/*
* add PostingItem to a non-leaf page.
*/
void
GinDataPageAddPostingItem(Page page, PostingItem *data, OffsetNumber offset)
{
OffsetNumber maxoff = GinPageGetOpaque(page)->maxoff;
char *ptr;
Assert(PostingItemGetBlockNumber(data) != InvalidBlockNumber);
Assert(!GinPageIsLeaf(page));
if (offset == InvalidOffsetNumber)
{
ptr = (char *) GinDataPageGetPostingItem(page, maxoff + 1);
}
else
{
ptr = (char *) GinDataPageGetPostingItem(page, offset);
if (maxoff + 1 - offset != 0)
memmove(ptr + sizeof(PostingItem),
ptr,
(maxoff - offset + 1) * sizeof(PostingItem));
}
memcpy(ptr, data, sizeof(PostingItem));
GinPageGetOpaque(page)->maxoff++;
}
/*
* Finds links to blkno on non-leaf page, returns
* offset of PostingItem
*/
static OffsetNumber
dataFindChildPtr(GinBtree btree, Page page, BlockNumber blkno, OffsetNumber storedOff)
{
OffsetNumber i,
maxoff = GinPageGetOpaque(page)->maxoff;
PostingItem *pitem;
Assert(!GinPageIsLeaf(page));
Assert(GinPageIsData(page));
/* if page isn't changed, we return storedOff */
if (storedOff >= FirstOffsetNumber && storedOff <= maxoff)
{
pitem = (PostingItem *) GinDataPageGetItem(page, storedOff);
if (PostingItemGetBlockNumber(pitem) == blkno)
return storedOff;
/*
* we hope, that needed pointer goes to right. It's true if there
* wasn't a deletion
*/
for (i = storedOff + 1; i <= maxoff; i++)
{
pitem = (PostingItem *) GinDataPageGetItem(page, i);
if (PostingItemGetBlockNumber(pitem) == blkno)
return i;
}
maxoff = storedOff - 1;
}
/* last chance */
for (i = FirstOffsetNumber; i <= maxoff; i++)
{
pitem = (PostingItem *) GinDataPageGetItem(page, i);
if (PostingItemGetBlockNumber(pitem) == blkno)
return i;
}
return InvalidOffsetNumber;
}
/*
* returns blkno of leftmost child
*/
static BlockNumber
dataGetLeftMostPage(GinBtree btree, Page page)
{
PostingItem *pitem;
Assert(!GinPageIsLeaf(page));
Assert(GinPageIsData(page));
Assert(GinPageGetOpaque(page)->maxoff >= FirstOffsetNumber);
pitem = (PostingItem *) GinDataPageGetItem(page, FirstOffsetNumber);
return PostingItemGetBlockNumber(pitem);
}
/*
* scans posting tree and deletes empty pages
*/
static bool
ginScanToDelete(GinVacuumState *gvs, BlockNumber blkno, bool isRoot, DataPageDeleteStack *parent, OffsetNumber myoff)
{
DataPageDeleteStack *me;
Buffer buffer;
Page page;
bool meDelete = FALSE;
if (isRoot)
{
me = parent;
}
else
{
if (!parent->child)
{
me = (DataPageDeleteStack *) palloc0(sizeof(DataPageDeleteStack));
me->parent = parent;
parent->child = me;
me->leftBlkno = InvalidBlockNumber;
}
else
me = parent->child;
}
buffer = ReadBufferExtended(gvs->index, MAIN_FORKNUM, blkno,
RBM_NORMAL, gvs->strategy);
page = BufferGetPage(buffer);
Assert(GinPageIsData(page));
if (!GinPageIsLeaf(page))
{
OffsetNumber i;
me->blkno = blkno;
for (i = FirstOffsetNumber; i <= GinPageGetOpaque(page)->maxoff; i++)
{
PostingItem *pitem = (PostingItem *) GinDataPageGetItem(page, i);
if (ginScanToDelete(gvs, PostingItemGetBlockNumber(pitem), FALSE, me, i))
i--;
}
}
if (GinPageGetOpaque(page)->maxoff < FirstOffsetNumber)
{
if (!(me->leftBlkno == InvalidBlockNumber && GinPageRightMost(page)))
{
/* we never delete right most branch */
Assert(!isRoot);
if (GinPageGetOpaque(page)->maxoff < FirstOffsetNumber)
{
ginDeletePage(gvs, blkno, me->leftBlkno, me->parent->blkno, myoff, me->parent->isRoot);
meDelete = TRUE;
}
}
}
ReleaseBuffer(buffer);
if (!meDelete)
me->leftBlkno = blkno;
return meDelete;
}
static void
ginDeletePage(GinVacuumState *gvs, BlockNumber deleteBlkno, BlockNumber leftBlkno,
BlockNumber parentBlkno, OffsetNumber myoff, bool isParentRoot)
{
Buffer dBuffer;
Buffer lBuffer;
Buffer pBuffer;
Page page,
parentPage;
dBuffer = ReadBufferExtended(gvs->index, MAIN_FORKNUM, deleteBlkno,
RBM_NORMAL, gvs->strategy);
if (leftBlkno != InvalidBlockNumber)
lBuffer = ReadBufferExtended(gvs->index, MAIN_FORKNUM, leftBlkno,
RBM_NORMAL, gvs->strategy);
else
lBuffer = InvalidBuffer;
pBuffer = ReadBufferExtended(gvs->index, MAIN_FORKNUM, parentBlkno,
RBM_NORMAL, gvs->strategy);
LockBuffer(dBuffer, GIN_EXCLUSIVE);
if (!isParentRoot) /* parent is already locked by
* LockBufferForCleanup() */
LockBuffer(pBuffer, GIN_EXCLUSIVE);
if (leftBlkno != InvalidBlockNumber)
LockBuffer(lBuffer, GIN_EXCLUSIVE);
START_CRIT_SECTION();
if (leftBlkno != InvalidBlockNumber)
{
BlockNumber rightlink;
page = BufferGetPage(dBuffer);
rightlink = GinPageGetOpaque(page)->rightlink;
page = BufferGetPage(lBuffer);
GinPageGetOpaque(page)->rightlink = rightlink;
}
parentPage = BufferGetPage(pBuffer);
#ifdef USE_ASSERT_CHECKING
do
{
PostingItem *tod = (PostingItem *) GinDataPageGetItem(parentPage, myoff);
Assert(PostingItemGetBlockNumber(tod) == deleteBlkno);
} while (0);
#endif
GinPageDeletePostingItem(parentPage, myoff);
page = BufferGetPage(dBuffer);
/*
* we shouldn't change rightlink field to save workability of running
* search scan
*/
GinPageGetOpaque(page)->flags = GIN_DELETED;
MarkBufferDirty(pBuffer);
if (leftBlkno != InvalidBlockNumber)
MarkBufferDirty(lBuffer);
MarkBufferDirty(dBuffer);
if (RelationNeedsWAL(gvs->index))
{
XLogRecPtr recptr;
XLogRecData rdata[4];
ginxlogDeletePage data;
int n;
data.node = gvs->index->rd_node;
data.blkno = deleteBlkno;
data.parentBlkno = parentBlkno;
data.parentOffset = myoff;
data.leftBlkno = leftBlkno;
data.rightLink = GinPageGetOpaque(page)->rightlink;
rdata[0].buffer = dBuffer;
rdata[0].buffer_std = FALSE;
rdata[0].data = NULL;
rdata[0].len = 0;
rdata[0].next = rdata + 1;
rdata[1].buffer = pBuffer;
rdata[1].buffer_std = FALSE;
rdata[1].data = NULL;
rdata[1].len = 0;
rdata[1].next = rdata + 2;
if (leftBlkno != InvalidBlockNumber)
{
rdata[2].buffer = lBuffer;
rdata[2].buffer_std = FALSE;
rdata[2].data = NULL;
rdata[2].len = 0;
rdata[2].next = rdata + 3;
n = 3;
}
else
n = 2;
rdata[n].buffer = InvalidBuffer;
rdata[n].buffer_std = FALSE;
rdata[n].len = sizeof(ginxlogDeletePage);
rdata[n].data = (char *) &data;
rdata[n].next = NULL;
recptr = XLogInsert(RM_GIN_ID, XLOG_GIN_DELETE_PAGE, rdata);
PageSetLSN(page, recptr);
PageSetTLI(page, ThisTimeLineID);
PageSetLSN(parentPage, recptr);
PageSetTLI(parentPage, ThisTimeLineID);
if (leftBlkno != InvalidBlockNumber)
{
page = BufferGetPage(lBuffer);
PageSetLSN(page, recptr);
PageSetTLI(page, ThisTimeLineID);
}
}
if (!isParentRoot)
LockBuffer(pBuffer, GIN_UNLOCK);
ReleaseBuffer(pBuffer);
if (leftBlkno != InvalidBlockNumber)
UnlockReleaseBuffer(lBuffer);
UnlockReleaseBuffer(dBuffer);
END_CRIT_SECTION();
gvs->result->pages_deleted++;
}
static bool
ginVacuumPostingTreeLeaves(GinVacuumState *gvs, BlockNumber blkno, bool isRoot, Buffer *rootBuffer)
{
Buffer buffer;
Page page;
bool hasVoidPage = FALSE;
buffer = ReadBufferExtended(gvs->index, MAIN_FORKNUM, blkno,
RBM_NORMAL, gvs->strategy);
page = BufferGetPage(buffer);
/*
* We should be sure that we don't concurrent with inserts, insert process
* never release root page until end (but it can unlock it and lock
* again). New scan can't start but previously started ones work
* concurrently.
*/
if (isRoot)
LockBufferForCleanup(buffer);
else
LockBuffer(buffer, GIN_EXCLUSIVE);
Assert(GinPageIsData(page));
if (GinPageIsLeaf(page))
{
OffsetNumber newMaxOff,
oldMaxOff = GinPageGetOpaque(page)->maxoff;
ItemPointerData *cleaned = NULL;
newMaxOff = ginVacuumPostingList(gvs,
(ItemPointer) GinDataPageGetData(page), oldMaxOff, &cleaned);
/* saves changes about deleted tuple ... */
if (oldMaxOff != newMaxOff)
{
START_CRIT_SECTION();
if (newMaxOff > 0)
memcpy(GinDataPageGetData(page), cleaned, sizeof(ItemPointerData) * newMaxOff);
pfree(cleaned);
GinPageGetOpaque(page)->maxoff = newMaxOff;
MarkBufferDirty(buffer);
xlogVacuumPage(gvs->index, buffer);
END_CRIT_SECTION();
/* if root is a leaf page, we don't desire further processing */
if (!isRoot && GinPageGetOpaque(page)->maxoff < FirstOffsetNumber)
hasVoidPage = TRUE;
}
}
else
{
OffsetNumber i;
bool isChildHasVoid = FALSE;
for (i = FirstOffsetNumber; i <= GinPageGetOpaque(page)->maxoff; i++)
{
PostingItem *pitem = (PostingItem *) GinDataPageGetItem(page, i);
if (ginVacuumPostingTreeLeaves(gvs, PostingItemGetBlockNumber(pitem), FALSE, NULL))
isChildHasVoid = TRUE;
}
if (isChildHasVoid)
hasVoidPage = TRUE;
}
/*
* if we have root and theres void pages in tree, then we don't release
* lock to go further processing and guarantee that tree is unused
*/
if (!(isRoot && hasVoidPage))
{
UnlockReleaseBuffer(buffer);
}
else
{
Assert(rootBuffer);
*rootBuffer = buffer;
}
return hasVoidPage;
}
/*
* Delete a posting tree page.
*/
static void
ginDeletePage(GinVacuumState *gvs, BlockNumber deleteBlkno, BlockNumber leftBlkno,
BlockNumber parentBlkno, OffsetNumber myoff, bool isParentRoot)
{
Buffer dBuffer;
Buffer lBuffer;
Buffer pBuffer;
Page page,
parentPage;
BlockNumber rightlink;
/*
* Lock the pages in the same order as an insertion would, to avoid
* deadlocks: left, then right, then parent.
*/
lBuffer = ReadBufferExtended(gvs->index, MAIN_FORKNUM, leftBlkno,
RBM_NORMAL, gvs->strategy);
dBuffer = ReadBufferExtended(gvs->index, MAIN_FORKNUM, deleteBlkno,
RBM_NORMAL, gvs->strategy);
pBuffer = ReadBufferExtended(gvs->index, MAIN_FORKNUM, parentBlkno,
RBM_NORMAL, gvs->strategy);
LockBuffer(lBuffer, GIN_EXCLUSIVE);
LockBuffer(dBuffer, GIN_EXCLUSIVE);
if (!isParentRoot) /* parent is already locked by
* LockBufferForCleanup() */
LockBuffer(pBuffer, GIN_EXCLUSIVE);
START_CRIT_SECTION();
/* Unlink the page by changing left sibling's rightlink */
page = BufferGetPage(dBuffer);
rightlink = GinPageGetOpaque(page)->rightlink;
page = BufferGetPage(lBuffer);
GinPageGetOpaque(page)->rightlink = rightlink;
/* Delete downlink from parent */
parentPage = BufferGetPage(pBuffer);
#ifdef USE_ASSERT_CHECKING
do
{
PostingItem *tod = GinDataPageGetPostingItem(parentPage, myoff);
Assert(PostingItemGetBlockNumber(tod) == deleteBlkno);
} while (0);
#endif
GinPageDeletePostingItem(parentPage, myoff);
page = BufferGetPage(dBuffer);
/*
* we shouldn't change rightlink field to save workability of running
* search scan
*/
GinPageGetOpaque(page)->flags = GIN_DELETED;
MarkBufferDirty(pBuffer);
MarkBufferDirty(lBuffer);
MarkBufferDirty(dBuffer);
if (RelationNeedsWAL(gvs->index))
{
XLogRecPtr recptr;
ginxlogDeletePage data;
/*
* We can't pass REGBUF_STANDARD for the deleted page, because we
* didn't set pd_lower on pre-9.4 versions. The page might've been
* binary-upgraded from an older version, and hence not have pd_lower
* set correctly. Ditto for the left page, but removing the item from
* the parent updated its pd_lower, so we know that's OK at this
* point.
*/
XLogBeginInsert();
XLogRegisterBuffer(0, dBuffer, 0);
XLogRegisterBuffer(1, pBuffer, REGBUF_STANDARD);
XLogRegisterBuffer(2, lBuffer, 0);
data.parentOffset = myoff;
data.rightLink = GinPageGetOpaque(page)->rightlink;
XLogRegisterData((char *) &data, sizeof(ginxlogDeletePage));
recptr = XLogInsert(RM_GIN_ID, XLOG_GIN_DELETE_PAGE);
PageSetLSN(page, recptr);
PageSetLSN(parentPage, recptr);
PageSetLSN(BufferGetPage(lBuffer), recptr);
}
if (!isParentRoot)
LockBuffer(pBuffer, GIN_UNLOCK);
ReleaseBuffer(pBuffer);
UnlockReleaseBuffer(lBuffer);
UnlockReleaseBuffer(dBuffer);
END_CRIT_SECTION();
gvs->result->pages_deleted++;
}
static bool
ginVacuumPostingTreeLeaves(GinVacuumState *gvs, BlockNumber blkno, bool isRoot, Buffer *rootBuffer)
{
Buffer buffer;
Page page;
bool hasVoidPage = FALSE;
MemoryContext oldCxt;
buffer = ReadBufferExtended(gvs->index, MAIN_FORKNUM, blkno,
RBM_NORMAL, gvs->strategy);
page = BufferGetPage(buffer);
/*
* We should be sure that we don't concurrent with inserts, insert process
* never release root page until end (but it can unlock it and lock
* again). New scan can't start but previously started ones work
* concurrently.
*/
if (isRoot)
LockBufferForCleanup(buffer);
else
LockBuffer(buffer, GIN_EXCLUSIVE);
Assert(GinPageIsData(page));
if (GinPageIsLeaf(page))
{
oldCxt = MemoryContextSwitchTo(gvs->tmpCxt);
ginVacuumPostingTreeLeaf(gvs->index, buffer, gvs);
MemoryContextSwitchTo(oldCxt);
MemoryContextReset(gvs->tmpCxt);
/* if root is a leaf page, we don't desire further processing */
if (!isRoot && !hasVoidPage && GinDataLeafPageIsEmpty(page))
hasVoidPage = TRUE;
}
else
{
OffsetNumber i;
bool isChildHasVoid = FALSE;
for (i = FirstOffsetNumber; i <= GinPageGetOpaque(page)->maxoff; i++)
{
PostingItem *pitem = GinDataPageGetPostingItem(page, i);
if (ginVacuumPostingTreeLeaves(gvs, PostingItemGetBlockNumber(pitem), FALSE, NULL))
isChildHasVoid = TRUE;
}
if (isChildHasVoid)
hasVoidPage = TRUE;
}
/*
* if we have root and there are empty pages in tree, then we don't
* release lock to go further processing and guarantee that tree is unused
*/
if (!(isRoot && hasVoidPage))
{
UnlockReleaseBuffer(buffer);
}
else
{
Assert(rootBuffer);
*rootBuffer = buffer;
}
return hasVoidPage;
}
/*
* Find correct PostingItem in non-leaf page. It supposed that page
* correctly chosen and searching value SHOULD be on page
*/
static BlockNumber
dataLocateItem(GinBtree btree, GinBtreeStack *stack)
{
OffsetNumber low,
high,
maxoff;
PostingItem *pitem = NULL;
int result;
Page page = BufferGetPage(stack->buffer);
Assert(!GinPageIsLeaf(page));
Assert(GinPageIsData(page));
if (btree->fullScan)
{
stack->off = FirstOffsetNumber;
stack->predictNumber *= GinPageGetOpaque(page)->maxoff;
return btree->getLeftMostPage(btree, page);
}
low = FirstOffsetNumber;
maxoff = high = GinPageGetOpaque(page)->maxoff;
Assert(high >= low);
high++;
while (high > low)
{
OffsetNumber mid = low + ((high - low) / 2);
pitem = (PostingItem *) GinDataPageGetItem(page, mid);
if (mid == maxoff)
{
/*
* Right infinity, page already correctly chosen with a help of
* dataIsMoveRight
*/
result = -1;
}
else
{
pitem = (PostingItem *) GinDataPageGetItem(page, mid);
result = ginCompareItemPointers(btree->items + btree->curitem, &(pitem->key));
}
if (result == 0)
{
stack->off = mid;
return PostingItemGetBlockNumber(pitem);
}
else if (result > 0)
low = mid + 1;
else
high = mid;
}
Assert(high >= FirstOffsetNumber && high <= maxoff);
stack->off = high;
pitem = (PostingItem *) GinDataPageGetItem(page, high);
return PostingItemGetBlockNumber(pitem);
}
/*
* split page and fills WAL record. original buffer(lbuf) leaves untouched,
* returns shadow page of lbuf filled new data. In leaf page and build mode puts all
* ItemPointers to pages. Also, in build mode splits data by way to full fulled
* left page
*/
static Page
dataSplitPage(GinBtree btree, Buffer lbuf, Buffer rbuf, OffsetNumber off, XLogRecData **prdata)
{
char *ptr;
OffsetNumber separator;
ItemPointer bound;
Page lpage = PageGetTempPageCopy(BufferGetPage(lbuf));
ItemPointerData oldbound = *GinDataPageGetRightBound(lpage);
int sizeofitem = GinSizeOfDataPageItem(lpage);
OffsetNumber maxoff = GinPageGetOpaque(lpage)->maxoff;
Page rpage = BufferGetPage(rbuf);
Size pageSize = PageGetPageSize(lpage);
Size freeSpace;
uint32 nCopied = 1;
/* these must be static so they can be returned to caller */
static ginxlogSplit data;
static XLogRecData rdata[4];
static char vector[2 * BLCKSZ];
GinInitPage(rpage, GinPageGetOpaque(lpage)->flags, pageSize);
freeSpace = GinDataPageGetFreeSpace(rpage);
*prdata = rdata;
data.leftChildBlkno = (GinPageIsLeaf(lpage)) ?
InvalidOffsetNumber : PostingItemGetBlockNumber(&(btree->pitem));
data.updateBlkno = dataPrepareData(btree, lpage, off);
memcpy(vector, GinDataPageGetItem(lpage, FirstOffsetNumber),
maxoff * sizeofitem);
if (GinPageIsLeaf(lpage) && GinPageRightMost(lpage) && off > GinPageGetOpaque(lpage)->maxoff)
{
nCopied = 0;
while (btree->curitem < btree->nitem &&
maxoff * sizeof(ItemPointerData) < 2 * (freeSpace - sizeof(ItemPointerData)))
{
memcpy(vector + maxoff * sizeof(ItemPointerData),
btree->items + btree->curitem,
sizeof(ItemPointerData));
maxoff++;
nCopied++;
btree->curitem++;
}
}
else
{
ptr = vector + (off - 1) * sizeofitem;
if (maxoff + 1 - off != 0)
memmove(ptr + sizeofitem, ptr, (maxoff - off + 1) * sizeofitem);
if (GinPageIsLeaf(lpage))
{
memcpy(ptr, btree->items + btree->curitem, sizeofitem);
btree->curitem++;
}
else
memcpy(ptr, &(btree->pitem), sizeofitem);
maxoff++;
}
/*
* we suppose that during index creation table scaned from begin to end,
* so ItemPointers are monotonically increased..
*/
if (btree->isBuild && GinPageRightMost(lpage))
separator = freeSpace / sizeofitem;
else
separator = maxoff / 2;
GinInitPage(rpage, GinPageGetOpaque(lpage)->flags, pageSize);
GinInitPage(lpage, GinPageGetOpaque(rpage)->flags, pageSize);
memcpy(GinDataPageGetItem(lpage, FirstOffsetNumber), vector, separator * sizeofitem);
GinPageGetOpaque(lpage)->maxoff = separator;
memcpy(GinDataPageGetItem(rpage, FirstOffsetNumber),
vector + separator * sizeofitem, (maxoff - separator) * sizeofitem);
GinPageGetOpaque(rpage)->maxoff = maxoff - separator;
PostingItemSetBlockNumber(&(btree->pitem), BufferGetBlockNumber(lbuf));
if (GinPageIsLeaf(lpage))
btree->pitem.key = *(ItemPointerData *) GinDataPageGetItem(lpage,
GinPageGetOpaque(lpage)->maxoff);
else
btree->pitem.key = ((PostingItem *) GinDataPageGetItem(lpage,
GinPageGetOpaque(lpage)->maxoff))->key;
btree->rightblkno = BufferGetBlockNumber(rbuf);
/* set up right bound for left page */
bound = GinDataPageGetRightBound(lpage);
*bound = btree->pitem.key;
/* set up right bound for right page */
bound = GinDataPageGetRightBound(rpage);
*bound = oldbound;
data.node = btree->index->rd_node;
data.rootBlkno = InvalidBlockNumber;
data.lblkno = BufferGetBlockNumber(lbuf);
data.rblkno = BufferGetBlockNumber(rbuf);
data.separator = separator;
data.nitem = maxoff;
data.isData = TRUE;
data.isLeaf = GinPageIsLeaf(lpage) ? TRUE : FALSE;
data.isRootSplit = FALSE;
data.rightbound = oldbound;
rdata[0].buffer = InvalidBuffer;
rdata[0].data = (char *) &data;
rdata[0].len = sizeof(ginxlogSplit);
rdata[0].next = &rdata[1];
rdata[1].buffer = InvalidBuffer;
rdata[1].data = vector;
rdata[1].len = MAXALIGN(maxoff * sizeofitem);
rdata[1].next = NULL;
return lpage;
}
/*
* Scan through posting tree, delete empty tuples from leaf pages.
* Also, this function collects empty subtrees (with all empty leafs).
* For parents of these subtrees CleanUp lock is taken, then we call
* ScanToDelete. This is done for every inner page, which points to
* empty subtree.
*/
static bool
ginVacuumPostingTreeLeaves(GinVacuumState *gvs, BlockNumber blkno, bool isRoot)
{
Buffer buffer;
Page page;
bool hasVoidPage = FALSE;
MemoryContext oldCxt;
buffer = ReadBufferExtended(gvs->index, MAIN_FORKNUM, blkno,
RBM_NORMAL, gvs->strategy);
page = BufferGetPage(buffer);
ginTraverseLock(buffer, false);
Assert(GinPageIsData(page));
if (GinPageIsLeaf(page))
{
oldCxt = MemoryContextSwitchTo(gvs->tmpCxt);
ginVacuumPostingTreeLeaf(gvs->index, buffer, gvs);
MemoryContextSwitchTo(oldCxt);
MemoryContextReset(gvs->tmpCxt);
/* if root is a leaf page, we don't desire further processing */
if (GinDataLeafPageIsEmpty(page))
hasVoidPage = TRUE;
UnlockReleaseBuffer(buffer);
return hasVoidPage;
}
else
{
OffsetNumber i;
bool hasEmptyChild = FALSE;
bool hasNonEmptyChild = FALSE;
OffsetNumber maxoff = GinPageGetOpaque(page)->maxoff;
BlockNumber *children = palloc(sizeof(BlockNumber) * (maxoff + 1));
/*
* Read all children BlockNumbers. Not sure it is safe if there are
* many concurrent vacuums.
*/
for (i = FirstOffsetNumber; i <= maxoff; i++)
{
PostingItem *pitem = GinDataPageGetPostingItem(page, i);
children[i] = PostingItemGetBlockNumber(pitem);
}
UnlockReleaseBuffer(buffer);
for (i = FirstOffsetNumber; i <= maxoff; i++)
{
if (ginVacuumPostingTreeLeaves(gvs, children[i], FALSE))
hasEmptyChild = TRUE;
else
hasNonEmptyChild = TRUE;
}
pfree(children);
vacuum_delay_point();
/*
* All subtree is empty - just return TRUE to indicate that parent
* must do a cleanup. Unless we are ROOT an there is way to go upper.
*/
if (hasEmptyChild && !hasNonEmptyChild && !isRoot)
return TRUE;
if (hasEmptyChild)
{
DataPageDeleteStack root,
*ptr,
*tmp;
buffer = ReadBufferExtended(gvs->index, MAIN_FORKNUM, blkno,
RBM_NORMAL, gvs->strategy);
LockBufferForCleanup(buffer);
memset(&root, 0, sizeof(DataPageDeleteStack));
root.leftBlkno = InvalidBlockNumber;
root.isRoot = TRUE;
ginScanToDelete(gvs, blkno, TRUE, &root, InvalidOffsetNumber);
ptr = root.child;
while (ptr)
{
tmp = ptr->child;
pfree(ptr);
ptr = tmp;
}
UnlockReleaseBuffer(buffer);
}
/* Here we have deleted all empty subtrees */
return FALSE;
}
}
/*
* Delete a posting tree page.
*/
static void
ginDeletePage(GinVacuumState *gvs, BlockNumber deleteBlkno, BlockNumber leftBlkno,
BlockNumber parentBlkno, OffsetNumber myoff, bool isParentRoot)
{
Buffer dBuffer;
Buffer lBuffer;
Buffer pBuffer;
Page page,
parentPage;
BlockNumber rightlink;
/*
* This function MUST be called only if someone of parent pages hold
* exclusive cleanup lock. This guarantees that no insertions currently
* happen in this subtree. Caller also acquire Exclusive lock on deletable
* page and is acquiring and releasing exclusive lock on left page before.
* Left page was locked and released. Then parent and this page are
* locked. We acquire left page lock here only to mark page dirty after
* changing right pointer.
*/
lBuffer = ReadBufferExtended(gvs->index, MAIN_FORKNUM, leftBlkno,
RBM_NORMAL, gvs->strategy);
dBuffer = ReadBufferExtended(gvs->index, MAIN_FORKNUM, deleteBlkno,
RBM_NORMAL, gvs->strategy);
pBuffer = ReadBufferExtended(gvs->index, MAIN_FORKNUM, parentBlkno,
RBM_NORMAL, gvs->strategy);
LockBuffer(lBuffer, GIN_EXCLUSIVE);
START_CRIT_SECTION();
/* Unlink the page by changing left sibling's rightlink */
page = BufferGetPage(dBuffer);
rightlink = GinPageGetOpaque(page)->rightlink;
page = BufferGetPage(lBuffer);
GinPageGetOpaque(page)->rightlink = rightlink;
/* Delete downlink from parent */
parentPage = BufferGetPage(pBuffer);
#ifdef USE_ASSERT_CHECKING
do
{
PostingItem *tod = GinDataPageGetPostingItem(parentPage, myoff);
Assert(PostingItemGetBlockNumber(tod) == deleteBlkno);
} while (0);
#endif
GinPageDeletePostingItem(parentPage, myoff);
page = BufferGetPage(dBuffer);
/*
* we shouldn't change rightlink field to save workability of running
* search scan
*/
GinPageGetOpaque(page)->flags = GIN_DELETED;
MarkBufferDirty(pBuffer);
MarkBufferDirty(lBuffer);
MarkBufferDirty(dBuffer);
if (RelationNeedsWAL(gvs->index))
{
XLogRecPtr recptr;
ginxlogDeletePage data;
/*
* We can't pass REGBUF_STANDARD for the deleted page, because we
* didn't set pd_lower on pre-9.4 versions. The page might've been
* binary-upgraded from an older version, and hence not have pd_lower
* set correctly. Ditto for the left page, but removing the item from
* the parent updated its pd_lower, so we know that's OK at this
* point.
*/
XLogBeginInsert();
XLogRegisterBuffer(0, dBuffer, 0);
XLogRegisterBuffer(1, pBuffer, REGBUF_STANDARD);
XLogRegisterBuffer(2, lBuffer, 0);
data.parentOffset = myoff;
data.rightLink = GinPageGetOpaque(page)->rightlink;
XLogRegisterData((char *) &data, sizeof(ginxlogDeletePage));
recptr = XLogInsert(RM_GIN_ID, XLOG_GIN_DELETE_PAGE);
PageSetLSN(page, recptr);
PageSetLSN(parentPage, recptr);
PageSetLSN(BufferGetPage(lBuffer), recptr);
}
ReleaseBuffer(pBuffer);
UnlockReleaseBuffer(lBuffer);
ReleaseBuffer(dBuffer);
END_CRIT_SECTION();
gvs->result->pages_deleted++;
}
void
gin_desc(StringInfo buf, uint8 xl_info, char *rec)
{
uint8 info = xl_info & ~XLR_INFO_MASK;
switch (info)
{
case XLOG_GIN_CREATE_INDEX:
appendStringInfoString(buf, "Create index, ");
desc_node(buf, *(RelFileNode *) rec, GIN_ROOT_BLKNO);
break;
case XLOG_GIN_CREATE_PTREE:
appendStringInfoString(buf, "Create posting tree, ");
desc_node(buf, ((ginxlogCreatePostingTree *) rec)->node, ((ginxlogCreatePostingTree *) rec)->blkno);
break;
case XLOG_GIN_INSERT:
{
ginxlogInsert *xlrec = (ginxlogInsert *) rec;
char *payload = rec + sizeof(ginxlogInsert);
appendStringInfoString(buf, "Insert item, ");
desc_node(buf, xlrec->node, xlrec->blkno);
appendStringInfo(buf, " isdata: %c isleaf: %c",
(xlrec->flags & GIN_INSERT_ISDATA) ? 'T' : 'F',
(xlrec->flags & GIN_INSERT_ISLEAF) ? 'T' : 'F');
if (!(xlrec->flags & GIN_INSERT_ISLEAF))
{
BlockNumber leftChildBlkno;
BlockNumber rightChildBlkno;
leftChildBlkno = BlockIdGetBlockNumber((BlockId) payload);
payload += sizeof(BlockIdData);
rightChildBlkno = BlockIdGetBlockNumber((BlockId) payload);
payload += sizeof(BlockNumber);
appendStringInfo(buf, " children: %u/%u",
leftChildBlkno, rightChildBlkno);
}
if (!(xlrec->flags & GIN_INSERT_ISDATA))
appendStringInfo(buf, " isdelete: %c",
(((ginxlogInsertEntry *) payload)->isDelete) ? 'T' : 'F');
else if (xlrec->flags & GIN_INSERT_ISLEAF)
{
ginxlogRecompressDataLeaf *insertData =
(ginxlogRecompressDataLeaf *) payload;
appendStringInfo(buf, " unmodified: %u length: %u (compressed)",
insertData->unmodifiedsize,
insertData->length);
}
else
{
ginxlogInsertDataInternal *insertData = (ginxlogInsertDataInternal *) payload;
appendStringInfo(buf, " pitem: %u-%u/%u",
PostingItemGetBlockNumber(&insertData->newitem),
ItemPointerGetBlockNumber(&insertData->newitem.key),
ItemPointerGetOffsetNumber(&insertData->newitem.key));
}
}
break;
case XLOG_GIN_SPLIT:
{
ginxlogSplit *xlrec = (ginxlogSplit *) rec;
appendStringInfoString(buf, "Page split, ");
desc_node(buf, ((ginxlogSplit *) rec)->node, ((ginxlogSplit *) rec)->lblkno);
appendStringInfo(buf, " isrootsplit: %c", (((ginxlogSplit *) rec)->flags & GIN_SPLIT_ROOT) ? 'T' : 'F');
appendStringInfo(buf, " isdata: %c isleaf: %c",
(xlrec->flags & GIN_INSERT_ISDATA) ? 'T' : 'F',
(xlrec->flags & GIN_INSERT_ISLEAF) ? 'T' : 'F');
}
break;
case XLOG_GIN_VACUUM_PAGE:
appendStringInfoString(buf, "Vacuum page, ");
desc_node(buf, ((ginxlogVacuumPage *) rec)->node, ((ginxlogVacuumPage *) rec)->blkno);
break;
case XLOG_GIN_VACUUM_DATA_LEAF_PAGE:
{
ginxlogVacuumDataLeafPage *xlrec = (ginxlogVacuumDataLeafPage *) rec;
appendStringInfoString(buf, "Vacuum data leaf page, ");
desc_node(buf, xlrec->node, xlrec->blkno);
appendStringInfo(buf, " unmodified: %u length: %u",
xlrec->data.unmodifiedsize,
xlrec->data.length);
}
break;
case XLOG_GIN_DELETE_PAGE:
appendStringInfoString(buf, "Delete page, ");
desc_node(buf, ((ginxlogDeletePage *) rec)->node, ((ginxlogDeletePage *) rec)->blkno);
break;
case XLOG_GIN_UPDATE_META_PAGE:
appendStringInfoString(buf, "Update metapage, ");
desc_node(buf, ((ginxlogUpdateMeta *) rec)->node, GIN_METAPAGE_BLKNO);
break;
case XLOG_GIN_INSERT_LISTPAGE:
appendStringInfoString(buf, "Insert new list page, ");
desc_node(buf, ((ginxlogInsertListPage *) rec)->node, ((ginxlogInsertListPage *) rec)->blkno);
break;
case XLOG_GIN_DELETE_LISTPAGE:
appendStringInfo(buf, "Delete list pages (%d), ", ((ginxlogDeleteListPages *) rec)->ndeleted);
desc_node(buf, ((ginxlogDeleteListPages *) rec)->node, GIN_METAPAGE_BLKNO);
break;
default:
appendStringInfo(buf, "unknown gin op code %u", info);
break;
}
}
/*
* Delete a posting tree page.
*/
static void
ginDeletePage(GinVacuumState *gvs, BlockNumber deleteBlkno, BlockNumber leftBlkno,
BlockNumber parentBlkno, OffsetNumber myoff, bool isParentRoot)
{
Buffer dBuffer;
Buffer lBuffer;
Buffer pBuffer;
Page page,
parentPage;
BlockNumber rightlink;
/*
* Lock the pages in the same order as an insertion would, to avoid
* deadlocks: left, then right, then parent.
*/
lBuffer = ReadBufferExtended(gvs->index, MAIN_FORKNUM, leftBlkno,
RBM_NORMAL, gvs->strategy);
dBuffer = ReadBufferExtended(gvs->index, MAIN_FORKNUM, deleteBlkno,
RBM_NORMAL, gvs->strategy);
pBuffer = ReadBufferExtended(gvs->index, MAIN_FORKNUM, parentBlkno,
RBM_NORMAL, gvs->strategy);
LockBuffer(lBuffer, GIN_EXCLUSIVE);
LockBuffer(dBuffer, GIN_EXCLUSIVE);
if (!isParentRoot) /* parent is already locked by
* LockBufferForCleanup() */
LockBuffer(pBuffer, GIN_EXCLUSIVE);
START_CRIT_SECTION();
/* Unlink the page by changing left sibling's rightlink */
page = BufferGetPage(dBuffer);
rightlink = GinPageGetOpaque(page)->rightlink;
page = BufferGetPage(lBuffer);
GinPageGetOpaque(page)->rightlink = rightlink;
/* Delete downlink from parent */
parentPage = BufferGetPage(pBuffer);
#ifdef USE_ASSERT_CHECKING
do
{
PostingItem *tod = GinDataPageGetPostingItem(parentPage, myoff);
Assert(PostingItemGetBlockNumber(tod) == deleteBlkno);
} while (0);
#endif
GinPageDeletePostingItem(parentPage, myoff);
page = BufferGetPage(dBuffer);
/*
* we shouldn't change rightlink field to save workability of running
* search scan
*/
GinPageGetOpaque(page)->flags = GIN_DELETED;
MarkBufferDirty(pBuffer);
MarkBufferDirty(lBuffer);
MarkBufferDirty(dBuffer);
if (RelationNeedsWAL(gvs->index))
{
XLogRecPtr recptr;
XLogRecData rdata[4];
ginxlogDeletePage data;
data.node = gvs->index->rd_node;
data.blkno = deleteBlkno;
data.parentBlkno = parentBlkno;
data.parentOffset = myoff;
data.leftBlkno = leftBlkno;
data.rightLink = GinPageGetOpaque(page)->rightlink;
/*
* We can't pass buffer_std = TRUE, because we didn't set pd_lower on
* pre-9.4 versions. The page might've been binary-upgraded from an
* older version, and hence not have pd_lower set correctly. Ditto for
* the left page, but removing the item from the parent updated its
* pd_lower, so we know that's OK at this point.
*/
rdata[0].buffer = dBuffer;
rdata[0].buffer_std = FALSE;
rdata[0].data = NULL;
rdata[0].len = 0;
rdata[0].next = rdata + 1;
rdata[1].buffer = pBuffer;
rdata[1].buffer_std = TRUE;
rdata[1].data = NULL;
rdata[1].len = 0;
rdata[1].next = rdata + 2;
rdata[2].buffer = lBuffer;
rdata[2].buffer_std = FALSE;
rdata[2].data = NULL;
rdata[2].len = 0;
rdata[2].next = rdata + 3;
rdata[3].buffer = InvalidBuffer;
rdata[3].buffer_std = FALSE;
rdata[3].len = sizeof(ginxlogDeletePage);
rdata[3].data = (char *) &data;
rdata[3].next = NULL;
recptr = XLogInsert(RM_GIN_ID, XLOG_GIN_DELETE_PAGE, rdata);
PageSetLSN(page, recptr);
PageSetLSN(parentPage, recptr);
PageSetLSN(BufferGetPage(lBuffer), recptr);
}
if (!isParentRoot)
LockBuffer(pBuffer, GIN_UNLOCK);
ReleaseBuffer(pBuffer);
UnlockReleaseBuffer(lBuffer);
UnlockReleaseBuffer(dBuffer);
END_CRIT_SECTION();
gvs->result->pages_deleted++;
}