the revmap by one page. This might not happen for a number of
* reasons; caller is expected to retry until the expected outcome is obtained.
*/
static void
revmap_physical_extend(BrinRevmap *revmap)
{
Buffer buf;
Page page;
Page metapage;
BrinMetaPageData *metadata;
BlockNumber mapBlk;
BlockNumber nblocks;
Relation irel = revmap->rm_irel;
bool needLock = !RELATION_IS_LOCAL(irel);
/*
* Lock the metapage. This locks out concurrent extensions of the revmap,
* but note that we still need to grab the relation extension lock because
* another backend can extend the index with regular BRIN pages.
*/
LockBuffer(revmap->rm_metaBuf, BUFFER_LOCK_EXCLUSIVE);
metapage = BufferGetPage(revmap->rm_metaBuf);
metadata = (BrinMetaPageData *) PageGetContents(metapage);
/*
* Check that our cached lastRevmapPage value was up-to-date; if it
* wasn't, update the cached copy and have caller start over.
*/
if (metadata->lastRevmapPage != revmap->rm_lastRevmapPage)
{
revmap->rm_lastRevmapPage = metadata->lastRevmapPage;
LockBuffer(revmap->rm_metaBuf, BUFFER_LOCK_UNLOCK);
return;
}
mapBlk = metadata->lastRevmapPage + 1;
nblocks = RelationGetNumberOfBlocks(irel);
if (mapBlk < nblocks)
{
buf = ReadBuffer(irel, mapBlk);
LockBuffer(buf, BUFFER_LOCK_EXCLUSIVE);
page = BufferGetPage(buf);
}
else
{
if (needLock)
LockRelationForExtension(irel, ExclusiveLock);
buf = ReadBuffer(irel, P_NEW);
if (BufferGetBlockNumber(buf) != mapBlk)
{
/*
* Very rare corner case: somebody extended the relation
* concurrently after we read its length. If this happens, give
* up and have caller start over. We will have to evacuate that
* page from under whoever is using it.
*/
if (needLock)
UnlockRelationForExtension(irel, ExclusiveLock);
LockBuffer(revmap->rm_metaBuf, BUFFER_LOCK_UNLOCK);
return;
}
LockBuffer(buf, BUFFER_LOCK_EXCLUSIVE);
page = BufferGetPage(buf);
if (needLock)
UnlockRelationForExtension(irel, ExclusiveLock);
}
/* Check that it's a regular block (or an empty page) */
if (!PageIsNew(page) && !BRIN_IS_REGULAR_PAGE(page))
ereport(ERROR,
(errcode(ERRCODE_INDEX_CORRUPTED),
errmsg("unexpected page type 0x%04X in BRIN index \"%s\" block %u",
BRIN_PAGE_TYPE(page),
RelationGetRelationName(irel),
BufferGetBlockNumber(buf))));
/* If the page is in use, evacuate it and restart */
if (brin_start_evacuating_page(irel, buf))
{
LockBuffer(revmap->rm_metaBuf, BUFFER_LOCK_UNLOCK);
brin_evacuate_page(irel, revmap->rm_pagesPerRange, revmap, buf);
/* have caller start over */
return;
}
/*
* Ok, we have now locked the metapage and the target block. Re-initialize
* it as a revmap page.
*/
START_CRIT_SECTION();
/* the rm_tids array is initialized to all invalid by PageInit */
brin_page_init(page, BRIN_PAGETYPE_REVMAP);
MarkBufferDirty(buf);
metadata->lastRevmapPage = mapBlk;
MarkBufferDirty(revmap->rm_metaBuf);
if (RelationNeedsWAL(revmap->rm_irel))
{
xl_brin_revmap_extend xlrec;
XLogRecPtr recptr;
XLogRecData rdata[2];
xlrec.node = revmap->rm_irel->rd_node;
xlrec.targetBlk = mapBlk;
rdata[0].data = (char *) &xlrec;
rdata[0].len = SizeOfBrinRevmapExtend;
rdata[0].buffer = InvalidBuffer;
rdata[0].buffer_std = false;
rdata[0].next = &(rdata[1]);
rdata[1].data = (char *) NULL;
rdata[1].len = 0;
rdata[1].buffer = revmap->rm_metaBuf;
rdata[1].buffer_std = false;
rdata[1].next = NULL;
recptr = XLogInsert(RM_BRIN_ID, XLOG_BRIN_REVMAP_EXTEND, rdata);
PageSetLSN(metapage, recptr);
PageSetLSN(page, recptr);
}
END_CRIT_SECTION();
LockBuffer(revmap->rm_metaBuf, BUFFER_LOCK_UNLOCK);
UnlockReleaseBuffer(buf);
}
Datum
gistvacuumcleanup(PG_FUNCTION_ARGS)
{
IndexVacuumInfo *info = (IndexVacuumInfo *) PG_GETARG_POINTER(0);
GistBulkDeleteResult *stats = (GistBulkDeleteResult *) PG_GETARG_POINTER(1);
Relation rel = info->index;
BlockNumber npages,
blkno;
BlockNumber totFreePages,
nFreePages,
*freePages,
maxFreePages;
BlockNumber lastBlock = GIST_ROOT_BLKNO,
lastFilledBlock = GIST_ROOT_BLKNO;
bool needLock;
/* Set up all-zero stats if gistbulkdelete wasn't called */
if (stats == NULL)
{
stats = (GistBulkDeleteResult *) palloc0(sizeof(GistBulkDeleteResult));
/* use heap's tuple count */
Assert(info->num_heap_tuples >= 0);
stats->std.num_index_tuples = info->num_heap_tuples;
/*
* XXX the above is wrong if index is partial. Would it be OK to just
* return NULL, or is there work we must do below?
*/
}
/* gistVacuumUpdate may cause hard work */
if (info->vacuum_full)
{
GistVacuum gv;
ArrayTuple res;
/* note: vacuum.c already acquired AccessExclusiveLock on index */
gv.index = rel;
initGISTstate(&(gv.giststate), rel);
gv.opCtx = createTempGistContext();
gv.result = stats;
gv.strategy = info->strategy;
/* walk through the entire index for update tuples */
res = gistVacuumUpdate(&gv, GIST_ROOT_BLKNO, false);
/* cleanup */
if (res.itup)
{
int i;
for (i = 0; i < res.ituplen; i++)
pfree(res.itup[i]);
pfree(res.itup);
}
freeGISTstate(&(gv.giststate));
MemoryContextDelete(gv.opCtx);
}
else if (stats->needFullVacuum)
ereport(NOTICE,
(errmsg("index \"%s\" needs VACUUM FULL or REINDEX to finish crash recovery",
RelationGetRelationName(rel))));
/*
* If vacuum full, we already have exclusive lock on the index. Otherwise,
* need lock unless it's local to this backend.
*/
if (info->vacuum_full)
needLock = false;
else
needLock = !RELATION_IS_LOCAL(rel);
/* try to find deleted pages */
if (needLock)
LockRelationForExtension(rel, ExclusiveLock);
npages = RelationGetNumberOfBlocks(rel);
if (needLock)
UnlockRelationForExtension(rel, ExclusiveLock);
maxFreePages = npages;
if (maxFreePages > MaxFSMPages)
maxFreePages = MaxFSMPages;
totFreePages = nFreePages = 0;
freePages = (BlockNumber *) palloc(sizeof(BlockNumber) * maxFreePages);
for (blkno = GIST_ROOT_BLKNO + 1; blkno < npages; blkno++)
{
Buffer buffer;
Page page;
vacuum_delay_point();
buffer = ReadBufferWithStrategy(rel, blkno, info->strategy);
LockBuffer(buffer, GIST_SHARE);
page = (Page) BufferGetPage(buffer);
if (PageIsNew(page) || GistPageIsDeleted(page))
{
if (nFreePages < maxFreePages)
freePages[nFreePages++] = blkno;
totFreePages++;
}
else
lastFilledBlock = blkno;
UnlockReleaseBuffer(buffer);
}
lastBlock = npages - 1;
if (info->vacuum_full && nFreePages > 0)
{ /* try to truncate index */
int i;
for (i = 0; i < nFreePages; i++)
if (freePages[i] >= lastFilledBlock)
{
totFreePages = nFreePages = i;
break;
}
if (lastBlock > lastFilledBlock)
RelationTruncate(rel, lastFilledBlock + 1);
stats->std.pages_removed = lastBlock - lastFilledBlock;
}
RecordIndexFreeSpace(&rel->rd_node, totFreePages, nFreePages, freePages);
pfree(freePages);
/* return statistics */
stats->std.pages_free = totFreePages;
if (needLock)
LockRelationForExtension(rel, ExclusiveLock);
stats->std.num_pages = RelationGetNumberOfBlocks(rel);
if (needLock)
UnlockRelationForExtension(rel, ExclusiveLock);
PG_RETURN_POINTER(stats);
}
/**
* @brief Write block buffer contents. Number of block buffer to be
* written is specified by num argument.
*
* Flow:
* <ol>
* <li>If no more space is available in the data file, switch to a new one.</li>
* <li>Compute block number which can be written to the current file.</li>
* <li>Save the last block number in the load status file.</li>
* <li>Write to the current file.</li>
* <li>If there are other data, write them too.</li>
* </ol>
*
* @param loader [in] Direct Writer.
* @return File descriptor for the current data file.
*/
static void
flush_pages(DirectWriter *loader)
{
int i;
int num;
LoadStatus *ls = &loader->ls;
num = loader->curblk;
if (!PageIsEmpty(GetCurrentPage(loader)))
num += 1;
if (num <= 0)
return; /* no work */
/*
* Add WAL entry (only the first page) to ensure the current xid will
* be recorded in xlog. We must flush some xlog records with XLogFlush()
* before write any data blocks to follow the WAL protocol.
*
* If postgres process, such as loader and COPY, is killed by "kill -9",
* database will be rewound to the last checkpoint and recovery will
* be performed using WAL.
*
* After the recovery, if there are xid's which have not been recorded
* to WAL, such xid's will be reused.
*
* However, in the loader and COPY, data file is actually updated and
* xid must not be reused.
*
* WAL entry with such xid can be added using XLogInsert(). However,
* such entries are not really written to the disk immediately.
* WAL entries are flushed to the disk by XLogFlush(), typically
* when a transaction is commited. COPY prevents xid reuse by
* this method.
*/
#if PG_VERSION_NUM >= 90100
if (ls->ls.create_cnt == 0 && !RELATION_IS_LOCAL(loader->base.rel)
&& !(loader->base.rel->rd_rel->relpersistence == RELPERSISTENCE_UNLOGGED) )
{
XLogRecPtr recptr;
recptr = log_newpage(&ls->ls.rnode, MAIN_FORKNUM,
ls->ls.exist_cnt, loader->blocks);
XLogFlush(recptr);
}
#else
if (ls->ls.create_cnt == 0 && !RELATION_IS_LOCAL(loader->base.rel) )
{
XLogRecPtr recptr;
recptr = log_newpage(&ls->ls.rnode, MAIN_FORKNUM,
ls->ls.exist_cnt, loader->blocks);
XLogFlush(recptr);
}
#endif
/*
* Write blocks. We might need to write multiple files on boundary of
* relation segments.
*/
for (i = 0; i < num;)
{
char *buffer;
int total;
int written;
int flush_num;
BlockNumber relblks = LS_TOTAL_CNT(ls);
/* Switch to the next file if the current file has been filled up. */
if (relblks % RELSEG_SIZE == 0)
close_data_file(loader);
if (loader->datafd == -1)
loader->datafd = open_data_file(ls->ls.rnode,
RELATION_IS_LOCAL(loader->base.rel),
relblks);
/* Number of blocks to be added to the current file. */
flush_num = Min(num - i, RELSEG_SIZE - relblks % RELSEG_SIZE);
Assert(flush_num > 0);
/* Write the last block number to the load status file. */
UpdateLSF(loader, flush_num);
#if PG_VERSION_NUM >= 90300
/* If we need a checksum, add it */
if (DataChecksumsEnabled()){
int j = 0;
Page contained_page;
for ( j=0; j<flush_num; j++ ) {
contained_page = GetTargetPage(loader,j);
((PageHeader) contained_page)->pd_checksum =
pg_checksum_page((char *) contained_page, LS_TOTAL_CNT(ls) - 1 - j);
}
}
#endif
/*
* Flush flush_num data block to the current file.
* Then the current file size becomes RELSEG_SIZE self->blocks.
*/
buffer = loader->blocks + BLCKSZ * i;
total = BLCKSZ * flush_num;
written = 0;
while (total > 0)
{
int len = write(loader->datafd, buffer + written, total);
if (len == -1)
{
/* fatal error, do not want to write blocks anymore */
ereport(ERROR, (errcode_for_file_access(),
errmsg("could not write to data file: %m")));
}
written += len;
total -= len;
}
i += flush_num;
}
/*
* NOTICE: Be sure reset curblk to 0 and reinitialize recycled page
* if you will continue to use blocks.
*/
}
/*
* btvacuumscan --- scan the index for VACUUMing purposes
*
* This combines the functions of looking for leaf tuples that are deletable
* according to the vacuum callback, looking for empty pages that can be
* deleted, and looking for old deleted pages that can be recycled. Both
* btbulkdelete and btvacuumcleanup invoke this (the latter only if no
* btbulkdelete call occurred).
*
* The caller is responsible for initially allocating/zeroing a stats struct
* and for obtaining a vacuum cycle ID if necessary.
*/
static void
btvacuumscan(IndexVacuumInfo *info, IndexBulkDeleteResult *stats,
IndexBulkDeleteCallback callback, void *callback_state,
BTCycleId cycleid)
{
MIRROREDLOCK_BUFMGR_VERIFY_NO_LOCK_LEAK_DECLARE;
Relation rel = info->index;
BTVacState vstate;
BlockNumber num_pages;
BlockNumber blkno;
bool needLock;
MIRROREDLOCK_BUFMGR_VERIFY_NO_LOCK_LEAK_ENTER;
/*
* Reset counts that will be incremented during the scan; needed in case
* of multiple scans during a single VACUUM command
*/
stats->num_index_tuples = 0;
stats->pages_deleted = 0;
/* Set up info to pass down to btvacuumpage */
vstate.info = info;
vstate.stats = stats;
vstate.callback = callback;
vstate.callback_state = callback_state;
vstate.cycleid = cycleid;
vstate.freePages = NULL; /* temporarily */
vstate.nFreePages = 0;
vstate.maxFreePages = 0;
vstate.totFreePages = 0;
/* Create a temporary memory context to run _bt_pagedel in */
vstate.pagedelcontext = AllocSetContextCreate(CurrentMemoryContext,
"_bt_pagedel",
ALLOCSET_DEFAULT_MINSIZE,
ALLOCSET_DEFAULT_INITSIZE,
ALLOCSET_DEFAULT_MAXSIZE);
/*
* The outer loop iterates over all index pages except the metapage, in
* physical order (we hope the kernel will cooperate in providing
* read-ahead for speed). It is critical that we visit all leaf pages,
* including ones added after we start the scan, else we might fail to
* delete some deletable tuples. Hence, we must repeatedly check the
* relation length. We must acquire the relation-extension lock while
* doing so to avoid a race condition: if someone else is extending the
* relation, there is a window where bufmgr/smgr have created a new
* all-zero page but it hasn't yet been write-locked by _bt_getbuf(). If
* we manage to scan such a page here, we'll improperly assume it can be
* recycled. Taking the lock synchronizes things enough to prevent a
* problem: either num_pages won't include the new page, or _bt_getbuf
* already has write lock on the buffer and it will be fully initialized
* before we can examine it. (See also vacuumlazy.c, which has the same
* issue.) Also, we need not worry if a page is added immediately after
* we look; the page splitting code already has write-lock on the left
* page before it adds a right page, so we must already have processed any
* tuples due to be moved into such a page.
*
* We can skip locking for new or temp relations, however, since no one
* else could be accessing them.
*/
needLock = !RELATION_IS_LOCAL(rel);
blkno = BTREE_METAPAGE + 1;
for (;;)
{
/* Get the current relation length */
if (needLock)
LockRelationForExtension(rel, ExclusiveLock);
num_pages = RelationGetNumberOfBlocks(rel);
if (needLock)
UnlockRelationForExtension(rel, ExclusiveLock);
/* Allocate freePages after we read num_pages the first time */
if (vstate.freePages == NULL)
{
/* No point in remembering more than MaxFSMPages pages */
vstate.maxFreePages = MaxFSMPages;
if ((BlockNumber) vstate.maxFreePages > num_pages)
vstate.maxFreePages = (int) num_pages;
vstate.freePages = (BlockNumber *)
palloc(vstate.maxFreePages * sizeof(BlockNumber));
}
/* Quit if we've scanned the whole relation */
if (blkno >= num_pages)
break;
/* Iterate over pages, then loop back to recheck length */
for (; blkno < num_pages; blkno++)
{
btvacuumpage(&vstate, blkno, blkno);
}
}
/*
* During VACUUM FULL, we truncate off any recyclable pages at the end of
* the index. In a normal vacuum it'd be unsafe to do this except by
* acquiring exclusive lock on the index and then rechecking all the
* pages; doesn't seem worth it.
*/
if (info->vacuum_full && vstate.nFreePages > 0)
{
BlockNumber new_pages = num_pages;
while (vstate.nFreePages > 0 &&
vstate.freePages[vstate.nFreePages - 1] == new_pages - 1)
{
new_pages--;
stats->pages_deleted--;
vstate.nFreePages--;
vstate.totFreePages = vstate.nFreePages; /* can't be more */
}
if (new_pages != num_pages)
{
/*
* Okay to truncate.
*/
RelationTruncate(rel, new_pages,
/* markPersistentAsPhysicallyTruncated */ true);
/* update statistics */
stats->pages_removed += num_pages - new_pages;
num_pages = new_pages;
}
}
/*
* Update the shared Free Space Map with the info we now have about free
* pages in the index, discarding any old info the map may have. We do not
* need to sort the page numbers; they're in order already.
*/
RecordIndexFreeSpace(&rel->rd_node, vstate.totFreePages,
vstate.nFreePages, vstate.freePages);
pfree(vstate.freePages);
MemoryContextDelete(vstate.pagedelcontext);
/* update statistics */
stats->num_pages = num_pages;
stats->pages_free = vstate.totFreePages;
MIRROREDLOCK_BUFMGR_VERIFY_NO_LOCK_LEAK_EXIT;
}
Datum
ginvacuumcleanup(PG_FUNCTION_ARGS)
{
MIRROREDLOCK_BUFMGR_DECLARE;
IndexVacuumInfo *info = (IndexVacuumInfo *) PG_GETARG_POINTER(0);
IndexBulkDeleteResult *stats = (IndexBulkDeleteResult *) PG_GETARG_POINTER(1);
Relation index = info->index;
bool needLock;
BlockNumber npages,
blkno;
BlockNumber totFreePages,
nFreePages,
*freePages,
maxFreePages;
BlockNumber lastBlock = GIN_ROOT_BLKNO,
lastFilledBlock = GIN_ROOT_BLKNO;
/* Set up all-zero stats if ginbulkdelete wasn't called */
if (stats == NULL)
stats = (IndexBulkDeleteResult *) palloc0(sizeof(IndexBulkDeleteResult));
/*
* XXX we always report the heap tuple count as the number of index
* entries. This is bogus if the index is partial, but it's real hard to
* tell how many distinct heap entries are referenced by a GIN index.
*/
stats->num_index_tuples = info->num_heap_tuples;
/*
* If vacuum full, we already have exclusive lock on the index. Otherwise,
* need lock unless it's local to this backend.
*/
if (info->vacuum_full)
needLock = false;
else
needLock = !RELATION_IS_LOCAL(index);
if (needLock)
LockRelationForExtension(index, ExclusiveLock);
npages = RelationGetNumberOfBlocks(index);
if (needLock)
UnlockRelationForExtension(index, ExclusiveLock);
maxFreePages = npages;
if (maxFreePages > MaxFSMPages)
maxFreePages = MaxFSMPages;
totFreePages = nFreePages = 0;
freePages = (BlockNumber *) palloc(sizeof(BlockNumber) * maxFreePages);
for (blkno = GIN_ROOT_BLKNO + 1; blkno < npages; blkno++)
{
Buffer buffer;
Page page;
vacuum_delay_point();
// -------- MirroredLock ----------
MIRROREDLOCK_BUFMGR_LOCK;
buffer = ReadBuffer(index, blkno);
LockBuffer(buffer, GIN_SHARE);
page = (Page) BufferGetPage(buffer);
if (GinPageIsDeleted(page))
{
if (nFreePages < maxFreePages)
freePages[nFreePages++] = blkno;
totFreePages++;
}
else
lastFilledBlock = blkno;
UnlockReleaseBuffer(buffer);
MIRROREDLOCK_BUFMGR_UNLOCK;
// -------- MirroredLock ----------
}
lastBlock = npages - 1;
if (info->vacuum_full && nFreePages > 0)
{
/* try to truncate index */
int i;
for (i = 0; i < nFreePages; i++)
if (freePages[i] >= lastFilledBlock)
{
totFreePages = nFreePages = i;
break;
}
if (lastBlock > lastFilledBlock)
RelationTruncate(
index,
lastFilledBlock + 1,
/* markPersistentAsPhysicallyTruncated */ true);
stats->pages_removed = lastBlock - lastFilledBlock;
}
RecordIndexFreeSpace(&index->rd_node, totFreePages, nFreePages, freePages);
stats->pages_free = totFreePages;
if (needLock)
LockRelationForExtension(index, ExclusiveLock);
stats->num_pages = RelationGetNumberOfBlocks(index);
if (needLock)
UnlockRelationForExtension(index, ExclusiveLock);
PG_RETURN_POINTER(stats);
}
/*
* Return a pinned and exclusively locked buffer which can be used to insert an
* index item of size itemsz (caller must ensure not to request sizes
* impossible to fulfill). If oldbuf is a valid buffer, it is also locked (in
* an order determined to avoid deadlocks.)
*
* If we find that the old page is no longer a regular index page (because
* of a revmap extension), the old buffer is unlocked and we return
* InvalidBuffer.
*
* If there's no existing page with enough free space to accommodate the new
* item, the relation is extended. If this happens, *extended is set to true,
* and it is the caller's responsibility to initialize the page (and WAL-log
* that fact) prior to use.
*
* Note that in some corner cases it is possible for this routine to extend the
* relation and then not return the buffer. It is this routine's
* responsibility to WAL-log the page initialization and to record the page in
* FSM if that happens. Such a buffer may later be reused by this routine.
*/
static Buffer
brin_getinsertbuffer(Relation irel, Buffer oldbuf, Size itemsz,
bool *extended)
{
BlockNumber oldblk;
BlockNumber newblk;
Page page;
int freespace;
/* callers must have checked */
Assert(itemsz <= BrinMaxItemSize);
*extended = false;
if (BufferIsValid(oldbuf))
oldblk = BufferGetBlockNumber(oldbuf);
else
oldblk = InvalidBlockNumber;
/*
* Loop until we find a page with sufficient free space. By the time we
* return to caller out of this loop, both buffers are valid and locked;
* if we have to restart here, neither buffer is locked and buf is not a
* pinned buffer.
*/
newblk = RelationGetTargetBlock(irel);
if (newblk == InvalidBlockNumber)
newblk = GetPageWithFreeSpace(irel, itemsz);
for (;;)
{
Buffer buf;
bool extensionLockHeld = false;
CHECK_FOR_INTERRUPTS();
if (newblk == InvalidBlockNumber)
{
/*
* There's not enough free space in any existing index page,
* according to the FSM: extend the relation to obtain a shiny new
* page.
*/
if (!RELATION_IS_LOCAL(irel))
{
LockRelationForExtension(irel, ExclusiveLock);
extensionLockHeld = true;
}
buf = ReadBuffer(irel, P_NEW);
newblk = BufferGetBlockNumber(buf);
*extended = true;
BRIN_elog((DEBUG2, "brin_getinsertbuffer: extending to page %u",
BufferGetBlockNumber(buf)));
}
else if (newblk == oldblk)
{
/*
* There's an odd corner-case here where the FSM is out-of-date,
* and gave us the old page.
*/
buf = oldbuf;
}
else
{
buf = ReadBuffer(irel, newblk);
}
/*
* We lock the old buffer first, if it's earlier than the new one; but
* before we do, we need to check that it hasn't been turned into a
* revmap page concurrently; if we detect that it happened, give up
* and tell caller to start over.
*/
if (BufferIsValid(oldbuf) && oldblk < newblk)
{
LockBuffer(oldbuf, BUFFER_LOCK_EXCLUSIVE);
if (!BRIN_IS_REGULAR_PAGE(BufferGetPage(oldbuf)))
{
LockBuffer(oldbuf, BUFFER_LOCK_UNLOCK);
/*
* It is possible that the new page was obtained from
* extending the relation. In that case, we must be sure to
* record it in the FSM before leaving, because otherwise the
* space would be lost forever. However, we cannot let an
* uninitialized page get in the FSM, so we need to initialize
* it first.
*/
if (*extended)
{
brin_initialize_empty_new_buffer(irel, buf);
/* shouldn't matter, but don't confuse caller */
*extended = false;
}
if (extensionLockHeld)
UnlockRelationForExtension(irel, ExclusiveLock);
ReleaseBuffer(buf);
return InvalidBuffer;
}
}
LockBuffer(buf, BUFFER_LOCK_EXCLUSIVE);
if (extensionLockHeld)
UnlockRelationForExtension(irel, ExclusiveLock);
page = BufferGetPage(buf);
/*
* We have a new buffer to insert into. Check that the new page has
* enough free space, and return it if it does; otherwise start over.
* Note that we allow for the FSM to be out of date here, and in that
* case we update it and move on.
*
* (br_page_get_freespace also checks that the FSM didn't hand us a
* page that has since been repurposed for the revmap.)
*/
freespace = *extended ?
BrinMaxItemSize : br_page_get_freespace(page);
if (freespace >= itemsz)
{
RelationSetTargetBlock(irel, BufferGetBlockNumber(buf));
/*
* Since the target block specification can get lost on cache
* invalidations, make sure we update the more permanent FSM with
* data about it before going away.
*/
if (*extended)
RecordPageWithFreeSpace(irel, BufferGetBlockNumber(buf),
freespace);
/*
* Lock the old buffer if not locked already. Note that in this
* case we know for sure it's a regular page: it's later than the
* new page we just got, which is not a revmap page, and revmap
* pages are always consecutive.
*/
if (BufferIsValid(oldbuf) && oldblk > newblk)
{
LockBuffer(oldbuf, BUFFER_LOCK_EXCLUSIVE);
Assert(BRIN_IS_REGULAR_PAGE(BufferGetPage(oldbuf)));
}
return buf;
}
/* This page is no good. */
/*
* If an entirely new page does not contain enough free space for the
* new item, then surely that item is oversized. Complain loudly; but
* first make sure we initialize the page and record it as free, for
* next time.
*/
if (*extended)
{
brin_initialize_empty_new_buffer(irel, buf);
ereport(ERROR,
(errcode(ERRCODE_PROGRAM_LIMIT_EXCEEDED),
errmsg("index row size %lu exceeds maximum %lu for index \"%s\"",
(unsigned long) itemsz,
(unsigned long) freespace,
RelationGetRelationName(irel))));
return InvalidBuffer; /* keep compiler quiet */
}
if (newblk != oldblk)
UnlockReleaseBuffer(buf);
if (BufferIsValid(oldbuf) && oldblk <= newblk)
LockBuffer(oldbuf, BUFFER_LOCK_UNLOCK);
newblk = RecordAndGetPageWithFreeSpace(irel, newblk, freespace, itemsz);
}
}
/*
* RelationGetBufferForTuple
*
* Returns pinned and exclusive-locked buffer of a page in given relation
* with free space >= given len.
*
* If otherBuffer is not InvalidBuffer, then it references a previously
* pinned buffer of another page in the same relation; on return, this
* buffer will also be exclusive-locked. (This case is used by heap_update;
* the otherBuffer contains the tuple being updated.)
*
* The reason for passing otherBuffer is that if two backends are doing
* concurrent heap_update operations, a deadlock could occur if they try
* to lock the same two buffers in opposite orders. To ensure that this
* can't happen, we impose the rule that buffers of a relation must be
* locked in increasing page number order. This is most conveniently done
* by having RelationGetBufferForTuple lock them both, with suitable care
* for ordering.
*
* NOTE: it is unlikely, but not quite impossible, for otherBuffer to be the
* same buffer we select for insertion of the new tuple (this could only
* happen if space is freed in that page after heap_update finds there's not
* enough there). In that case, the page will be pinned and locked only once.
*
* If use_fsm is true (the normal case), we use FSM to help us find free
* space. If use_fsm is false, we always append a new empty page to the
* end of the relation if the tuple won't fit on the current target page.
* This can save some cycles when we know the relation is new and doesn't
* contain useful amounts of free space.
*
* The use_fsm = false case is also useful for non-WAL-logged additions to a
* relation, if the caller holds exclusive lock and is careful to invalidate
* relation->rd_targblock before the first insertion --- that ensures that
* all insertions will occur into newly added pages and not be intermixed
* with tuples from other transactions. That way, a crash can't risk losing
* any committed data of other transactions. (See heap_insert's comments
* for additional constraints needed for safe usage of this behavior.)
*
* We always try to avoid filling existing pages further than the fillfactor.
* This is OK since this routine is not consulted when updating a tuple and
* keeping it on the same page, which is the scenario fillfactor is meant
* to reserve space for.
*
* ereport(ERROR) is allowed here, so this routine *must* be called
* before any (unlogged) changes are made in buffer pool.
*/
Buffer
RelationGetBufferForTuple(Relation relation, Size len,
Buffer otherBuffer, bool use_fsm)
{
Buffer buffer = InvalidBuffer;
Page pageHeader;
Size pageFreeSpace,
saveFreeSpace;
BlockNumber targetBlock,
otherBlock;
bool needLock;
MIRROREDLOCK_BUFMGR_MUST_ALREADY_BE_HELD;
len = MAXALIGN(len); /* be conservative */
/*
* If we're gonna fail for oversize tuple, do it right away
*/
if (len > MaxHeapTupleSize)
ereport(ERROR,
(errcode(ERRCODE_PROGRAM_LIMIT_EXCEEDED),
errmsg("row is too big: size %lu, maximum size %lu",
(unsigned long) len,
(unsigned long) MaxHeapTupleSize)));
/* Compute desired extra freespace due to fillfactor option */
saveFreeSpace = RelationGetTargetPageFreeSpace(relation,
HEAP_DEFAULT_FILLFACTOR);
if (otherBuffer != InvalidBuffer)
otherBlock = BufferGetBlockNumber(otherBuffer);
else
otherBlock = InvalidBlockNumber; /* just to keep compiler quiet */
/*
* We first try to put the tuple on the same page we last inserted a tuple
* on, as cached in the relcache entry. If that doesn't work, we ask the
* shared Free Space Map to locate a suitable page. Since the FSM's info
* might be out of date, we have to be prepared to loop around and retry
* multiple times. (To insure this isn't an infinite loop, we must update
* the FSM with the correct amount of free space on each page that proves
* not to be suitable.) If the FSM has no record of a page with enough
* free space, we give up and extend the relation.
*
* When use_fsm is false, we either put the tuple onto the existing target
* page or extend the relation.
*/
if (len + saveFreeSpace <= MaxHeapTupleSize)
targetBlock = relation->rd_targblock;
else
{
/* can't fit, don't screw up FSM request tracking by trying */
targetBlock = InvalidBlockNumber;
use_fsm = false;
}
if (targetBlock == InvalidBlockNumber && use_fsm)
{
/*
* We have no cached target page, so ask the FSM for an initial
* target.
*/
targetBlock = GetPageWithFreeSpace(&relation->rd_node,
len + saveFreeSpace);
/*
* If the FSM knows nothing of the rel, try the last page before we
* give up and extend. This avoids one-tuple-per-page syndrome during
* bootstrapping or in a recently-started system.
*/
if (targetBlock == InvalidBlockNumber)
{
BlockNumber nblocks = RelationGetNumberOfBlocks(relation);
if (nblocks > 0)
targetBlock = nblocks - 1;
}
}
while (targetBlock != InvalidBlockNumber)
{
/*
* Read and exclusive-lock the target block, as well as the other
* block if one was given, taking suitable care with lock ordering and
* the possibility they are the same block.
*/
if (otherBuffer == InvalidBuffer)
{
/* easy case */
buffer = ReadBuffer(relation, targetBlock);
LockBuffer(buffer, BUFFER_LOCK_EXCLUSIVE);
}
else if (otherBlock == targetBlock)
{
/* also easy case */
buffer = otherBuffer;
LockBuffer(buffer, BUFFER_LOCK_EXCLUSIVE);
}
else if (otherBlock < targetBlock)
{
/* lock other buffer first */
buffer = ReadBuffer(relation, targetBlock);
LockBuffer(otherBuffer, BUFFER_LOCK_EXCLUSIVE);
LockBuffer(buffer, BUFFER_LOCK_EXCLUSIVE);
}
else
{
/* lock target buffer first */
buffer = ReadBuffer(relation, targetBlock);
LockBuffer(buffer, BUFFER_LOCK_EXCLUSIVE);
LockBuffer(otherBuffer, BUFFER_LOCK_EXCLUSIVE);
}
/*
* Now we can check to see if there's enough free space here. If so,
* we're done.
*/
pageHeader = (Page) BufferGetPage(buffer);
pageFreeSpace = PageGetFreeSpace(pageHeader);
if (len + saveFreeSpace <= pageFreeSpace)
{
/* use this page as future insert target, too */
relation->rd_targblock = targetBlock;
return buffer;
}
/*
* Not enough space, so we must give up our page locks and pin (if
* any) and prepare to look elsewhere. We don't care which order we
* unlock the two buffers in, so this can be slightly simpler than the
* code above.
*/
LockBuffer(buffer, BUFFER_LOCK_UNLOCK);
if (otherBuffer == InvalidBuffer)
ReleaseBuffer(buffer);
else if (otherBlock != targetBlock)
{
LockBuffer(otherBuffer, BUFFER_LOCK_UNLOCK);
ReleaseBuffer(buffer);
}
/* Without FSM, always fall out of the loop and extend */
if (!use_fsm)
break;
/*
* Update FSM as to condition of this page, and ask for another page
* to try.
*/
targetBlock = RecordAndGetPageWithFreeSpace(&relation->rd_node,
targetBlock,
pageFreeSpace,
len + saveFreeSpace);
}
/*
* Have to extend the relation.
*
* We have to use a lock to ensure no one else is extending the rel at the
* same time, else we will both try to initialize the same new page. We
* can skip locking for new or temp relations, however, since no one else
* could be accessing them.
*/
needLock = !RELATION_IS_LOCAL(relation);
if (needLock)
LockRelationForExtension(relation, ExclusiveLock);
/*
* XXX This does an lseek - rather expensive - but at the moment it is the
* only way to accurately determine how many blocks are in a relation. Is
* it worth keeping an accurate file length in shared memory someplace,
* rather than relying on the kernel to do it for us?
*/
buffer = ReadBuffer(relation, P_NEW);
/*
* We can be certain that locking the otherBuffer first is OK, since it
* must have a lower page number.
*/
if (otherBuffer != InvalidBuffer)
LockBuffer(otherBuffer, BUFFER_LOCK_EXCLUSIVE);
/*
* Now acquire lock on the new page.
*/
LockBuffer(buffer, BUFFER_LOCK_EXCLUSIVE);
/*
* Release the file-extension lock; it's now OK for someone else to extend
* the relation some more. Note that we cannot release this lock before
* we have buffer lock on the new page, or we risk a race condition
* against vacuumlazy.c --- see comments therein.
*/
if (needLock)
UnlockRelationForExtension(relation, ExclusiveLock);
/*
* We need to initialize the empty new page. Double-check that it really
* is empty (this should never happen, but if it does we don't want to
* risk wiping out valid data).
*/
pageHeader = (Page) BufferGetPage(buffer);
if (!PageIsNew((PageHeader) pageHeader))
elog(ERROR, "page %u of relation \"%s\" should be empty but is not",
BufferGetBlockNumber(buffer),
RelationGetRelationName(relation));
PageInit(pageHeader, BufferGetPageSize(buffer), 0);
if (len > PageGetFreeSpace(pageHeader))
{
/* We should not get here given the test at the top */
elog(PANIC, "tuple is too big: size %lu", (unsigned long) len);
}
/*
* Remember the new page as our target for future insertions.
*
* XXX should we enter the new page into the free space map immediately,
* or just keep it for this backend's exclusive use in the short run
* (until VACUUM sees it)? Seems to depend on whether you expect the
* current backend to make more insertions or not, which is probably a
* good bet most of the time. So for now, don't add it to FSM yet.
*/
relation->rd_targblock = BufferGetBlockNumber(buffer);
return buffer;
}
/*
* VACUUM cleanup: update FSM
*/
Datum
gistvacuumcleanup(PG_FUNCTION_ARGS)
{
IndexVacuumInfo *info = (IndexVacuumInfo *) PG_GETARG_POINTER(0);
GistBulkDeleteResult *stats = (GistBulkDeleteResult *) PG_GETARG_POINTER(1);
Relation rel = info->index;
BlockNumber npages,
blkno;
BlockNumber totFreePages;
BlockNumber lastBlock = GIST_ROOT_BLKNO,
lastFilledBlock = GIST_ROOT_BLKNO;
bool needLock;
/* No-op in ANALYZE ONLY mode */
if (info->analyze_only)
PG_RETURN_POINTER(stats);
/* Set up all-zero stats if gistbulkdelete wasn't called */
if (stats == NULL)
{
stats = (GistBulkDeleteResult *) palloc0(sizeof(GistBulkDeleteResult));
/* use heap's tuple count */
stats->std.num_index_tuples = info->num_heap_tuples;
stats->std.estimated_count = info->estimated_count;
/*
* XXX the above is wrong if index is partial. Would it be OK to just
* return NULL, or is there work we must do below?
*/
}
if (stats->needReindex)
ereport(NOTICE,
(errmsg("index \"%s\" needs VACUUM FULL or REINDEX to finish crash recovery",
RelationGetRelationName(rel))));
/*
* Need lock unless it's local to this backend.
*/
needLock = !RELATION_IS_LOCAL(rel);
/* try to find deleted pages */
if (needLock)
LockRelationForExtension(rel, ExclusiveLock);
npages = RelationGetNumberOfBlocks(rel);
if (needLock)
UnlockRelationForExtension(rel, ExclusiveLock);
totFreePages = 0;
for (blkno = GIST_ROOT_BLKNO + 1; blkno < npages; blkno++)
{
Buffer buffer;
Page page;
vacuum_delay_point();
buffer = ReadBufferExtended(rel, MAIN_FORKNUM, blkno, RBM_NORMAL,
info->strategy);
LockBuffer(buffer, GIST_SHARE);
page = (Page) BufferGetPage(buffer);
if (PageIsNew(page) || GistPageIsDeleted(page))
{
totFreePages++;
RecordFreeIndexPage(rel, blkno);
}
else
lastFilledBlock = blkno;
UnlockReleaseBuffer(buffer);
}
lastBlock = npages - 1;
/* Finally, vacuum the FSM */
IndexFreeSpaceMapVacuum(info->index);
/* return statistics */
stats->std.pages_free = totFreePages;
if (needLock)
LockRelationForExtension(rel, ExclusiveLock);
stats->std.num_pages = RelationGetNumberOfBlocks(rel);
if (needLock)
UnlockRelationForExtension(rel, ExclusiveLock);
PG_RETURN_POINTER(stats);
}
Datum
ginvacuumcleanup(PG_FUNCTION_ARGS)
{
IndexVacuumInfo *info = (IndexVacuumInfo *) PG_GETARG_POINTER(0);
IndexBulkDeleteResult *stats = (IndexBulkDeleteResult *) PG_GETARG_POINTER(1);
Relation index = info->index;
bool needLock;
BlockNumber npages,
blkno;
BlockNumber totFreePages;
GinState ginstate;
GinStatsData idxStat;
/*
* In an autovacuum analyze, we want to clean up pending insertions.
* Otherwise, an ANALYZE-only call is a no-op.
*/
if (info->analyze_only)
{
if (IsAutoVacuumWorkerProcess())
{
initGinState(&ginstate, index);
ginInsertCleanup(&ginstate, true, stats);
}
PG_RETURN_POINTER(stats);
}
/*
* Set up all-zero stats and cleanup pending inserts if ginbulkdelete
* wasn't called
*/
if (stats == NULL)
{
stats = (IndexBulkDeleteResult *) palloc0(sizeof(IndexBulkDeleteResult));
initGinState(&ginstate, index);
ginInsertCleanup(&ginstate, true, stats);
}
memset(&idxStat, 0, sizeof(idxStat));
/*
* XXX we always report the heap tuple count as the number of index
* entries. This is bogus if the index is partial, but it's real hard to
* tell how many distinct heap entries are referenced by a GIN index.
*/
stats->num_index_tuples = info->num_heap_tuples;
stats->estimated_count = info->estimated_count;
/*
* Need lock unless it's local to this backend.
*/
needLock = !RELATION_IS_LOCAL(index);
if (needLock)
LockRelationForExtension(index, ExclusiveLock);
npages = RelationGetNumberOfBlocks(index);
if (needLock)
UnlockRelationForExtension(index, ExclusiveLock);
totFreePages = 0;
for (blkno = GIN_ROOT_BLKNO; blkno < npages; blkno++)
{
Buffer buffer;
Page page;
vacuum_delay_point();
buffer = ReadBufferExtended(index, MAIN_FORKNUM, blkno,
RBM_NORMAL, info->strategy);
LockBuffer(buffer, GIN_SHARE);
page = (Page) BufferGetPage(buffer);
if (GinPageIsDeleted(page))
{
Assert(blkno != GIN_ROOT_BLKNO);
RecordFreeIndexPage(index, blkno);
totFreePages++;
}
else if (GinPageIsData(page))
{
idxStat.nDataPages++;
}
else if (!GinPageIsList(page))
{
idxStat.nEntryPages++;
if (GinPageIsLeaf(page))
idxStat.nEntries += PageGetMaxOffsetNumber(page);
}
UnlockReleaseBuffer(buffer);
}
/* Update the metapage with accurate page and entry counts */
idxStat.nTotalPages = npages;
ginUpdateStats(info->index, &idxStat);
/* Finally, vacuum the FSM */
IndexFreeSpaceMapVacuum(info->index);
stats->pages_free = totFreePages;
if (needLock)
LockRelationForExtension(index, ExclusiveLock);
stats->num_pages = RelationGetNumberOfBlocks(index);
if (needLock)
UnlockRelationForExtension(index, ExclusiveLock);
PG_RETURN_POINTER(stats);
}
/*
* _bt_getbuf() -- Get a buffer by block number for read or write.
*
* blkno == P_NEW means to get an unallocated index page. The page
* will be initialized before returning it.
*
* When this routine returns, the appropriate lock is set on the
* requested buffer and its reference count has been incremented
* (ie, the buffer is "locked and pinned"). Also, we apply
* _bt_checkpage to sanity-check the page (except in P_NEW case).
*/
Buffer
_bt_getbuf(Relation rel, BlockNumber blkno, int access)
{
Buffer buf;
MIRROREDLOCK_BUFMGR_MUST_ALREADY_BE_HELD;
if (blkno != P_NEW)
{
/* Read an existing block of the relation */
buf = ReadBuffer(rel, blkno);
LockBuffer(buf, access);
_bt_checkpage(rel, buf);
}
else
{
bool needLock;
Page page;
Assert(access == BT_WRITE);
/*
* First see if the FSM knows of any free pages.
*
* We can't trust the FSM's report unreservedly; we have to check that
* the page is still free. (For example, an already-free page could
* have been re-used between the time the last VACUUM scanned it and
* the time the VACUUM made its FSM updates.)
*
* In fact, it's worse than that: we can't even assume that it's safe
* to take a lock on the reported page. If somebody else has a lock
* on it, or even worse our own caller does, we could deadlock. (The
* own-caller scenario is actually not improbable. Consider an index
* on a serial or timestamp column. Nearly all splits will be at the
* rightmost page, so it's entirely likely that _bt_split will call us
* while holding a lock on the page most recently acquired from FSM. A
* VACUUM running concurrently with the previous split could well have
* placed that page back in FSM.)
*
* To get around that, we ask for only a conditional lock on the
* reported page. If we fail, then someone else is using the page,
* and we may reasonably assume it's not free. (If we happen to be
* wrong, the worst consequence is the page will be lost to use till
* the next VACUUM, which is no big problem.)
*/
for (;;)
{
blkno = GetFreeIndexPage(&rel->rd_node);
if (blkno == InvalidBlockNumber)
break;
buf = ReadBuffer(rel, blkno);
if (ConditionalLockBuffer(buf))
{
page = BufferGetPage(buf);
if (_bt_page_recyclable(page))
{
/* Okay to use page. Re-initialize and return it */
_bt_pageinit(page, BufferGetPageSize(buf));
return buf;
}
elog(DEBUG2, "FSM returned nonrecyclable page");
_bt_relbuf(rel, buf);
}
else
{
elog(DEBUG2, "FSM returned nonlockable page");
/* couldn't get lock, so just drop pin */
ReleaseBuffer(buf);
}
}
/*
* Extend the relation by one page.
*
* We have to use a lock to ensure no one else is extending the rel at
* the same time, else we will both try to initialize the same new
* page. We can skip locking for new or temp relations, however,
* since no one else could be accessing them.
*/
needLock = !RELATION_IS_LOCAL(rel);
if (needLock)
LockRelationForExtension(rel, ExclusiveLock);
buf = ReadBuffer(rel, P_NEW);
/* Acquire buffer lock on new page */
LockBuffer(buf, BT_WRITE);
/*
* Release the file-extension lock; it's now OK for someone else to
* extend the relation some more. Note that we cannot release this
* lock before we have buffer lock on the new page, or we risk a race
* condition against btvacuumscan --- see comments therein.
*/
if (needLock)
UnlockRelationForExtension(rel, ExclusiveLock);
/* Initialize the new page before returning it */
page = BufferGetPage(buf);
Assert(PageIsNew((PageHeader) page));
_bt_pageinit(page, BufferGetPageSize(buf));
}
/* ref count and lock type are correct */
return buf;
}
/*
* btvacuumscan --- scan the index for VACUUMing purposes
*
* This combines the functions of looking for leaf tuples that are deletable
* according to the vacuum callback, looking for empty pages that can be
* deleted, and looking for old deleted pages that can be recycled. Both
* btbulkdelete and btvacuumcleanup invoke this (the latter only if no
* btbulkdelete call occurred).
*
* The caller is responsible for initially allocating/zeroing a stats struct
* and for obtaining a vacuum cycle ID if necessary.
*/
static void
btvacuumscan(IndexVacuumInfo *info, IndexBulkDeleteResult *stats,
IndexBulkDeleteCallback callback, void *callback_state,
BTCycleId cycleid)
{
Relation rel = info->index;
BTVacState vstate;
BlockNumber num_pages;
BlockNumber blkno;
bool needLock;
/*
* Reset counts that will be incremented during the scan; needed in case
* of multiple scans during a single VACUUM command
*/
stats->estimated_count = false;
stats->num_index_tuples = 0;
stats->pages_deleted = 0;
/* Set up info to pass down to btvacuumpage */
vstate.info = info;
vstate.stats = stats;
vstate.callback = callback;
vstate.callback_state = callback_state;
vstate.cycleid = cycleid;
vstate.lastBlockVacuumed = BTREE_METAPAGE; /* Initialise at first block */
vstate.lastUsedPage = BTREE_METAPAGE;
vstate.totFreePages = 0;
/* Create a temporary memory context to run _bt_pagedel in */
vstate.pagedelcontext = AllocSetContextCreate(CurrentMemoryContext,
"_bt_pagedel",
ALLOCSET_DEFAULT_MINSIZE,
ALLOCSET_DEFAULT_INITSIZE,
ALLOCSET_DEFAULT_MAXSIZE);
/*
* The outer loop iterates over all index pages except the metapage, in
* physical order (we hope the kernel will cooperate in providing
* read-ahead for speed). It is critical that we visit all leaf pages,
* including ones added after we start the scan, else we might fail to
* delete some deletable tuples. Hence, we must repeatedly check the
* relation length. We must acquire the relation-extension lock while
* doing so to avoid a race condition: if someone else is extending the
* relation, there is a window where bufmgr/smgr have created a new
* all-zero page but it hasn't yet been write-locked by _bt_getbuf(). If
* we manage to scan such a page here, we'll improperly assume it can be
* recycled. Taking the lock synchronizes things enough to prevent a
* problem: either num_pages won't include the new page, or _bt_getbuf
* already has write lock on the buffer and it will be fully initialized
* before we can examine it. (See also vacuumlazy.c, which has the same
* issue.) Also, we need not worry if a page is added immediately after
* we look; the page splitting code already has write-lock on the left
* page before it adds a right page, so we must already have processed any
* tuples due to be moved into such a page.
*
* We can skip locking for new or temp relations, however, since no one
* else could be accessing them.
*/
needLock = !RELATION_IS_LOCAL(rel);
blkno = BTREE_METAPAGE + 1;
for (;;)
{
/* Get the current relation length */
if (needLock)
LockRelationForExtension(rel, ExclusiveLock);
num_pages = RelationGetNumberOfBlocks(rel);
if (needLock)
UnlockRelationForExtension(rel, ExclusiveLock);
/* Quit if we've scanned the whole relation */
if (blkno >= num_pages)
break;
/* Iterate over pages, then loop back to recheck length */
for (; blkno < num_pages; blkno++)
{
btvacuumpage(&vstate, blkno, blkno);
}
}
/*
* InHotStandby we need to scan right up to the end of the index for
* correct locking, so we may need to write a WAL record for the final
* block in the index if it was not vacuumed. It's possible that VACUUMing
* has actually removed zeroed pages at the end of the index so we need to
* take care to issue the record for last actual block and not for the
* last block that was scanned. Ignore empty indexes.
*/
if (XLogStandbyInfoActive() &&
num_pages > 1 && vstate.lastBlockVacuumed < (num_pages - 1))
{
Buffer buf;
/*
* We can't use _bt_getbuf() here because it always applies
* _bt_checkpage(), which will barf on an all-zero page. We want to
* recycle all-zero pages, not fail. Also, we want to use a
* nondefault buffer access strategy.
*/
buf = ReadBufferExtended(rel, MAIN_FORKNUM, num_pages - 1, RBM_NORMAL,
info->strategy);
LockBufferForCleanup(buf);
_bt_delitems_vacuum(rel, buf, NULL, 0, vstate.lastBlockVacuumed);
_bt_relbuf(rel, buf);
}
MemoryContextDelete(vstate.pagedelcontext);
/* update statistics */
stats->num_pages = num_pages;
stats->pages_free = vstate.totFreePages;
}
/*
* btvacuumscan --- scan the index for VACUUMing purposes
*
* This combines the functions of looking for leaf tuples that are deletable
* according to the vacuum callback, looking for empty pages that can be
* deleted, and looking for old deleted pages that can be recycled. Both
* btbulkdelete and btvacuumcleanup invoke this (the latter only if no
* btbulkdelete call occurred).
*
* The caller is responsible for initially allocating/zeroing a stats struct
* and for obtaining a vacuum cycle ID if necessary.
*/
static void
btvacuumscan(IndexVacuumInfo *info, IndexBulkDeleteResult *stats,
IndexBulkDeleteCallback callback, void *callback_state,
BTCycleId cycleid)
{
Relation rel = info->index;
BTVacState vstate;
BlockNumber num_pages;
BlockNumber blkno;
bool needLock;
/*
* Reset counts that will be incremented during the scan; needed in case
* of multiple scans during a single VACUUM command
*/
stats->estimated_count = false;
stats->num_index_tuples = 0;
stats->pages_deleted = 0;
/* Set up info to pass down to btvacuumpage */
vstate.info = info;
vstate.stats = stats;
vstate.callback = callback;
vstate.callback_state = callback_state;
vstate.cycleid = cycleid;
vstate.lastBlockVacuumed = BTREE_METAPAGE; /* Initialise at first block */
vstate.lastBlockLocked = BTREE_METAPAGE;
vstate.totFreePages = 0;
/* Create a temporary memory context to run _bt_pagedel in */
vstate.pagedelcontext = AllocSetContextCreate(CurrentMemoryContext,
"_bt_pagedel",
ALLOCSET_DEFAULT_SIZES);
/*
* The outer loop iterates over all index pages except the metapage, in
* physical order (we hope the kernel will cooperate in providing
* read-ahead for speed). It is critical that we visit all leaf pages,
* including ones added after we start the scan, else we might fail to
* delete some deletable tuples. Hence, we must repeatedly check the
* relation length. We must acquire the relation-extension lock while
* doing so to avoid a race condition: if someone else is extending the
* relation, there is a window where bufmgr/smgr have created a new
* all-zero page but it hasn't yet been write-locked by _bt_getbuf(). If
* we manage to scan such a page here, we'll improperly assume it can be
* recycled. Taking the lock synchronizes things enough to prevent a
* problem: either num_pages won't include the new page, or _bt_getbuf
* already has write lock on the buffer and it will be fully initialized
* before we can examine it. (See also vacuumlazy.c, which has the same
* issue.) Also, we need not worry if a page is added immediately after
* we look; the page splitting code already has write-lock on the left
* page before it adds a right page, so we must already have processed any
* tuples due to be moved into such a page.
*
* We can skip locking for new or temp relations, however, since no one
* else could be accessing them.
*/
needLock = !RELATION_IS_LOCAL(rel);
blkno = BTREE_METAPAGE + 1;
for (;;)
{
/* Get the current relation length */
if (needLock)
LockRelationForExtension(rel, ExclusiveLock);
num_pages = RelationGetNumberOfBlocks(rel);
if (needLock)
UnlockRelationForExtension(rel, ExclusiveLock);
/* Quit if we've scanned the whole relation */
if (blkno >= num_pages)
break;
/* Iterate over pages, then loop back to recheck length */
for (; blkno < num_pages; blkno++)
{
btvacuumpage(&vstate, blkno, blkno);
}
}
/*
* Check to see if we need to issue one final WAL record for this index,
* which may be needed for correctness on a hot standby node when non-MVCC
* index scans could take place.
*
* If the WAL is replayed in hot standby, the replay process needs to get
* cleanup locks on all index leaf pages, just as we've been doing here.
* However, we won't issue any WAL records about pages that have no items
* to be deleted. For pages between pages we've vacuumed, the replay code
* will take locks under the direction of the lastBlockVacuumed fields in
* the XLOG_BTREE_VACUUM WAL records. To cover pages after the last one
* we vacuum, we need to issue a dummy XLOG_BTREE_VACUUM WAL record
* against the last leaf page in the index, if that one wasn't vacuumed.
*/
if (XLogStandbyInfoActive() &&
vstate.lastBlockVacuumed < vstate.lastBlockLocked)
{
Buffer buf;
/*
* The page should be valid, but we can't use _bt_getbuf() because we
* want to use a nondefault buffer access strategy. Since we aren't
* going to delete any items, getting cleanup lock again is probably
* overkill, but for consistency do that anyway.
*/
buf = ReadBufferExtended(rel, MAIN_FORKNUM, vstate.lastBlockLocked,
RBM_NORMAL, info->strategy);
LockBufferForCleanup(buf);
_bt_checkpage(rel, buf);
_bt_delitems_vacuum(rel, buf, NULL, 0, vstate.lastBlockVacuumed);
_bt_relbuf(rel, buf);
}
MemoryContextDelete(vstate.pagedelcontext);
/* update statistics */
stats->num_pages = num_pages;
stats->pages_free = vstate.totFreePages;
}
/*
* RelationGetBufferForTuple
*
* Returns pinned and exclusive-locked buffer of a page in given relation
* with free space >= given len.
*
* If otherBuffer is not InvalidBuffer, then it references a previously
* pinned buffer of another page in the same relation; on return, this
* buffer will also be exclusive-locked. (This case is used by heap_update;
* the otherBuffer contains the tuple being updated.)
*
* The reason for passing otherBuffer is that if two backends are doing
* concurrent heap_update operations, a deadlock could occur if they try
* to lock the same two buffers in opposite orders. To ensure that this
* can't happen, we impose the rule that buffers of a relation must be
* locked in increasing page number order. This is most conveniently done
* by having RelationGetBufferForTuple lock them both, with suitable care
* for ordering.
*
* NOTE: it is unlikely, but not quite impossible, for otherBuffer to be the
* same buffer we select for insertion of the new tuple (this could only
* happen if space is freed in that page after heap_update finds there's not
* enough there). In that case, the page will be pinned and locked only once.
*
* For the vmbuffer and vmbuffer_other arguments, we avoid deadlock by
* locking them only after locking the corresponding heap page, and taking
* no further lwlocks while they are locked.
*
* We normally use FSM to help us find free space. However,
* if HEAP_INSERT_SKIP_FSM is specified, we just append a new empty page to
* the end of the relation if the tuple won't fit on the current target page.
* This can save some cycles when we know the relation is new and doesn't
* contain useful amounts of free space.
*
* HEAP_INSERT_SKIP_FSM is also useful for non-WAL-logged additions to a
* relation, if the caller holds exclusive lock and is careful to invalidate
* relation's smgr_targblock before the first insertion --- that ensures that
* all insertions will occur into newly added pages and not be intermixed
* with tuples from other transactions. That way, a crash can't risk losing
* any committed data of other transactions. (See heap_insert's comments
* for additional constraints needed for safe usage of this behavior.)
*
* The caller can also provide a BulkInsertState object to optimize many
* insertions into the same relation. This keeps a pin on the current
* insertion target page (to save pin/unpin cycles) and also passes a
* BULKWRITE buffer selection strategy object to the buffer manager.
* Passing NULL for bistate selects the default behavior.
*
* We always try to avoid filling existing pages further than the fillfactor.
* This is OK since this routine is not consulted when updating a tuple and
* keeping it on the same page, which is the scenario fillfactor is meant
* to reserve space for.
*
* ereport(ERROR) is allowed here, so this routine *must* be called
* before any (unlogged) changes are made in buffer pool.
*/
Buffer
RelationGetBufferForTuple(Relation relation, Size len,
Buffer otherBuffer, int options,
BulkInsertState bistate,
Buffer *vmbuffer, Buffer *vmbuffer_other)
{
bool use_fsm = !(options & HEAP_INSERT_SKIP_FSM);
Buffer buffer = InvalidBuffer;
Page page;
Size pageFreeSpace,
saveFreeSpace;
BlockNumber targetBlock,
otherBlock;
bool needLock;
len = MAXALIGN(len); /* be conservative */
/* Bulk insert is not supported for updates, only inserts. */
Assert(otherBuffer == InvalidBuffer || !bistate);
/*
* If we're gonna fail for oversize tuple, do it right away
*/
if (len > MaxHeapTupleSize)
ereport(ERROR,
(errcode(ERRCODE_PROGRAM_LIMIT_EXCEEDED),
errmsg("row is too big: size %lu, maximum size %lu",
(unsigned long) len,
(unsigned long) MaxHeapTupleSize)));
/* Compute desired extra freespace due to fillfactor option */
saveFreeSpace = RelationGetTargetPageFreeSpace(relation,
HEAP_DEFAULT_FILLFACTOR);
if (otherBuffer != InvalidBuffer)
otherBlock = BufferGetBlockNumber(otherBuffer);
else
otherBlock = InvalidBlockNumber; /* just to keep compiler quiet */
/*
* We first try to put the tuple on the same page we last inserted a tuple
* on, as cached in the BulkInsertState or relcache entry. If that
* doesn't work, we ask the Free Space Map to locate a suitable page.
* Since the FSM's info might be out of date, we have to be prepared to
* loop around and retry multiple times. (To insure this isn't an infinite
* loop, we must update the FSM with the correct amount of free space on
* each page that proves not to be suitable.) If the FSM has no record of
* a page with enough free space, we give up and extend the relation.
*
* When use_fsm is false, we either put the tuple onto the existing target
* page or extend the relation.
*/
if (len + saveFreeSpace > MaxHeapTupleSize)
{
/* can't fit, don't bother asking FSM */
targetBlock = InvalidBlockNumber;
use_fsm = false;
}
else if (bistate && bistate->current_buf != InvalidBuffer)
targetBlock = BufferGetBlockNumber(bistate->current_buf);
else
targetBlock = RelationGetTargetBlock(relation);
if (targetBlock == InvalidBlockNumber && use_fsm)
{
/*
* We have no cached target page, so ask the FSM for an initial
* target.
*/
targetBlock = GetPageWithFreeSpace(relation, len + saveFreeSpace);
/*
* If the FSM knows nothing of the rel, try the last page before we
* give up and extend. This avoids one-tuple-per-page syndrome during
* bootstrapping or in a recently-started system.
*/
if (targetBlock == InvalidBlockNumber)
{
BlockNumber nblocks = RelationGetNumberOfBlocks(relation);
if (nblocks > 0)
targetBlock = nblocks - 1;
}
}
while (targetBlock != InvalidBlockNumber)
{
/*
* Read and exclusive-lock the target block, as well as the other
* block if one was given, taking suitable care with lock ordering and
* the possibility they are the same block.
*
* If the page-level all-visible flag is set, caller will need to
* clear both that and the corresponding visibility map bit. However,
* by the time we return, we'll have x-locked the buffer, and we don't
* want to do any I/O while in that state. So we check the bit here
* before taking the lock, and pin the page if it appears necessary.
* Checking without the lock creates a risk of getting the wrong
* answer, so we'll have to recheck after acquiring the lock.
*/
if (otherBuffer == InvalidBuffer)
{
/* easy case */
buffer = ReadBufferBI(relation, targetBlock, bistate);
if (PageIsAllVisible(BufferGetPage(buffer)))
visibilitymap_pin(relation, targetBlock, vmbuffer);
LockBuffer(buffer, BUFFER_LOCK_EXCLUSIVE);
}
else if (otherBlock == targetBlock)
{
/* also easy case */
buffer = otherBuffer;
if (PageIsAllVisible(BufferGetPage(buffer)))
visibilitymap_pin(relation, targetBlock, vmbuffer);
LockBuffer(buffer, BUFFER_LOCK_EXCLUSIVE);
}
else if (otherBlock < targetBlock)
{
/* lock other buffer first */
buffer = ReadBuffer(relation, targetBlock);
if (PageIsAllVisible(BufferGetPage(buffer)))
visibilitymap_pin(relation, targetBlock, vmbuffer);
LockBuffer(otherBuffer, BUFFER_LOCK_EXCLUSIVE);
LockBuffer(buffer, BUFFER_LOCK_EXCLUSIVE);
}
else
{
/* lock target buffer first */
buffer = ReadBuffer(relation, targetBlock);
if (PageIsAllVisible(BufferGetPage(buffer)))
visibilitymap_pin(relation, targetBlock, vmbuffer);
LockBuffer(buffer, BUFFER_LOCK_EXCLUSIVE);
LockBuffer(otherBuffer, BUFFER_LOCK_EXCLUSIVE);
}
/*
* We now have the target page (and the other buffer, if any) pinned
* and locked. However, since our initial PageIsAllVisible checks
* were performed before acquiring the lock, the results might now be
* out of date, either for the selected victim buffer, or for the
* other buffer passed by the caller. In that case, we'll need to
* give up our locks, go get the pin(s) we failed to get earlier, and
* re-lock. That's pretty painful, but hopefully shouldn't happen
* often.
*
* Note that there's a small possibility that we didn't pin the page
* above but still have the correct page pinned anyway, either because
* we've already made a previous pass through this loop, or because
* caller passed us the right page anyway.
*
* Note also that it's possible that by the time we get the pin and
* retake the buffer locks, the visibility map bit will have been
* cleared by some other backend anyway. In that case, we'll have
* done a bit of extra work for no gain, but there's no real harm
* done.
*/
if (otherBuffer == InvalidBuffer || buffer <= otherBuffer)
GetVisibilityMapPins(relation, buffer, otherBuffer,
targetBlock, otherBlock, vmbuffer,
vmbuffer_other);
else
GetVisibilityMapPins(relation, otherBuffer, buffer,
otherBlock, targetBlock, vmbuffer_other,
vmbuffer);
/*
* Now we can check to see if there's enough free space here. If so,
* we're done.
*/
page = BufferGetPage(buffer);
pageFreeSpace = PageGetHeapFreeSpace(page);
if (len + saveFreeSpace <= pageFreeSpace)
{
/* use this page as future insert target, too */
RelationSetTargetBlock(relation, targetBlock);
return buffer;
}
/*
* Not enough space, so we must give up our page locks and pin (if
* any) and prepare to look elsewhere. We don't care which order we
* unlock the two buffers in, so this can be slightly simpler than the
* code above.
*/
LockBuffer(buffer, BUFFER_LOCK_UNLOCK);
if (otherBuffer == InvalidBuffer)
ReleaseBuffer(buffer);
else if (otherBlock != targetBlock)
{
LockBuffer(otherBuffer, BUFFER_LOCK_UNLOCK);
ReleaseBuffer(buffer);
}
/* Without FSM, always fall out of the loop and extend */
if (!use_fsm)
break;
/*
* Update FSM as to condition of this page, and ask for another page
* to try.
*/
targetBlock = RecordAndGetPageWithFreeSpace(relation,
targetBlock,
pageFreeSpace,
len + saveFreeSpace);
}
/*
* Have to extend the relation.
*
* We have to use a lock to ensure no one else is extending the rel at the
* same time, else we will both try to initialize the same new page. We
* can skip locking for new or temp relations, however, since no one else
* could be accessing them.
*/
needLock = !RELATION_IS_LOCAL(relation);
if (needLock)
LockRelationForExtension(relation, ExclusiveLock);
/*
* XXX This does an lseek - rather expensive - but at the moment it is the
* only way to accurately determine how many blocks are in a relation. Is
* it worth keeping an accurate file length in shared memory someplace,
* rather than relying on the kernel to do it for us?
*/
buffer = ReadBufferBI(relation, P_NEW, bistate);
/*
* We can be certain that locking the otherBuffer first is OK, since it
* must have a lower page number.
*/
if (otherBuffer != InvalidBuffer)
LockBuffer(otherBuffer, BUFFER_LOCK_EXCLUSIVE);
/*
* Now acquire lock on the new page.
*/
LockBuffer(buffer, BUFFER_LOCK_EXCLUSIVE);
/*
* Release the file-extension lock; it's now OK for someone else to extend
* the relation some more. Note that we cannot release this lock before
* we have buffer lock on the new page, or we risk a race condition
* against vacuumlazy.c --- see comments therein.
*/
if (needLock)
UnlockRelationForExtension(relation, ExclusiveLock);
/*
* We need to initialize the empty new page. Double-check that it really
* is empty (this should never happen, but if it does we don't want to
* risk wiping out valid data).
*/
page = BufferGetPage(buffer);
if (!PageIsNew(page))
elog(ERROR, "page %u of relation \"%s\" should be empty but is not",
BufferGetBlockNumber(buffer),
RelationGetRelationName(relation));
PageInit(page, BufferGetPageSize(buffer), 0);
if (len > PageGetHeapFreeSpace(page))
{
/* We should not get here given the test at the top */
elog(PANIC, "tuple is too big: size %lu", (unsigned long) len);
}
/*
* Remember the new page as our target for future insertions.
*
* XXX should we enter the new page into the free space map immediately,
* or just keep it for this backend's exclusive use in the short run
* (until VACUUM sees it)? Seems to depend on whether you expect the
* current backend to make more insertions or not, which is probably a
* good bet most of the time. So for now, don't add it to FSM yet.
*/
RelationSetTargetBlock(relation, BufferGetBlockNumber(buffer));
return buffer;
}