/* ----------------------------------------------------------------
* caql_endscan()
* free all resources associated with the scan, including tuples,
* tables and locks.
* NOTE: this function is *not* a "drop-in" replacement for
* ReleaseSysCache. ReleaseSysCache is only called for valid tuples,
* but you must always call endscan, even if getnext never returned a
* valid tuple.
* ----------------------------------------------------------------
*/
void caql_endscan(cqContext *pCtx)
{
if (pCtx->cq_indstate) /* close the indexes if they were open */
CatalogCloseIndexes(pCtx->cq_indstate);
pCtx->cq_indstate = NULL;
pCtx->cq_bScanBlock = false; /* scan block has ended */
if (pCtx->cq_freeScan)
{
if (!pCtx->cq_usesyscache)
systable_endscan(pCtx->cq_sysScan);
else
{
/* XXX XXX: no need to release if never fetched a valid tuple */
if (HeapTupleIsValid(pCtx->cq_lasttup))
ReleaseSysCache(pCtx->cq_lasttup);
}
}
caql_heapclose(pCtx);
if (pCtx->cq_free) /* free dynamic context */
pfree(pCtx);
pCtx->cq_freeScan = false;
pCtx->cq_free = false;
}
/*
* CatalogUpdateIndexes - do all the indexing work for a new catalog tuple
*
* This is a convenience routine for the common case where we only need
* to insert or update a single tuple in a system catalog. Avoid using it for
* multiple tuples, since opening the indexes and building the index info
* structures is moderately expensive.
*/
void
CatalogUpdateIndexes(Relation heapRel, HeapTuple heapTuple)
{
CatalogIndexState indstate;
indstate = CatalogOpenIndexes(heapRel);
CatalogIndexInsert(indstate, heapTuple);
CatalogCloseIndexes(indstate);
}
/*
* CatalogTupleUpdate - do heap and indexing work for updating a catalog tuple
*
* Update the tuple identified by "otid", replacing it with the data in "tup".
*
* This is a convenience routine for the common case of updating a single
* tuple in a system catalog; it updates one heap tuple, keeping indexes
* current. Avoid using it for multiple tuples, since opening the indexes
* and building the index info structures is moderately expensive.
* (Use CatalogTupleUpdateWithInfo in such cases.)
*/
void
CatalogTupleUpdate(Relation heapRel, ItemPointer otid, HeapTuple tup)
{
CatalogIndexState indstate;
indstate = CatalogOpenIndexes(heapRel);
simple_heap_update(heapRel, otid, tup);
CatalogIndexInsert(indstate, tup);
CatalogCloseIndexes(indstate);
}
/*
* CatalogTupleInsert - do heap and indexing work for a new catalog tuple
*
* Insert the tuple data in "tup" into the specified catalog relation.
* The Oid of the inserted tuple is returned.
*
* This is a convenience routine for the common case of inserting a single
* tuple in a system catalog; it inserts a new heap tuple, keeping indexes
* current. Avoid using it for multiple tuples, since opening the indexes
* and building the index info structures is moderately expensive.
* (Use CatalogTupleInsertWithInfo in such cases.)
*/
void
CatalogTupleInsert(Relation heapRel, HeapTuple tup)
{
CatalogIndexState indstate;
indstate = CatalogOpenIndexes(heapRel);
simple_heap_insert(heapRel, tup);
CatalogIndexInsert(indstate, tup);
CatalogCloseIndexes(indstate);
}
/* ----------------------------------------------------------------
* AppendAttributeTuples
* ----------------------------------------------------------------
*/
static void
AppendAttributeTuples(Relation indexRelation, int numatts)
{
Relation pg_attribute;
CatalogIndexState indstate;
TupleDesc indexTupDesc;
HeapTuple new_tuple;
int i;
/*
* open the attribute relation and its indexes
*/
pg_attribute = heap_openr(AttributeRelationName, RowExclusiveLock);
indstate = CatalogOpenIndexes(pg_attribute);
/*
* insert data from new index's tupdesc into pg_attribute
*/
indexTupDesc = RelationGetDescr(indexRelation);
for (i = 0; i < numatts; i++)
{
/*
* There used to be very grotty code here to set these fields, but
* I think it's unnecessary. They should be set already.
*/
Assert(indexTupDesc->attrs[i]->attnum == i + 1);
Assert(indexTupDesc->attrs[i]->attcacheoff == -1);
new_tuple = heap_addheader(Natts_pg_attribute,
false,
ATTRIBUTE_TUPLE_SIZE,
(void *) indexTupDesc->attrs[i]);
simple_heap_insert(pg_attribute, new_tuple);
CatalogIndexInsert(indstate, new_tuple);
heap_freetuple(new_tuple);
}
CatalogCloseIndexes(indstate);
heap_close(pg_attribute, RowExclusiveLock);
}
/*
* This is a copy of swap_relation_files in cluster.c, but it also swaps
* relfrozenxid.
*/
static void
swap_heap_or_index_files(Oid r1, Oid r2)
{
Relation relRelation;
HeapTuple reltup1,
reltup2;
Form_pg_class relform1,
relform2;
Oid swaptemp;
CatalogIndexState indstate;
/* We need writable copies of both pg_class tuples. */
relRelation = heap_open(RelationRelationId, RowExclusiveLock);
reltup1 = SearchSysCacheCopy(RELOID,
ObjectIdGetDatum(r1),
0, 0, 0);
if (!HeapTupleIsValid(reltup1))
elog(ERROR, "cache lookup failed for relation %u", r1);
relform1 = (Form_pg_class) GETSTRUCT(reltup1);
reltup2 = SearchSysCacheCopy(RELOID,
ObjectIdGetDatum(r2),
0, 0, 0);
if (!HeapTupleIsValid(reltup2))
elog(ERROR, "cache lookup failed for relation %u", r2);
relform2 = (Form_pg_class) GETSTRUCT(reltup2);
Assert(relform1->relkind == relform2->relkind);
/*
* Actually swap the fields in the two tuples
*/
swaptemp = relform1->relfilenode;
relform1->relfilenode = relform2->relfilenode;
relform2->relfilenode = swaptemp;
swaptemp = relform1->reltablespace;
relform1->reltablespace = relform2->reltablespace;
relform2->reltablespace = swaptemp;
swaptemp = relform1->reltoastrelid;
relform1->reltoastrelid = relform2->reltoastrelid;
relform2->reltoastrelid = swaptemp;
/* set rel1's frozen Xid to larger one */
if (TransactionIdIsNormal(relform1->relfrozenxid))
{
if (TransactionIdFollows(relform1->relfrozenxid,
relform2->relfrozenxid))
relform1->relfrozenxid = relform2->relfrozenxid;
else
relform2->relfrozenxid = relform1->relfrozenxid;
}
/* swap size statistics too, since new rel has freshly-updated stats */
{
#if PG_VERSION_NUM >= 90300
int32 swap_pages;
#else
int4 swap_pages;
#endif
float4 swap_tuples;
swap_pages = relform1->relpages;
relform1->relpages = relform2->relpages;
relform2->relpages = swap_pages;
swap_tuples = relform1->reltuples;
relform1->reltuples = relform2->reltuples;
relform2->reltuples = swap_tuples;
}
/* Update the tuples in pg_class */
simple_heap_update(relRelation, &reltup1->t_self, reltup1);
simple_heap_update(relRelation, &reltup2->t_self, reltup2);
/* Keep system catalogs current */
indstate = CatalogOpenIndexes(relRelation);
CatalogIndexInsert(indstate, reltup1);
CatalogIndexInsert(indstate, reltup2);
CatalogCloseIndexes(indstate);
/*
* If we have toast tables associated with the relations being swapped,
* change their dependency links to re-associate them with their new
* owning relations. Otherwise the wrong one will get dropped ...
*
* NOTE: it is possible that only one table has a toast table; this can
* happen in CLUSTER if there were dropped columns in the old table, and
* in ALTER TABLE when adding or changing type of columns.
*
* NOTE: at present, a TOAST table's only dependency is the one on its
* owning table. If more are ever created, we'd need to use something
* more selective than deleteDependencyRecordsFor() to get rid of only the
* link we want.
*/
if (relform1->reltoastrelid || relform2->reltoastrelid)
{
ObjectAddress baseobject,
toastobject;
long count;
/* Delete old dependencies */
if (relform1->reltoastrelid)
{
count = deleteDependencyRecordsFor(RelationRelationId,
relform1->reltoastrelid,
false);
if (count != 1)
elog(ERROR, "expected one dependency record for TOAST table, found %ld",
count);
}
if (relform2->reltoastrelid)
{
count = deleteDependencyRecordsFor(RelationRelationId,
relform2->reltoastrelid,
false);
if (count != 1)
elog(ERROR, "expected one dependency record for TOAST table, found %ld",
count);
}
/* Register new dependencies */
baseobject.classId = RelationRelationId;
baseobject.objectSubId = 0;
toastobject.classId = RelationRelationId;
toastobject.objectSubId = 0;
if (relform1->reltoastrelid)
{
baseobject.objectId = r1;
toastobject.objectId = relform1->reltoastrelid;
recordDependencyOn(&toastobject, &baseobject, DEPENDENCY_INTERNAL);
}
if (relform2->reltoastrelid)
{
baseobject.objectId = r2;
toastobject.objectId = relform2->reltoastrelid;
recordDependencyOn(&toastobject, &baseobject, DEPENDENCY_INTERNAL);
}
}
/*
* Blow away the old relcache entries now. We need this kluge because
* relcache.c keeps a link to the smgr relation for the physical file, and
* that will be out of date as soon as we do CommandCounterIncrement.
* Whichever of the rels is the second to be cleared during cache
* invalidation will have a dangling reference to an already-deleted smgr
* relation. Rather than trying to avoid this by ordering operations just
* so, it's easiest to not have the relcache entries there at all.
* (Fortunately, since one of the entries is local in our transaction,
* it's sufficient to clear out our own relcache this way; the problem
* cannot arise for other backends when they see our update on the
* non-local relation.)
*/
RelationForgetRelation(r1);
RelationForgetRelation(r2);
/* Clean up. */
heap_freetuple(reltup1);
heap_freetuple(reltup2);
heap_close(relRelation, RowExclusiveLock);
}
void
inv_truncate(LargeObjectDesc *obj_desc, int len)
{
int32 pageno = (int32) (len / LOBLKSIZE);
int off;
ScanKeyData skey[2];
SysScanDesc sd;
HeapTuple oldtuple;
Form_pg_largeobject olddata;
struct
{
bytea hdr;
char data[LOBLKSIZE]; /* make struct big enough */
int32 align_it; /* ensure struct is aligned well enough */
} workbuf;
char *workb = VARDATA(&workbuf.hdr);
HeapTuple newtup;
Datum values[Natts_pg_largeobject];
bool nulls[Natts_pg_largeobject];
bool replace[Natts_pg_largeobject];
CatalogIndexState indstate;
Assert(PointerIsValid(obj_desc));
/* enforce writability because snapshot is probably wrong otherwise */
if ((obj_desc->flags & IFS_WRLOCK) == 0)
ereport(ERROR,
(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
errmsg("large object %u was not opened for writing",
obj_desc->id)));
/* check existence of the target largeobject */
if (!LargeObjectExists(obj_desc->id))
ereport(ERROR,
(errcode(ERRCODE_UNDEFINED_OBJECT),
errmsg("large object %u was already dropped", obj_desc->id)));
open_lo_relation();
indstate = CatalogOpenIndexes(lo_heap_r);
/*
* Set up to find all pages with desired loid and pageno >= target
*/
ScanKeyInit(&skey[0],
Anum_pg_largeobject_loid,
BTEqualStrategyNumber, F_OIDEQ,
ObjectIdGetDatum(obj_desc->id));
ScanKeyInit(&skey[1],
Anum_pg_largeobject_pageno,
BTGreaterEqualStrategyNumber, F_INT4GE,
Int32GetDatum(pageno));
sd = systable_beginscan_ordered(lo_heap_r, lo_index_r,
obj_desc->snapshot, 2, skey);
/*
* If possible, get the page the truncation point is in. The truncation
* point may be beyond the end of the LO or in a hole.
*/
olddata = NULL;
if ((oldtuple = systable_getnext_ordered(sd, ForwardScanDirection)) != NULL)
{
if (HeapTupleHasNulls(oldtuple)) /* paranoia */
elog(ERROR, "null field found in pg_largeobject");
olddata = (Form_pg_largeobject) GETSTRUCT(oldtuple);
Assert(olddata->pageno >= pageno);
}
/*
* If we found the page of the truncation point we need to truncate the
* data in it. Otherwise if we're in a hole, we need to create a page to
* mark the end of data.
*/
if (olddata != NULL && olddata->pageno == pageno)
{
/* First, load old data into workbuf */
bytea *datafield = &(olddata->data); /* see note at top of
* file */
bool pfreeit = false;
int pagelen;
if (VARATT_IS_EXTENDED(datafield))
{
datafield = (bytea *)
heap_tuple_untoast_attr((struct varlena *) datafield);
pfreeit = true;
}
pagelen = getbytealen(datafield);
Assert(pagelen <= LOBLKSIZE);
memcpy(workb, VARDATA(datafield), pagelen);
if (pfreeit)
pfree(datafield);
/*
* Fill any hole
*/
off = len % LOBLKSIZE;
if (off > pagelen)
MemSet(workb + pagelen, 0, off - pagelen);
/* compute length of new page */
SET_VARSIZE(&workbuf.hdr, off + VARHDRSZ);
/*
* Form and insert updated tuple
*/
memset(values, 0, sizeof(values));
memset(nulls, false, sizeof(nulls));
memset(replace, false, sizeof(replace));
values[Anum_pg_largeobject_data - 1] = PointerGetDatum(&workbuf);
replace[Anum_pg_largeobject_data - 1] = true;
newtup = heap_modify_tuple(oldtuple, RelationGetDescr(lo_heap_r),
values, nulls, replace);
simple_heap_update(lo_heap_r, &newtup->t_self, newtup);
CatalogIndexInsert(indstate, newtup);
heap_freetuple(newtup);
}
else
{
/*
* If the first page we found was after the truncation point, we're in
* a hole that we'll fill, but we need to delete the later page
* because the loop below won't visit it again.
*/
if (olddata != NULL)
{
Assert(olddata->pageno > pageno);
simple_heap_delete(lo_heap_r, &oldtuple->t_self);
}
/*
* Write a brand new page.
*
* Fill the hole up to the truncation point
*/
off = len % LOBLKSIZE;
if (off > 0)
MemSet(workb, 0, off);
/* compute length of new page */
SET_VARSIZE(&workbuf.hdr, off + VARHDRSZ);
/*
* Form and insert new tuple
*/
memset(values, 0, sizeof(values));
memset(nulls, false, sizeof(nulls));
values[Anum_pg_largeobject_loid - 1] = ObjectIdGetDatum(obj_desc->id);
values[Anum_pg_largeobject_pageno - 1] = Int32GetDatum(pageno);
values[Anum_pg_largeobject_data - 1] = PointerGetDatum(&workbuf);
newtup = heap_form_tuple(lo_heap_r->rd_att, values, nulls);
simple_heap_insert(lo_heap_r, newtup);
CatalogIndexInsert(indstate, newtup);
heap_freetuple(newtup);
}
/*
* Delete any pages after the truncation point. If the initial search
* didn't find a page, then of course there's nothing more to do.
*/
if (olddata != NULL)
{
while ((oldtuple = systable_getnext_ordered(sd, ForwardScanDirection)) != NULL)
{
simple_heap_delete(lo_heap_r, &oldtuple->t_self);
}
}
systable_endscan_ordered(sd);
CatalogCloseIndexes(indstate);
/*
* Advance command counter so that tuple updates will be seen by later
* large-object operations in this transaction.
*/
CommandCounterIncrement();
}
int
inv_write(LargeObjectDesc *obj_desc, const char *buf, int nbytes)
{
int nwritten = 0;
int n;
int off;
int len;
int32 pageno = (int32) (obj_desc->offset / LOBLKSIZE);
ScanKeyData skey[2];
SysScanDesc sd;
HeapTuple oldtuple;
Form_pg_largeobject olddata;
bool neednextpage;
bytea *datafield;
bool pfreeit;
struct
{
bytea hdr;
char data[LOBLKSIZE]; /* make struct big enough */
int32 align_it; /* ensure struct is aligned well enough */
} workbuf;
char *workb = VARDATA(&workbuf.hdr);
HeapTuple newtup;
Datum values[Natts_pg_largeobject];
bool nulls[Natts_pg_largeobject];
bool replace[Natts_pg_largeobject];
CatalogIndexState indstate;
Assert(PointerIsValid(obj_desc));
Assert(buf != NULL);
/* enforce writability because snapshot is probably wrong otherwise */
if ((obj_desc->flags & IFS_WRLOCK) == 0)
ereport(ERROR,
(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
errmsg("large object %u was not opened for writing",
obj_desc->id)));
/* check existence of the target largeobject */
if (!LargeObjectExists(obj_desc->id))
ereport(ERROR,
(errcode(ERRCODE_UNDEFINED_OBJECT),
errmsg("large object %u was already dropped", obj_desc->id)));
if (nbytes <= 0)
return 0;
open_lo_relation();
indstate = CatalogOpenIndexes(lo_heap_r);
ScanKeyInit(&skey[0],
Anum_pg_largeobject_loid,
BTEqualStrategyNumber, F_OIDEQ,
ObjectIdGetDatum(obj_desc->id));
ScanKeyInit(&skey[1],
Anum_pg_largeobject_pageno,
BTGreaterEqualStrategyNumber, F_INT4GE,
Int32GetDatum(pageno));
sd = systable_beginscan_ordered(lo_heap_r, lo_index_r,
obj_desc->snapshot, 2, skey);
oldtuple = NULL;
olddata = NULL;
neednextpage = true;
while (nwritten < nbytes)
{
/*
* If possible, get next pre-existing page of the LO. We expect the
* indexscan will deliver these in order --- but there may be holes.
*/
if (neednextpage)
{
if ((oldtuple = systable_getnext_ordered(sd, ForwardScanDirection)) != NULL)
{
if (HeapTupleHasNulls(oldtuple)) /* paranoia */
elog(ERROR, "null field found in pg_largeobject");
olddata = (Form_pg_largeobject) GETSTRUCT(oldtuple);
Assert(olddata->pageno >= pageno);
}
neednextpage = false;
}
/*
* If we have a pre-existing page, see if it is the page we want to
* write, or a later one.
*/
if (olddata != NULL && olddata->pageno == pageno)
{
/*
* Update an existing page with fresh data.
*
* First, load old data into workbuf
*/
datafield = &(olddata->data); /* see note at top of file */
pfreeit = false;
if (VARATT_IS_EXTENDED(datafield))
{
datafield = (bytea *)
heap_tuple_untoast_attr((struct varlena *) datafield);
pfreeit = true;
}
len = getbytealen(datafield);
Assert(len <= LOBLKSIZE);
memcpy(workb, VARDATA(datafield), len);
if (pfreeit)
pfree(datafield);
/*
* Fill any hole
*/
off = (int) (obj_desc->offset % LOBLKSIZE);
if (off > len)
MemSet(workb + len, 0, off - len);
/*
* Insert appropriate portion of new data
*/
n = LOBLKSIZE - off;
n = (n <= (nbytes - nwritten)) ? n : (nbytes - nwritten);
memcpy(workb + off, buf + nwritten, n);
nwritten += n;
obj_desc->offset += n;
off += n;
/* compute valid length of new page */
len = (len >= off) ? len : off;
SET_VARSIZE(&workbuf.hdr, len + VARHDRSZ);
/*
* Form and insert updated tuple
*/
memset(values, 0, sizeof(values));
memset(nulls, false, sizeof(nulls));
memset(replace, false, sizeof(replace));
values[Anum_pg_largeobject_data - 1] = PointerGetDatum(&workbuf);
replace[Anum_pg_largeobject_data - 1] = true;
newtup = heap_modify_tuple(oldtuple, RelationGetDescr(lo_heap_r),
values, nulls, replace);
simple_heap_update(lo_heap_r, &newtup->t_self, newtup);
CatalogIndexInsert(indstate, newtup);
heap_freetuple(newtup);
/*
* We're done with this old page.
*/
oldtuple = NULL;
olddata = NULL;
neednextpage = true;
}
else
{
/*
* Write a brand new page.
*
* First, fill any hole
*/
off = (int) (obj_desc->offset % LOBLKSIZE);
if (off > 0)
MemSet(workb, 0, off);
/*
* Insert appropriate portion of new data
*/
n = LOBLKSIZE - off;
n = (n <= (nbytes - nwritten)) ? n : (nbytes - nwritten);
memcpy(workb + off, buf + nwritten, n);
nwritten += n;
obj_desc->offset += n;
/* compute valid length of new page */
len = off + n;
SET_VARSIZE(&workbuf.hdr, len + VARHDRSZ);
/*
* Form and insert updated tuple
*/
memset(values, 0, sizeof(values));
memset(nulls, false, sizeof(nulls));
values[Anum_pg_largeobject_loid - 1] = ObjectIdGetDatum(obj_desc->id);
values[Anum_pg_largeobject_pageno - 1] = Int32GetDatum(pageno);
values[Anum_pg_largeobject_data - 1] = PointerGetDatum(&workbuf);
newtup = heap_form_tuple(lo_heap_r->rd_att, values, nulls);
simple_heap_insert(lo_heap_r, newtup);
CatalogIndexInsert(indstate, newtup);
heap_freetuple(newtup);
}
pageno++;
}
systable_endscan_ordered(sd);
CatalogCloseIndexes(indstate);
/*
* Advance command counter so that my tuple updates will be seen by later
* large-object operations in this transaction.
*/
CommandCounterIncrement();
return nwritten;
}
/*
* Record multiple dependencies (of the same kind) for a single dependent
* object. This has a little less overhead than recording each separately.
*/
void
recordMultipleDependencies(const ObjectAddress *depender,
const ObjectAddress *referenced,
int nreferenced,
DependencyType behavior)
{
Relation dependDesc;
CatalogIndexState indstate;
HeapTuple tup;
int i;
bool nulls[Natts_pg_depend];
Datum values[Natts_pg_depend];
if (nreferenced <= 0)
return; /* nothing to do */
/*
* During bootstrap, do nothing since pg_depend may not exist yet. initdb
* will fill in appropriate pg_depend entries after bootstrap.
*/
if (IsBootstrapProcessingMode())
return;
dependDesc = heap_open(DependRelationId, RowExclusiveLock);
/* Don't open indexes unless we need to make an update */
indstate = NULL;
memset(nulls, false, sizeof(nulls));
for (i = 0; i < nreferenced; i++, referenced++)
{
/*
* If the referenced object is pinned by the system, there's no real
* need to record dependencies on it. This saves lots of space in
* pg_depend, so it's worth the time taken to check.
*/
if (!isObjectPinned(referenced, dependDesc))
{
/*
* Record the Dependency. Note we don't bother to check for
* duplicate dependencies; there's no harm in them.
*/
values[Anum_pg_depend_classid - 1] = ObjectIdGetDatum(depender->classId);
values[Anum_pg_depend_objid - 1] = ObjectIdGetDatum(depender->objectId);
values[Anum_pg_depend_objsubid - 1] = Int32GetDatum(depender->objectSubId);
values[Anum_pg_depend_refclassid - 1] = ObjectIdGetDatum(referenced->classId);
values[Anum_pg_depend_refobjid - 1] = ObjectIdGetDatum(referenced->objectId);
values[Anum_pg_depend_refobjsubid - 1] = Int32GetDatum(referenced->objectSubId);
values[Anum_pg_depend_deptype - 1] = CharGetDatum((char) behavior);
tup = heap_form_tuple(dependDesc->rd_att, values, nulls);
simple_heap_insert(dependDesc, tup);
/* keep indexes current */
if (indstate == NULL)
indstate = CatalogOpenIndexes(dependDesc);
CatalogIndexInsert(indstate, tup);
heap_freetuple(tup);
}
}
if (indstate != NULL)
CatalogCloseIndexes(indstate);
heap_close(dependDesc, RowExclusiveLock);
}
int
inv_write(LargeObjectDesc *obj_desc, char *buf, int nbytes)
{
int nwritten = 0;
int n;
int off;
int len;
int32 pageno = (int32) (obj_desc->offset / LOBLKSIZE);
ScanKeyData skey[2];
IndexScanDesc sd;
HeapTuple oldtuple;
Form_pg_largeobject olddata;
bool neednextpage;
bytea *datafield;
bool pfreeit;
struct
{
bytea hdr;
char data[LOBLKSIZE];
} workbuf;
char *workb = VARATT_DATA(&workbuf.hdr);
HeapTuple newtup;
Datum values[Natts_pg_largeobject];
char nulls[Natts_pg_largeobject];
char replace[Natts_pg_largeobject];
CatalogIndexState indstate;
Assert(PointerIsValid(obj_desc));
Assert(buf != NULL);
if (nbytes <= 0)
return 0;
open_lo_relation();
indstate = CatalogOpenIndexes(lo_heap_r);
ScanKeyInit(&skey[0],
Anum_pg_largeobject_loid,
BTEqualStrategyNumber, F_OIDEQ,
ObjectIdGetDatum(obj_desc->id));
ScanKeyInit(&skey[1],
Anum_pg_largeobject_pageno,
BTGreaterEqualStrategyNumber, F_INT4GE,
Int32GetDatum(pageno));
sd = index_beginscan(lo_heap_r, lo_index_r,
SnapshotNow, 2, skey);
oldtuple = NULL;
olddata = NULL;
neednextpage = true;
while (nwritten < nbytes)
{
/*
* If possible, get next pre-existing page of the LO. We assume
* the indexscan will deliver these in order --- but there may be
* holes.
*/
if (neednextpage)
{
if ((oldtuple = index_getnext(sd, ForwardScanDirection)) != NULL)
{
olddata = (Form_pg_largeobject) GETSTRUCT(oldtuple);
Assert(olddata->pageno >= pageno);
}
neednextpage = false;
}
/*
* If we have a pre-existing page, see if it is the page we want
* to write, or a later one.
*/
if (olddata != NULL && olddata->pageno == pageno)
{
/*
* Update an existing page with fresh data.
*
* First, load old data into workbuf
*/
datafield = &(olddata->data);
pfreeit = false;
if (VARATT_IS_EXTENDED(datafield))
{
datafield = (bytea *)
heap_tuple_untoast_attr((varattrib *) datafield);
pfreeit = true;
}
len = getbytealen(datafield);
Assert(len <= LOBLKSIZE);
memcpy(workb, VARDATA(datafield), len);
if (pfreeit)
pfree(datafield);
/*
* Fill any hole
*/
off = (int) (obj_desc->offset % LOBLKSIZE);
if (off > len)
MemSet(workb + len, 0, off - len);
/*
* Insert appropriate portion of new data
*/
n = LOBLKSIZE - off;
n = (n <= (nbytes - nwritten)) ? n : (nbytes - nwritten);
memcpy(workb + off, buf + nwritten, n);
nwritten += n;
obj_desc->offset += n;
off += n;
/* compute valid length of new page */
len = (len >= off) ? len : off;
VARATT_SIZEP(&workbuf.hdr) = len + VARHDRSZ;
/*
* Form and insert updated tuple
*/
memset(values, 0, sizeof(values));
memset(nulls, ' ', sizeof(nulls));
memset(replace, ' ', sizeof(replace));
values[Anum_pg_largeobject_data - 1] = PointerGetDatum(&workbuf);
replace[Anum_pg_largeobject_data - 1] = 'r';
newtup = heap_modifytuple(oldtuple, lo_heap_r,
values, nulls, replace);
simple_heap_update(lo_heap_r, &newtup->t_self, newtup);
CatalogIndexInsert(indstate, newtup);
heap_freetuple(newtup);
/*
* We're done with this old page.
*/
oldtuple = NULL;
olddata = NULL;
neednextpage = true;
}
else
{
/*
* Write a brand new page.
*
* First, fill any hole
*/
off = (int) (obj_desc->offset % LOBLKSIZE);
if (off > 0)
MemSet(workb, 0, off);
/*
* Insert appropriate portion of new data
*/
n = LOBLKSIZE - off;
n = (n <= (nbytes - nwritten)) ? n : (nbytes - nwritten);
memcpy(workb + off, buf + nwritten, n);
nwritten += n;
obj_desc->offset += n;
/* compute valid length of new page */
len = off + n;
VARATT_SIZEP(&workbuf.hdr) = len + VARHDRSZ;
/*
* Form and insert updated tuple
*/
memset(values, 0, sizeof(values));
memset(nulls, ' ', sizeof(nulls));
values[Anum_pg_largeobject_loid - 1] = ObjectIdGetDatum(obj_desc->id);
values[Anum_pg_largeobject_pageno - 1] = Int32GetDatum(pageno);
values[Anum_pg_largeobject_data - 1] = PointerGetDatum(&workbuf);
newtup = heap_formtuple(lo_heap_r->rd_att, values, nulls);
simple_heap_insert(lo_heap_r, newtup);
CatalogIndexInsert(indstate, newtup);
heap_freetuple(newtup);
}
pageno++;
}
index_endscan(sd);
CatalogCloseIndexes(indstate);
/*
* Advance command counter so that my tuple updates will be seen by
* later large-object operations in this transaction.
*/
CommandCounterIncrement();
return nwritten;
}