/*
* MultiRelease -- release a multi-level lock
*
* Returns: TRUE if successful, FALSE otherwise.
*/
bool
MultiRelease(LockTableId tableId,
LOCKTAG *tag,
LOCKT lockt,
LOCK_LEVEL level)
{
LOCKT locks[N_LEVELS];
int i,status;
LOCKTAG xxTag, *tmpTag = &xxTag;
/*
* same level scheme as MultiAcquire().
*/
switch (level) {
case RELN_LEVEL:
locks[0] = lockt;
locks[1] = NO_LOCK;
locks[2] = NO_LOCK;
break;
case PAGE_LEVEL:
locks[0] = lockt + INTENT;
locks[1] = lockt;
locks[2] = NO_LOCK;
break;
case TUPLE_LEVEL:
locks[0] = lockt + INTENT;
locks[1] = lockt + INTENT;
locks[2] = lockt;
break;
default:
elog(WARN,"MultiRelease: bad lockt");
}
/*
* again, construct the tag on the fly. This time, however,
* we release the locks in the REVERSE order -- from lowest
* level to highest level.
*
* Must zero out the tag to set padding byes to zero and ensure
* hashing consistency.
*/
memset(tmpTag, 0, sizeof(*tmpTag));
tmpTag->relId = tag->relId;
tmpTag->dbId = tag->dbId;
for (i=(N_LEVELS-1); i>=0; i--) {
if (locks[i] != NO_LOCK) {
switch (i) {
case RELN_LEVEL:
/* -------------
* Set the block # and offset to invalid
* -------------
*/
BlockIdSet(&(tmpTag->tupleId.ip_blkid), InvalidBlockNumber);
tmpTag->tupleId.ip_posid = InvalidOffsetNumber;
break;
case PAGE_LEVEL:
/* -------------
* Copy the block #, set the offset to invalid
* -------------
*/
BlockIdCopy(&(tmpTag->tupleId.ip_blkid),
&(tag->tupleId.ip_blkid));
tmpTag->tupleId.ip_posid = InvalidOffsetNumber;
break;
case TUPLE_LEVEL:
ItemPointerCopy(&tmpTag->tupleId, &tag->tupleId);
break;
}
status = LockRelease(tableId, tmpTag, locks[i]);
if (! status) {
elog(WARN,"MultiRelease: couldn't release after error");
}
}
}
/* shouldn't reach here */
return false;
}
/*
* Insert a new item to a page.
*
* Returns true if the insertion was finished. On false, the page was split and
* the parent needs to be updated. (a root split returns true as it doesn't
* need any further action by the caller to complete)
*
* When inserting a downlink to a internal page, 'childbuf' contains the
* child page that was split. Its GIN_INCOMPLETE_SPLIT flag will be cleared
* atomically with the insert. Also, the existing item at the given location
* is updated to point to 'updateblkno'.
*
* stack->buffer is locked on entry, and is kept locked.
*/
static bool
ginPlaceToPage(GinBtree btree, GinBtreeStack *stack,
void *insertdata, BlockNumber updateblkno,
Buffer childbuf, GinStatsData *buildStats)
{
Page page = BufferGetPage(stack->buffer);
XLogRecData *payloadrdata;
bool fit;
uint16 xlflags = 0;
Page childpage = NULL;
if (GinPageIsData(page))
xlflags |= GIN_INSERT_ISDATA;
if (GinPageIsLeaf(page))
{
xlflags |= GIN_INSERT_ISLEAF;
Assert(!BufferIsValid(childbuf));
Assert(updateblkno == InvalidBlockNumber);
}
else
{
Assert(BufferIsValid(childbuf));
Assert(updateblkno != InvalidBlockNumber);
childpage = BufferGetPage(childbuf);
}
/*
* Try to put the incoming tuple on the page. If it doesn't fit,
* placeToPage method will return false and leave the page unmodified, and
* we'll have to split the page.
*/
START_CRIT_SECTION();
fit = btree->placeToPage(btree, stack->buffer, stack->off,
insertdata, updateblkno,
&payloadrdata);
if (fit)
{
MarkBufferDirty(stack->buffer);
/* An insert to an internal page finishes the split of the child. */
if (childbuf != InvalidBuffer)
{
GinPageGetOpaque(childpage)->flags &= ~GIN_INCOMPLETE_SPLIT;
MarkBufferDirty(childbuf);
}
if (RelationNeedsWAL(btree->index))
{
XLogRecPtr recptr;
XLogRecData rdata[3];
ginxlogInsert xlrec;
BlockIdData childblknos[2];
xlrec.node = btree->index->rd_node;
xlrec.blkno = BufferGetBlockNumber(stack->buffer);
xlrec.offset = stack->off;
xlrec.flags = xlflags;
rdata[0].buffer = InvalidBuffer;
rdata[0].data = (char *) &xlrec;
rdata[0].len = sizeof(ginxlogInsert);
/*
* Log information about child if this was an insertion of a
* downlink.
*/
if (childbuf != InvalidBuffer)
{
rdata[0].next = &rdata[1];
BlockIdSet(&childblknos[0], BufferGetBlockNumber(childbuf));
BlockIdSet(&childblknos[1], GinPageGetOpaque(childpage)->rightlink);
rdata[1].buffer = InvalidBuffer;
rdata[1].data = (char *) childblknos;
rdata[1].len = sizeof(BlockIdData) * 2;
rdata[1].next = &rdata[2];
rdata[2].buffer = childbuf;
rdata[2].buffer_std = false;
rdata[2].data = NULL;
rdata[2].len = 0;
rdata[2].next = payloadrdata;
}
else
rdata[0].next = payloadrdata;
recptr = XLogInsert(RM_GIN_ID, XLOG_GIN_INSERT, rdata);
PageSetLSN(page, recptr);
if (childbuf != InvalidBuffer)
PageSetLSN(childpage, recptr);
}
END_CRIT_SECTION();
return true;
}
else
{
/* Didn't fit, have to split */
Buffer rbuffer;
Page newlpage;
BlockNumber savedRightLink;
Page rpage;
XLogRecData rdata[2];
ginxlogSplit data;
Buffer lbuffer = InvalidBuffer;
Page newrootpg = NULL;
END_CRIT_SECTION();
rbuffer = GinNewBuffer(btree->index);
/* During index build, count the new page */
if (buildStats)
{
if (btree->isData)
buildStats->nDataPages++;
else
buildStats->nEntryPages++;
}
savedRightLink = GinPageGetOpaque(page)->rightlink;
/*
* newlpage is a pointer to memory page, it is not associated with a
* buffer. stack->buffer is not touched yet.
*/
newlpage = btree->splitPage(btree, stack->buffer, rbuffer, stack->off,
insertdata, updateblkno,
&payloadrdata);
data.node = btree->index->rd_node;
data.rblkno = BufferGetBlockNumber(rbuffer);
data.flags = xlflags;
if (childbuf != InvalidBuffer)
{
Page childpage = BufferGetPage(childbuf);
GinPageGetOpaque(childpage)->flags &= ~GIN_INCOMPLETE_SPLIT;
data.leftChildBlkno = BufferGetBlockNumber(childbuf);
data.rightChildBlkno = GinPageGetOpaque(childpage)->rightlink;
}
else
data.leftChildBlkno = data.rightChildBlkno = InvalidBlockNumber;
rdata[0].buffer = InvalidBuffer;
rdata[0].data = (char *) &data;
rdata[0].len = sizeof(ginxlogSplit);
if (childbuf != InvalidBuffer)
{
rdata[0].next = &rdata[1];
rdata[1].buffer = childbuf;
rdata[1].buffer_std = false;
rdata[1].data = NULL;
rdata[1].len = 0;
rdata[1].next = payloadrdata;
}
else
rdata[0].next = payloadrdata;
rpage = BufferGetPage(rbuffer);
if (stack->parent == NULL)
{
/*
* split root, so we need to allocate new left page and place
* pointer on root to left and right page
*/
lbuffer = GinNewBuffer(btree->index);
/* During index build, count the newly-added root page */
if (buildStats)
{
if (btree->isData)
buildStats->nDataPages++;
else
buildStats->nEntryPages++;
}
/*
* root never has a right-link, so we borrow the rrlink field to
* store the root block number.
*/
data.rrlink = BufferGetBlockNumber(stack->buffer);
data.lblkno = BufferGetBlockNumber(lbuffer);
data.flags |= GIN_SPLIT_ROOT;
GinPageGetOpaque(rpage)->rightlink = InvalidBlockNumber;
GinPageGetOpaque(newlpage)->rightlink = BufferGetBlockNumber(rbuffer);
/*
* Construct a new root page containing downlinks to the new left
* and right pages. (do this in a temporary copy first rather
* than overwriting the original page directly, so that we can still
* abort gracefully if this fails.)
*/
newrootpg = PageGetTempPage(rpage);
GinInitPage(newrootpg, GinPageGetOpaque(newlpage)->flags & ~GIN_LEAF, BLCKSZ);
btree->fillRoot(btree, newrootpg,
BufferGetBlockNumber(lbuffer), newlpage,
BufferGetBlockNumber(rbuffer), rpage);
}
else
{
/* split non-root page */
data.rrlink = savedRightLink;
data.lblkno = BufferGetBlockNumber(stack->buffer);
GinPageGetOpaque(rpage)->rightlink = savedRightLink;
GinPageGetOpaque(newlpage)->flags |= GIN_INCOMPLETE_SPLIT;
GinPageGetOpaque(newlpage)->rightlink = BufferGetBlockNumber(rbuffer);
}
/*
* Ok, we have the new contents of the left page in a temporary copy
* now (newlpage), and the newly-allocated right block has been filled
* in. The original page is still unchanged.
*
* If this is a root split, we also have a temporary page containing
* the new contents of the root. Copy the new left page to a
* newly-allocated block, and initialize the (original) root page the
* new copy. Otherwise, copy over the temporary copy of the new left
* page over the old left page.
*/
START_CRIT_SECTION();
MarkBufferDirty(rbuffer);
if (stack->parent == NULL)
{
PageRestoreTempPage(newlpage, BufferGetPage(lbuffer));
MarkBufferDirty(lbuffer);
newlpage = newrootpg;
}
PageRestoreTempPage(newlpage, BufferGetPage(stack->buffer));
MarkBufferDirty(stack->buffer);
/* write WAL record */
if (RelationNeedsWAL(btree->index))
{
XLogRecPtr recptr;
recptr = XLogInsert(RM_GIN_ID, XLOG_GIN_SPLIT, rdata);
PageSetLSN(BufferGetPage(stack->buffer), recptr);
PageSetLSN(rpage, recptr);
if (stack->parent == NULL)
PageSetLSN(BufferGetPage(lbuffer), recptr);
}
END_CRIT_SECTION();
/*
* We can release the lock on the right page now, but keep the
* original buffer locked.
*/
UnlockReleaseBuffer(rbuffer);
if (stack->parent == NULL)
UnlockReleaseBuffer(lbuffer);
/*
* If we split the root, we're done. Otherwise the split is not
* complete until the downlink for the new page has been inserted to
* the parent.
*/
if (stack->parent == NULL)
return true;
else
return false;
}
}
/*
* MultiAcquire -- acquire multi level lock at requested level
*
* Returns: TRUE if lock is set, FALSE if not
* Side Effects:
*/
bool
MultiAcquire(LockTableId tableId,
LOCKTAG *tag,
LOCKT lockt,
LOCK_LEVEL level)
{
LOCKT locks[N_LEVELS];
int i,status;
LOCKTAG xxTag, *tmpTag = &xxTag;
int retStatus = TRUE;
/*
* Three levels implemented. If we set a low level (e.g. Tuple)
* lock, we must set INTENT locks on the higher levels. The
* intent lock detects conflicts between the low level lock
* and an existing high level lock. For example, setting a
* write lock on a tuple in a relation is disallowed if there
* is an existing read lock on the entire relation. The
* write lock would set a WRITE + INTENT lock on the relation
* and that lock would conflict with the read.
*/
switch (level) {
case RELN_LEVEL:
locks[0] = lockt;
locks[1] = NO_LOCK;
locks[2] = NO_LOCK;
break;
case PAGE_LEVEL:
locks[0] = lockt + INTENT;
locks[1] = lockt;
locks[2] = NO_LOCK;
break;
case TUPLE_LEVEL:
locks[0] = lockt + INTENT;
locks[1] = lockt + INTENT;
locks[2] = lockt;
break;
default:
elog(WARN,"MultiAcquire: bad lock level");
return(FALSE);
}
/*
* construct a new tag as we go. Always loop through all levels,
* but if we arent' seting a low level lock, locks[i] is set to
* NO_LOCK for the lower levels. Always start from the highest
* level and go to the lowest level.
*/
memset(tmpTag,0,sizeof(*tmpTag));
tmpTag->relId = tag->relId;
tmpTag->dbId = tag->dbId;
for (i=0;i<N_LEVELS;i++) {
if (locks[i] != NO_LOCK) {
switch (i) {
case RELN_LEVEL:
/* -------------
* Set the block # and offset to invalid
* -------------
*/
BlockIdSet(&(tmpTag->tupleId.ip_blkid), InvalidBlockNumber);
tmpTag->tupleId.ip_posid = InvalidOffsetNumber;
break;
case PAGE_LEVEL:
/* -------------
* Copy the block #, set the offset to invalid
* -------------
*/
BlockIdCopy(&(tmpTag->tupleId.ip_blkid),
&(tag->tupleId.ip_blkid));
tmpTag->tupleId.ip_posid = InvalidOffsetNumber;
break;
case TUPLE_LEVEL:
/* --------------
* Copy the entire tuple id.
* --------------
*/
ItemPointerCopy(&tmpTag->tupleId, &tag->tupleId);
break;
}
status = LockAcquire(tableId, tmpTag, locks[i]);
if (! status) {
/* failed for some reason. Before returning we have
* to release all of the locks we just acquired.
* MultiRelease(xx,xx,xx, i) means release starting from
* the last level lock we successfully acquired
*/
retStatus = FALSE;
(void) MultiRelease(tableId, tag, lockt, i);
/* now leave the loop. Don't try for any more locks */
break;
}
}
}
return(retStatus);
}
/*
* Insert a new item to a page.
*
* Returns true if the insertion was finished. On false, the page was split and
* the parent needs to be updated. (A root split returns true as it doesn't
* need any further action by the caller to complete.)
*
* When inserting a downlink to an internal page, 'childbuf' contains the
* child page that was split. Its GIN_INCOMPLETE_SPLIT flag will be cleared
* atomically with the insert. Also, the existing item at offset stack->off
* in the target page is updated to point to updateblkno.
*
* stack->buffer is locked on entry, and is kept locked.
* Likewise for childbuf, if given.
*/
static bool
ginPlaceToPage(GinBtree btree, GinBtreeStack *stack,
void *insertdata, BlockNumber updateblkno,
Buffer childbuf, GinStatsData *buildStats)
{
Page page = BufferGetPage(stack->buffer);
bool result;
GinPlaceToPageRC rc;
uint16 xlflags = 0;
Page childpage = NULL;
Page newlpage = NULL,
newrpage = NULL;
void *ptp_workspace = NULL;
XLogRecData payloadrdata[10];
MemoryContext tmpCxt;
MemoryContext oldCxt;
/*
* We do all the work of this function and its subfunctions in a temporary
* memory context. This avoids leakages and simplifies APIs, since some
* subfunctions allocate storage that has to survive until we've finished
* the WAL insertion.
*/
tmpCxt = AllocSetContextCreate(CurrentMemoryContext,
"ginPlaceToPage temporary context",
ALLOCSET_DEFAULT_MINSIZE,
ALLOCSET_DEFAULT_INITSIZE,
ALLOCSET_DEFAULT_MAXSIZE);
oldCxt = MemoryContextSwitchTo(tmpCxt);
if (GinPageIsData(page))
xlflags |= GIN_INSERT_ISDATA;
if (GinPageIsLeaf(page))
{
xlflags |= GIN_INSERT_ISLEAF;
Assert(!BufferIsValid(childbuf));
Assert(updateblkno == InvalidBlockNumber);
}
else
{
Assert(BufferIsValid(childbuf));
Assert(updateblkno != InvalidBlockNumber);
childpage = BufferGetPage(childbuf);
}
/*
* See if the incoming tuple will fit on the page. beginPlaceToPage will
* decide if the page needs to be split, and will compute the split
* contents if so. See comments for beginPlaceToPage and execPlaceToPage
* functions for more details of the API here.
*/
rc = btree->beginPlaceToPage(btree, stack->buffer, stack,
insertdata, updateblkno,
&ptp_workspace,
&newlpage, &newrpage,
payloadrdata);
if (rc == GPTP_NO_WORK)
{
/* Nothing to do */
result = true;
}
else if (rc == GPTP_INSERT)
{
/* It will fit, perform the insertion */
START_CRIT_SECTION();
/* Perform the page update, and set up WAL data about it */
btree->execPlaceToPage(btree, stack->buffer, stack,
insertdata, updateblkno,
ptp_workspace, payloadrdata);
MarkBufferDirty(stack->buffer);
/* An insert to an internal page finishes the split of the child. */
if (BufferIsValid(childbuf))
{
GinPageGetOpaque(childpage)->flags &= ~GIN_INCOMPLETE_SPLIT;
MarkBufferDirty(childbuf);
}
if (RelationNeedsWAL(btree->index))
{
XLogRecPtr recptr;
XLogRecData rdata[3];
ginxlogInsert xlrec;
BlockIdData childblknos[2];
xlrec.node = btree->index->rd_node;
xlrec.blkno = BufferGetBlockNumber(stack->buffer);
xlrec.flags = xlflags;
rdata[0].buffer = InvalidBuffer;
rdata[0].data = (char *) &xlrec;
rdata[0].len = sizeof(ginxlogInsert);
/*
* Log information about child if this was an insertion of a
* downlink.
*/
if (BufferIsValid(childbuf))
{
rdata[0].next = &rdata[1];
BlockIdSet(&childblknos[0], BufferGetBlockNumber(childbuf));
BlockIdSet(&childblknos[1], GinPageGetOpaque(childpage)->rightlink);
rdata[1].buffer = InvalidBuffer;
rdata[1].data = (char *) childblknos;
rdata[1].len = sizeof(BlockIdData) * 2;
rdata[1].next = &rdata[2];
rdata[2].buffer = childbuf;
rdata[2].buffer_std = true;
rdata[2].data = NULL;
rdata[2].len = 0;
rdata[2].next = payloadrdata;
}
else
rdata[0].next = payloadrdata;
recptr = XLogInsert(RM_GIN_ID, XLOG_GIN_INSERT, rdata);
PageSetLSN(page, recptr);
if (BufferIsValid(childbuf))
PageSetLSN(childpage, recptr);
}
END_CRIT_SECTION();
/* Insertion is complete. */
result = true;
}
else if (rc == GPTP_SPLIT)
{
/*
* Didn't fit, need to split. The split has been computed in newlpage
* and newrpage, which are pointers to palloc'd pages, not associated
* with buffers. stack->buffer is not touched yet.
*/
Buffer rbuffer;
BlockNumber savedRightLink;
ginxlogSplit data;
Buffer lbuffer = InvalidBuffer;
Page newrootpg = NULL;
/* Get a new index page to become the right page */
rbuffer = GinNewBuffer(btree->index);
/* During index build, count the new page */
if (buildStats)
{
if (btree->isData)
buildStats->nDataPages++;
else
buildStats->nEntryPages++;
}
savedRightLink = GinPageGetOpaque(page)->rightlink;
/* Begin setting up WAL record (which we might not use) */
data.node = btree->index->rd_node;
data.rblkno = BufferGetBlockNumber(rbuffer);
data.flags = xlflags;
if (BufferIsValid(childbuf))
{
data.leftChildBlkno = BufferGetBlockNumber(childbuf);
data.rightChildBlkno = GinPageGetOpaque(childpage)->rightlink;
}
else
data.leftChildBlkno = data.rightChildBlkno = InvalidBlockNumber;
if (stack->parent == NULL)
{
/*
* splitting the root, so we need to allocate new left page and
* place pointers to left and right page on root page.
*/
lbuffer = GinNewBuffer(btree->index);
/* During index build, count the new left page */
if (buildStats)
{
if (btree->isData)
buildStats->nDataPages++;
else
buildStats->nEntryPages++;
}
/*
* root never has a right-link, so we borrow the rrlink field to
* store the root block number.
*/
data.rrlink = BufferGetBlockNumber(stack->buffer);
data.lblkno = BufferGetBlockNumber(lbuffer);
data.flags |= GIN_SPLIT_ROOT;
GinPageGetOpaque(newrpage)->rightlink = InvalidBlockNumber;
GinPageGetOpaque(newlpage)->rightlink = BufferGetBlockNumber(rbuffer);
/*
* Construct a new root page containing downlinks to the new left
* and right pages. (Do this in a temporary copy rather than
* overwriting the original page directly, since we're not in the
* critical section yet.)
*/
newrootpg = PageGetTempPage(newrpage);
GinInitPage(newrootpg, GinPageGetOpaque(newlpage)->flags & ~(GIN_LEAF | GIN_COMPRESSED), BLCKSZ);
btree->fillRoot(btree, newrootpg,
BufferGetBlockNumber(lbuffer), newlpage,
BufferGetBlockNumber(rbuffer), newrpage);
}
else
{
/* splitting a non-root page */
data.rrlink = savedRightLink;
data.lblkno = BufferGetBlockNumber(stack->buffer);
GinPageGetOpaque(newrpage)->rightlink = savedRightLink;
GinPageGetOpaque(newlpage)->flags |= GIN_INCOMPLETE_SPLIT;
GinPageGetOpaque(newlpage)->rightlink = BufferGetBlockNumber(rbuffer);
}
/*
* OK, we have the new contents of the left page in a temporary copy
* now (newlpage), and likewise for the new contents of the
* newly-allocated right block. The original page is still unchanged.
*
* If this is a root split, we also have a temporary page containing
* the new contents of the root.
*/
START_CRIT_SECTION();
MarkBufferDirty(rbuffer);
MarkBufferDirty(stack->buffer);
/*
* Restore the temporary copies over the real buffers.
*/
if (stack->parent == NULL)
{
/* Splitting the root, three pages to update */
MarkBufferDirty(lbuffer);
memcpy(page, newrootpg, BLCKSZ);
memcpy(BufferGetPage(lbuffer), newlpage, BLCKSZ);
memcpy(BufferGetPage(rbuffer), newrpage, BLCKSZ);
}
else
{
/* Normal split, only two pages to update */
memcpy(page, newlpage, BLCKSZ);
memcpy(BufferGetPage(rbuffer), newrpage, BLCKSZ);
}
/* We also clear childbuf's INCOMPLETE_SPLIT flag, if passed */
if (BufferIsValid(childbuf))
{
GinPageGetOpaque(childpage)->flags &= ~GIN_INCOMPLETE_SPLIT;
MarkBufferDirty(childbuf);
}
/* write WAL record */
if (RelationNeedsWAL(btree->index))
{
XLogRecData rdata[2];
XLogRecPtr recptr;
rdata[0].buffer = InvalidBuffer;
rdata[0].data = (char *) &data;
rdata[0].len = sizeof(ginxlogSplit);
if (BufferIsValid(childbuf))
{
rdata[0].next = &rdata[1];
rdata[1].buffer = childbuf;
rdata[1].buffer_std = true;
rdata[1].data = NULL;
rdata[1].len = 0;
rdata[1].next = payloadrdata;
}
else
rdata[0].next = payloadrdata;
recptr = XLogInsert(RM_GIN_ID, XLOG_GIN_SPLIT, rdata);
PageSetLSN(page, recptr);
PageSetLSN(BufferGetPage(rbuffer), recptr);
if (stack->parent == NULL)
PageSetLSN(BufferGetPage(lbuffer), recptr);
if (BufferIsValid(childbuf))
PageSetLSN(childpage, recptr);
}
END_CRIT_SECTION();
/*
* We can release the locks/pins on the new pages now, but keep
* stack->buffer locked. childbuf doesn't get unlocked either.
*/
UnlockReleaseBuffer(rbuffer);
if (stack->parent == NULL)
UnlockReleaseBuffer(lbuffer);
/*
* If we split the root, we're done. Otherwise the split is not
* complete until the downlink for the new page has been inserted to
* the parent.
*/
result = (stack->parent == NULL);
}
else
{
elog(ERROR, "invalid return code from GIN placeToPage method: %d", rc);
result = false; /* keep compiler quiet */
}
/* Clean up temp context */
MemoryContextSwitchTo(oldCxt);
MemoryContextDelete(tmpCxt);
return result;
}
