StringInfo rest_call_with_lock(char *method, char *url, char *params, StringInfo postData, int64 mutex, bool shared, bool allowCancel) {
CURL *curl;
struct curl_slist *headers = NULL;
char *errorbuff;
StringInfo response = makeStringInfo();
CURLcode ret;
int64 response_code;
errorbuff = (char *) palloc0(CURL_ERROR_SIZE);
curl = curl_easy_init();
if (curl) {
headers = curl_slist_append(headers, "Transfer-Encoding:");
headers = curl_slist_append(headers, "Expect:");
curl_easy_setopt(curl, CURLOPT_HTTPHEADER, headers);
curl_easy_setopt(curl, CURLOPT_FORBID_REUSE, 0L); /* allow connections to be reused */
if (allowCancel)
{
curl_easy_setopt(curl, CURLOPT_NOPROGRESS, 0); /* we want progress ... */
curl_easy_setopt(curl, CURLOPT_PROGRESSFUNCTION, curl_progress_func); /* to go here so we can detect a ^C within postgres */
}
curl_easy_setopt(curl, CURLOPT_USERAGENT, "zombodb for PostgreSQL");
curl_easy_setopt(curl, CURLOPT_MAXREDIRS, 0);
curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, curl_write_func);
curl_easy_setopt(curl, CURLOPT_FAILONERROR, 0);
curl_easy_setopt(curl, CURLOPT_ERRORBUFFER, errorbuff);
curl_easy_setopt(curl, CURLOPT_NOSIGNAL, 1);
curl_easy_setopt(curl, CURLOPT_TIMEOUT, 60 * 60L); /* timeout of 60 minutes */
curl_easy_setopt(curl, CURLOPT_URL, url);
curl_easy_setopt(curl, CURLOPT_CUSTOMREQUEST, method);
curl_easy_setopt(curl, CURLOPT_WRITEDATA, response);
curl_easy_setopt(curl, CURLOPT_POSTFIELDSIZE, postData ? postData->len : 0);
curl_easy_setopt(curl, CURLOPT_POSTFIELDS, postData ? postData->data : NULL);
curl_easy_setopt(curl, CURLOPT_POST, (strcmp(method, "POST") == 0) || (strcmp(method, "GET") != 0 && postData && postData->data) ? 1 : 0);
} else {
elog(IsTransactionState() ? ERROR : WARNING, "Unable to initialize libcurl");
}
// if (mutex != 0)
// {
// if (shared) DirectFunctionCall1(pg_advisory_lock_shared_int8, Int64GetDatum(mutex));
// else DirectFunctionCall1(pg_advisory_lock_int8, Int64GetDatum(mutex));
// }
ret = curl_easy_perform(curl);
// if (mutex != 0)
// {
// if (shared) DirectFunctionCall1(pg_advisory_unlock_shared_int8, Int64GetDatum(mutex));
// else DirectFunctionCall1(pg_advisory_unlock_int8, Int64GetDatum(mutex));
// }
if (allowCancel && IsTransactionState() && InterruptPending) {
/* we might have detected one in the progress function, so check for sure */
CHECK_FOR_INTERRUPTS();
}
if (ret != 0) {
/* curl messed up */
elog(IsTransactionState() ? ERROR : WARNING, "libcurl error-code: %s(%d); message: %s; req=-X%s %s ", curl_easy_strerror(ret), ret, errorbuff, method, url);
}
curl_easy_getinfo(curl, CURLINFO_RESPONSE_CODE, &response_code);
if (response_code < 200 || (response_code >=300 && response_code != 404)) {
text *errorText = DatumGetTextP(DirectFunctionCall2(json_object_field_text, CStringGetTextDatum(response->data), CStringGetTextDatum("error")));
elog(IsTransactionState() ? ERROR : WARNING, "rc=%ld; %s", response_code, errorText != NULL ? TextDatumGetCString(errorText) : response->data);
}
if (headers)
curl_slist_free_all(headers);
curl_easy_cleanup(curl);
pfree(errorbuff);
return response;
}
/*
* Initialize the TABLESAMPLE Descriptor and the TABLESAMPLE Method.
*/
TableSampleDesc *
tablesample_init(SampleScanState *scanstate, TableSampleClause *tablesample)
{
FunctionCallInfoData fcinfo;
int i;
List *args = tablesample->args;
ListCell *arg;
ExprContext *econtext = scanstate->ss.ps.ps_ExprContext;
TableSampleDesc *tsdesc = (TableSampleDesc *) palloc0(sizeof(TableSampleDesc));
/* Load functions */
fmgr_info(tablesample->tsminit, &(tsdesc->tsminit));
fmgr_info(tablesample->tsmnextblock, &(tsdesc->tsmnextblock));
fmgr_info(tablesample->tsmnexttuple, &(tsdesc->tsmnexttuple));
if (OidIsValid(tablesample->tsmexaminetuple))
fmgr_info(tablesample->tsmexaminetuple, &(tsdesc->tsmexaminetuple));
else
tsdesc->tsmexaminetuple.fn_oid = InvalidOid;
fmgr_info(tablesample->tsmreset, &(tsdesc->tsmreset));
fmgr_info(tablesample->tsmend, &(tsdesc->tsmend));
InitFunctionCallInfoData(fcinfo, &tsdesc->tsminit,
list_length(args) + 2,
InvalidOid, NULL, NULL);
tsdesc->tupDesc = scanstate->ss.ss_ScanTupleSlot->tts_tupleDescriptor;
tsdesc->heapScan = scanstate->ss.ss_currentScanDesc;
/* First argument for init function is always TableSampleDesc */
fcinfo.arg[0] = PointerGetDatum(tsdesc);
fcinfo.argnull[0] = false;
/*
* Second arg for init function is always REPEATABLE
* When tablesample->repeatable is NULL then REPEATABLE clause was not
* specified.
* When specified, the expression cannot evaluate to NULL.
*/
if (tablesample->repeatable)
{
ExprState *argstate = ExecInitExpr((Expr *) tablesample->repeatable,
(PlanState *) scanstate);
fcinfo.arg[1] = ExecEvalExpr(argstate, econtext,
&fcinfo.argnull[1], NULL);
if (fcinfo.argnull[1])
ereport(ERROR,
(errcode(ERRCODE_NULL_VALUE_NOT_ALLOWED),
errmsg("REPEATABLE clause must be NOT NULL numeric value")));
}
else
{
fcinfo.arg[1] = UInt32GetDatum(random());
fcinfo.argnull[1] = false;
}
/* Rest of the arguments come from user. */
i = 2;
foreach(arg, args)
{
Expr *argexpr = (Expr *) lfirst(arg);
ExprState *argstate = ExecInitExpr(argexpr, (PlanState *) scanstate);
if (argstate == NULL)
{
fcinfo.argnull[i] = true;
fcinfo.arg[i] = (Datum) 0;;
}
fcinfo.arg[i] = ExecEvalExpr(argstate, econtext,
&fcinfo.argnull[i], NULL);
i++;
}
/* ----------------------------------------------------------------
* ExecInitGather
* ----------------------------------------------------------------
*/
GatherMergeState *
ExecInitGatherMerge(GatherMerge *node, EState *estate, int eflags)
{
GatherMergeState *gm_state;
Plan *outerNode;
TupleDesc tupDesc;
/* Gather merge node doesn't have innerPlan node. */
Assert(innerPlan(node) == NULL);
/*
* create state structure
*/
gm_state = makeNode(GatherMergeState);
gm_state->ps.plan = (Plan *) node;
gm_state->ps.state = estate;
gm_state->ps.ExecProcNode = ExecGatherMerge;
gm_state->initialized = false;
gm_state->gm_initialized = false;
gm_state->tuples_needed = -1;
/*
* Miscellaneous initialization
*
* create expression context for node
*/
ExecAssignExprContext(estate, &gm_state->ps);
/*
* GatherMerge doesn't support checking a qual (it's always more efficient
* to do it in the child node).
*/
Assert(!node->plan.qual);
/*
* now initialize outer plan
*/
outerNode = outerPlan(node);
outerPlanState(gm_state) = ExecInitNode(outerNode, estate, eflags);
/*
* Leader may access ExecProcNode result directly (if
* need_to_scan_locally), or from workers via tuple queue. So we can't
* trivially rely on the slot type being fixed for expressions evaluated
* within this node.
*/
gm_state->ps.outeropsset = true;
gm_state->ps.outeropsfixed = false;
/*
* Store the tuple descriptor into gather merge state, so we can use it
* while initializing the gather merge slots.
*/
tupDesc = ExecGetResultType(outerPlanState(gm_state));
gm_state->tupDesc = tupDesc;
/*
* Initialize result type and projection.
*/
ExecInitResultTypeTL(&gm_state->ps);
ExecConditionalAssignProjectionInfo(&gm_state->ps, tupDesc, OUTER_VAR);
/*
* Without projections result slot type is not trivially known, see
* comment above.
*/
if (gm_state->ps.ps_ProjInfo == NULL)
{
gm_state->ps.resultopsset = true;
gm_state->ps.resultopsfixed = false;
}
/*
* initialize sort-key information
*/
if (node->numCols)
{
int i;
gm_state->gm_nkeys = node->numCols;
gm_state->gm_sortkeys =
palloc0(sizeof(SortSupportData) * node->numCols);
for (i = 0; i < node->numCols; i++)
{
SortSupport sortKey = gm_state->gm_sortkeys + i;
sortKey->ssup_cxt = CurrentMemoryContext;
sortKey->ssup_collation = node->collations[i];
sortKey->ssup_nulls_first = node->nullsFirst[i];
sortKey->ssup_attno = node->sortColIdx[i];
/*
* We don't perform abbreviated key conversion here, for the same
* reasons that it isn't used in MergeAppend
*/
sortKey->abbreviate = false;
PrepareSortSupportFromOrderingOp(node->sortOperators[i], sortKey);
}
}
/* Now allocate the workspace for gather merge */
gather_merge_setup(gm_state);
return gm_state;
}
Datum
lquery_in(PG_FUNCTION_ARGS)
{
char *buf = (char *) PG_GETARG_POINTER(0);
char *ptr;
int num = 0,
totallen = 0,
numOR = 0;
int state = LQPRS_WAITLEVEL;
lquery *result;
nodeitem *lptr = NULL;
lquery_level *cur,
*curqlevel,
*tmpql;
lquery_variant *lrptr = NULL;
bool hasnot = false;
bool wasbad = false;
int charlen;
int pos = 0;
ptr = buf;
while (*ptr)
{
charlen = pg_mblen(ptr);
if (charlen == 1)
{
if (t_iseq(ptr, '.'))
num++;
else if (t_iseq(ptr, '|'))
numOR++;
}
ptr += charlen;
}
num++;
if (num > MaxAllocSize / ITEMSIZE)
ereport(ERROR,
(errcode(ERRCODE_PROGRAM_LIMIT_EXCEEDED),
errmsg("number of levels (%d) exceeds the maximum allowed (%d)",
num, (int) (MaxAllocSize / ITEMSIZE))));
curqlevel = tmpql = (lquery_level *) palloc0(ITEMSIZE * num);
ptr = buf;
while (*ptr)
{
charlen = pg_mblen(ptr);
if (state == LQPRS_WAITLEVEL)
{
if (ISALNUM(ptr))
{
GETVAR(curqlevel) = lptr = (nodeitem *) palloc0(sizeof(nodeitem) * (numOR + 1));
lptr->start = ptr;
state = LQPRS_WAITDELIM;
curqlevel->numvar = 1;
}
else if (charlen == 1 && t_iseq(ptr, '!'))
{
GETVAR(curqlevel) = lptr = (nodeitem *) palloc0(sizeof(nodeitem) * (numOR + 1));
lptr->start = ptr + 1;
state = LQPRS_WAITDELIM;
curqlevel->numvar = 1;
curqlevel->flag |= LQL_NOT;
hasnot = true;
}
else if (charlen == 1 && t_iseq(ptr, '*'))
state = LQPRS_WAITOPEN;
else
UNCHAR;
}
else if (state == LQPRS_WAITVAR)
{
if (ISALNUM(ptr))
{
lptr++;
lptr->start = ptr;
state = LQPRS_WAITDELIM;
curqlevel->numvar++;
}
else
UNCHAR;
}
else if (state == LQPRS_WAITDELIM)
{
if (charlen == 1 && t_iseq(ptr, '@'))
{
if (lptr->start == ptr)
UNCHAR;
lptr->flag |= LVAR_INCASE;
curqlevel->flag |= LVAR_INCASE;
}
else if (charlen == 1 && t_iseq(ptr, '*'))
{
if (lptr->start == ptr)
UNCHAR;
lptr->flag |= LVAR_ANYEND;
curqlevel->flag |= LVAR_ANYEND;
}
else if (charlen == 1 && t_iseq(ptr, '%'))
{
if (lptr->start == ptr)
UNCHAR;
lptr->flag |= LVAR_SUBLEXEME;
curqlevel->flag |= LVAR_SUBLEXEME;
}
else if (charlen == 1 && t_iseq(ptr, '|'))
{
lptr->len = ptr - lptr->start -
((lptr->flag & LVAR_SUBLEXEME) ? 1 : 0) -
((lptr->flag & LVAR_INCASE) ? 1 : 0) -
((lptr->flag & LVAR_ANYEND) ? 1 : 0);
if (lptr->wlen > 255)
ereport(ERROR,
(errcode(ERRCODE_NAME_TOO_LONG),
errmsg("name of level is too long"),
errdetail("Name length is %d, must "
"be < 256, in position %d.",
lptr->wlen, pos)));
state = LQPRS_WAITVAR;
}
else if (charlen == 1 && t_iseq(ptr, '.'))
{
lptr->len = ptr - lptr->start -
((lptr->flag & LVAR_SUBLEXEME) ? 1 : 0) -
((lptr->flag & LVAR_INCASE) ? 1 : 0) -
((lptr->flag & LVAR_ANYEND) ? 1 : 0);
if (lptr->wlen > 255)
ereport(ERROR,
(errcode(ERRCODE_NAME_TOO_LONG),
errmsg("name of level is too long"),
errdetail("Name length is %d, must "
"be < 256, in position %d.",
lptr->wlen, pos)));
state = LQPRS_WAITLEVEL;
curqlevel = NEXTLEV(curqlevel);
}
else if (ISALNUM(ptr))
{
if (lptr->flag)
UNCHAR;
}
else
UNCHAR;
}
else if (state == LQPRS_WAITOPEN)
{
if (charlen == 1 && t_iseq(ptr, '{'))
state = LQPRS_WAITFNUM;
else if (charlen == 1 && t_iseq(ptr, '.'))
{
curqlevel->low = 0;
curqlevel->high = 0xffff;
curqlevel = NEXTLEV(curqlevel);
state = LQPRS_WAITLEVEL;
}
else
UNCHAR;
}
else if (state == LQPRS_WAITFNUM)
{
if (charlen == 1 && t_iseq(ptr, ','))
state = LQPRS_WAITSNUM;
else if (t_isdigit(ptr))
{
curqlevel->low = atoi(ptr);
state = LQPRS_WAITND;
}
else
UNCHAR;
}
else if (state == LQPRS_WAITSNUM)
{
if (t_isdigit(ptr))
{
curqlevel->high = atoi(ptr);
state = LQPRS_WAITCLOSE;
}
else if (charlen == 1 && t_iseq(ptr, '}'))
{
curqlevel->high = 0xffff;
state = LQPRS_WAITEND;
}
else
UNCHAR;
}
else if (state == LQPRS_WAITCLOSE)
{
if (charlen == 1 && t_iseq(ptr, '}'))
state = LQPRS_WAITEND;
else if (!t_isdigit(ptr))
UNCHAR;
}
else if (state == LQPRS_WAITND)
{
if (charlen == 1 && t_iseq(ptr, '}'))
{
curqlevel->high = curqlevel->low;
state = LQPRS_WAITEND;
}
else if (charlen == 1 && t_iseq(ptr, ','))
state = LQPRS_WAITSNUM;
else if (!t_isdigit(ptr))
UNCHAR;
}
else if (state == LQPRS_WAITEND)
{
if (charlen == 1 && t_iseq(ptr, '.'))
{
state = LQPRS_WAITLEVEL;
curqlevel = NEXTLEV(curqlevel);
}
else
UNCHAR;
}
else
/* internal error */
elog(ERROR, "internal error in parser");
ptr += charlen;
if (state == LQPRS_WAITDELIM)
lptr->wlen++;
pos++;
}
if (state == LQPRS_WAITDELIM)
{
if (lptr->start == ptr)
ereport(ERROR,
(errcode(ERRCODE_SYNTAX_ERROR),
errmsg("syntax error"),
errdetail("Unexpected end of line.")));
lptr->len = ptr - lptr->start -
((lptr->flag & LVAR_SUBLEXEME) ? 1 : 0) -
((lptr->flag & LVAR_INCASE) ? 1 : 0) -
((lptr->flag & LVAR_ANYEND) ? 1 : 0);
if (lptr->len == 0)
ereport(ERROR,
(errcode(ERRCODE_SYNTAX_ERROR),
errmsg("syntax error"),
errdetail("Unexpected end of line.")));
if (lptr->wlen > 255)
ereport(ERROR,
(errcode(ERRCODE_NAME_TOO_LONG),
errmsg("name of level is too long"),
errdetail("Name length is %d, must "
"be < 256, in position %d.",
lptr->wlen, pos)));
}
else if (state == LQPRS_WAITOPEN)
curqlevel->high = 0xffff;
else if (state != LQPRS_WAITEND)
ereport(ERROR,
(errcode(ERRCODE_SYNTAX_ERROR),
errmsg("syntax error"),
errdetail("Unexpected end of line.")));
curqlevel = tmpql;
totallen = LQUERY_HDRSIZE;
while ((char *) curqlevel - (char *) tmpql < num * ITEMSIZE)
{
totallen += LQL_HDRSIZE;
if (curqlevel->numvar)
{
lptr = GETVAR(curqlevel);
while (lptr - GETVAR(curqlevel) < curqlevel->numvar)
{
totallen += MAXALIGN(LVAR_HDRSIZE + lptr->len);
lptr++;
}
}
else if (curqlevel->low > curqlevel->high)
ereport(ERROR,
(errcode(ERRCODE_SYNTAX_ERROR),
errmsg("syntax error"),
errdetail("Low limit(%d) is greater than upper(%d).",
curqlevel->low, curqlevel->high)));
curqlevel = NEXTLEV(curqlevel);
}
result = (lquery *) palloc0(totallen);
SET_VARSIZE(result, totallen);
result->numlevel = num;
result->firstgood = 0;
result->flag = 0;
if (hasnot)
result->flag |= LQUERY_HASNOT;
cur = LQUERY_FIRST(result);
curqlevel = tmpql;
while ((char *) curqlevel - (char *) tmpql < num * ITEMSIZE)
{
memcpy(cur, curqlevel, LQL_HDRSIZE);
cur->totallen = LQL_HDRSIZE;
if (curqlevel->numvar)
{
lrptr = LQL_FIRST(cur);
lptr = GETVAR(curqlevel);
while (lptr - GETVAR(curqlevel) < curqlevel->numvar)
{
cur->totallen += MAXALIGN(LVAR_HDRSIZE + lptr->len);
lrptr->len = lptr->len;
lrptr->flag = lptr->flag;
lrptr->val = ltree_crc32_sz(lptr->start, lptr->len);
memcpy(lrptr->name, lptr->start, lptr->len);
lptr++;
lrptr = LVAR_NEXT(lrptr);
}
pfree(GETVAR(curqlevel));
if (cur->numvar > 1 || cur->flag != 0)
wasbad = true;
else if (wasbad == false)
(result->firstgood)++;
}
else
wasbad = true;
curqlevel = NEXTLEV(curqlevel);
cur = LQL_NEXT(cur);
}
pfree(tmpql);
PG_RETURN_POINTER(result);
}
/*
* get_relation_info -
* Retrieves catalog information for a given relation.
*
* Given the Oid of the relation, return the following info into fields
* of the RelOptInfo struct:
*
* min_attr lowest valid AttrNumber
* max_attr highest valid AttrNumber
* indexlist list of IndexOptInfos for relation's indexes
* fdwroutine if it's a foreign table, the FDW function pointers
* pages number of pages
* tuples number of tuples
*
* Also, initialize the attr_needed[] and attr_widths[] arrays. In most
* cases these are left as zeroes, but sometimes we need to compute attr
* widths here, and we may as well cache the results for costsize.c.
*
* If inhparent is true, all we need to do is set up the attr arrays:
* the RelOptInfo actually represents the appendrel formed by an inheritance
* tree, and so the parent rel's physical size and index information isn't
* important for it.
*/
void
get_relation_info(PlannerInfo *root, Oid relationObjectId, bool inhparent,
RelOptInfo *rel)
{
Index varno = rel->relid;
Relation relation;
bool hasindex;
List *indexinfos = NIL;
/*
* We need not lock the relation since it was already locked, either by
* the rewriter or when expand_inherited_rtentry() added it to the query's
* rangetable.
*/
relation = heap_open(relationObjectId, NoLock);
/* Temporary and unlogged relations are inaccessible during recovery. */
if (!RelationNeedsWAL(relation) && RecoveryInProgress())
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot access temporary or unlogged relations during recovery")));
rel->min_attr = FirstLowInvalidHeapAttributeNumber + 1;
rel->max_attr = RelationGetNumberOfAttributes(relation);
rel->reltablespace = RelationGetForm(relation)->reltablespace;
Assert(rel->max_attr >= rel->min_attr);
rel->attr_needed = (Relids *)
palloc0((rel->max_attr - rel->min_attr + 1) * sizeof(Relids));
rel->attr_widths = (int32 *)
palloc0((rel->max_attr - rel->min_attr + 1) * sizeof(int32));
/*
* Estimate relation size --- unless it's an inheritance parent, in which
* case the size will be computed later in set_append_rel_pathlist, and we
* must leave it zero for now to avoid bollixing the total_table_pages
* calculation.
*/
if (!inhparent)
estimate_rel_size(relation, rel->attr_widths - rel->min_attr,
&rel->pages, &rel->tuples, &rel->allvisfrac);
/*
* Make list of indexes. Ignore indexes on system catalogs if told to.
* Don't bother with indexes for an inheritance parent, either.
*/
if (inhparent ||
(IgnoreSystemIndexes && IsSystemRelation(relation)))
hasindex = false;
else
hasindex = relation->rd_rel->relhasindex;
if (hasindex)
{
List *indexoidlist;
ListCell *l;
LOCKMODE lmode;
indexoidlist = RelationGetIndexList(relation);
/*
* For each index, we get the same type of lock that the executor will
* need, and do not release it. This saves a couple of trips to the
* shared lock manager while not creating any real loss of
* concurrency, because no schema changes could be happening on the
* index while we hold lock on the parent rel, and neither lock type
* blocks any other kind of index operation.
*/
if (rel->relid == root->parse->resultRelation)
lmode = RowExclusiveLock;
else
lmode = AccessShareLock;
foreach(l, indexoidlist)
{
Oid indexoid = lfirst_oid(l);
Relation indexRelation;
Form_pg_index index;
IndexOptInfo *info;
int ncolumns;
int i;
/*
* Extract info from the relation descriptor for the index.
*/
indexRelation = index_open(indexoid, lmode);
index = indexRelation->rd_index;
/*
* Ignore invalid indexes, since they can't safely be used for
* queries. Note that this is OK because the data structure we
* are constructing is only used by the planner --- the executor
* still needs to insert into "invalid" indexes, if they're marked
* IndexIsReady.
*/
if (!IndexIsValid(index))
{
index_close(indexRelation, NoLock);
continue;
}
/*
* If the index is valid, but cannot yet be used, ignore it; but
* mark the plan we are generating as transient. See
* src/backend/access/heap/README.HOT for discussion.
*/
if (index->indcheckxmin &&
!TransactionIdPrecedes(HeapTupleHeaderGetXmin(indexRelation->rd_indextuple->t_data),
TransactionXmin))
{
root->glob->transientPlan = true;
index_close(indexRelation, NoLock);
continue;
}
info = makeNode(IndexOptInfo);
info->indexoid = index->indexrelid;
info->reltablespace =
RelationGetForm(indexRelation)->reltablespace;
info->rel = rel;
info->ncolumns = ncolumns = index->indnatts;
info->indexkeys = (int *) palloc(sizeof(int) * ncolumns);
info->indexcollations = (Oid *) palloc(sizeof(Oid) * ncolumns);
info->opfamily = (Oid *) palloc(sizeof(Oid) * ncolumns);
info->opcintype = (Oid *) palloc(sizeof(Oid) * ncolumns);
for (i = 0; i < ncolumns; i++)
{
info->indexkeys[i] = index->indkey.values[i];
info->indexcollations[i] = indexRelation->rd_indcollation[i];
info->opfamily[i] = indexRelation->rd_opfamily[i];
info->opcintype[i] = indexRelation->rd_opcintype[i];
}
info->relam = indexRelation->rd_rel->relam;
info->amcostestimate = indexRelation->rd_am->amcostestimate;
info->canreturn = index_can_return(indexRelation);
info->amcanorderbyop = indexRelation->rd_am->amcanorderbyop;
info->amoptionalkey = indexRelation->rd_am->amoptionalkey;
info->amsearcharray = indexRelation->rd_am->amsearcharray;
info->amsearchnulls = indexRelation->rd_am->amsearchnulls;
info->amhasgettuple = OidIsValid(indexRelation->rd_am->amgettuple);
info->amhasgetbitmap = OidIsValid(indexRelation->rd_am->amgetbitmap);
/*
* Fetch the ordering information for the index, if any.
*/
if (info->relam == BTREE_AM_OID)
{
/*
* If it's a btree index, we can use its opfamily OIDs
* directly as the sort ordering opfamily OIDs.
*/
Assert(indexRelation->rd_am->amcanorder);
info->sortopfamily = info->opfamily;
info->reverse_sort = (bool *) palloc(sizeof(bool) * ncolumns);
info->nulls_first = (bool *) palloc(sizeof(bool) * ncolumns);
for (i = 0; i < ncolumns; i++)
{
int16 opt = indexRelation->rd_indoption[i];
info->reverse_sort[i] = (opt & INDOPTION_DESC) != 0;
info->nulls_first[i] = (opt & INDOPTION_NULLS_FIRST) != 0;
}
}
else if (indexRelation->rd_am->amcanorder)
{
/*
* Otherwise, identify the corresponding btree opfamilies by
* trying to map this index's "<" operators into btree. Since
* "<" uniquely defines the behavior of a sort order, this is
* a sufficient test.
*
* XXX This method is rather slow and also requires the
* undesirable assumption that the other index AM numbers its
* strategies the same as btree. It'd be better to have a way
* to explicitly declare the corresponding btree opfamily for
* each opfamily of the other index type. But given the lack
* of current or foreseeable amcanorder index types, it's not
* worth expending more effort on now.
*/
info->sortopfamily = (Oid *) palloc(sizeof(Oid) * ncolumns);
info->reverse_sort = (bool *) palloc(sizeof(bool) * ncolumns);
info->nulls_first = (bool *) palloc(sizeof(bool) * ncolumns);
for (i = 0; i < ncolumns; i++)
{
int16 opt = indexRelation->rd_indoption[i];
Oid ltopr;
Oid btopfamily;
Oid btopcintype;
int16 btstrategy;
info->reverse_sort[i] = (opt & INDOPTION_DESC) != 0;
info->nulls_first[i] = (opt & INDOPTION_NULLS_FIRST) != 0;
ltopr = get_opfamily_member(info->opfamily[i],
info->opcintype[i],
info->opcintype[i],
BTLessStrategyNumber);
if (OidIsValid(ltopr) &&
get_ordering_op_properties(ltopr,
&btopfamily,
&btopcintype,
&btstrategy) &&
btopcintype == info->opcintype[i] &&
btstrategy == BTLessStrategyNumber)
{
/* Successful mapping */
info->sortopfamily[i] = btopfamily;
}
else
{
/* Fail ... quietly treat index as unordered */
info->sortopfamily = NULL;
info->reverse_sort = NULL;
info->nulls_first = NULL;
break;
}
}
}
else
{
info->sortopfamily = NULL;
info->reverse_sort = NULL;
info->nulls_first = NULL;
}
/*
* Fetch the index expressions and predicate, if any. We must
* modify the copies we obtain from the relcache to have the
* correct varno for the parent relation, so that they match up
* correctly against qual clauses.
*/
info->indexprs = RelationGetIndexExpressions(indexRelation);
info->indpred = RelationGetIndexPredicate(indexRelation);
if (info->indexprs && varno != 1)
ChangeVarNodes((Node *) info->indexprs, 1, varno, 0);
if (info->indpred && varno != 1)
ChangeVarNodes((Node *) info->indpred, 1, varno, 0);
/* Build targetlist using the completed indexprs data */
info->indextlist = build_index_tlist(root, info, relation);
info->predOK = false; /* set later in indxpath.c */
info->unique = index->indisunique;
info->immediate = index->indimmediate;
info->hypothetical = false;
/*
* Estimate the index size. If it's not a partial index, we lock
* the number-of-tuples estimate to equal the parent table; if it
* is partial then we have to use the same methods as we would for
* a table, except we can be sure that the index is not larger
* than the table.
*/
if (info->indpred == NIL)
{
info->pages = RelationGetNumberOfBlocks(indexRelation);
info->tuples = rel->tuples;
}
else
{
double allvisfrac; /* dummy */
estimate_rel_size(indexRelation, NULL,
&info->pages, &info->tuples, &allvisfrac);
if (info->tuples > rel->tuples)
info->tuples = rel->tuples;
}
if (info->relam == BTREE_AM_OID)
{
/* For btrees, get tree height while we have the index open */
info->tree_height = _bt_getrootheight(indexRelation);
}
else
{
/* For other index types, just set it to "unknown" for now */
info->tree_height = -1;
}
index_close(indexRelation, NoLock);
indexinfos = lcons(info, indexinfos);
}
list_free(indexoidlist);
}
/*
* Construct an inner tuple containing the given prefix and node array
*/
SpGistInnerTuple
spgFormInnerTuple(SpGistState *state, bool hasPrefix, Datum prefix,
int nNodes, SpGistNodeTuple *nodes)
{
SpGistInnerTuple tup;
unsigned int size;
unsigned int prefixSize;
int i;
char *ptr;
/* Compute size needed */
if (hasPrefix)
prefixSize = SpGistGetTypeSize(&state->attPrefixType, prefix);
else
prefixSize = 0;
size = SGITHDRSZ + prefixSize;
/* Note: we rely on node tuple sizes to be maxaligned already */
for (i = 0; i < nNodes; i++)
size += IndexTupleSize(nodes[i]);
/*
* Ensure that we can replace the tuple with a dead tuple later. This
* test is unnecessary given current tuple layouts, but let's be safe.
*/
if (size < SGDTSIZE)
size = SGDTSIZE;
/*
* Inner tuple should be small enough to fit on a page
*/
if (size > SPGIST_PAGE_CAPACITY - sizeof(ItemIdData))
ereport(ERROR,
(errcode(ERRCODE_PROGRAM_LIMIT_EXCEEDED),
errmsg("SP-GiST inner tuple size %zu exceeds maximum %zu",
(Size) size,
SPGIST_PAGE_CAPACITY - sizeof(ItemIdData)),
errhint("Values larger than a buffer page cannot be indexed.")));
/*
* Check for overflow of header fields --- probably can't fail if the
* above succeeded, but let's be paranoid
*/
if (size > SGITMAXSIZE ||
prefixSize > SGITMAXPREFIXSIZE ||
nNodes > SGITMAXNNODES)
elog(ERROR, "SPGiST inner tuple header field is too small");
/* OK, form the tuple */
tup = (SpGistInnerTuple) palloc0(size);
tup->nNodes = nNodes;
tup->prefixSize = prefixSize;
tup->size = size;
if (hasPrefix)
memcpyDatum(SGITDATAPTR(tup), &state->attPrefixType, prefix);
ptr = (char *) SGITNODEPTR(tup);
for (i = 0; i < nNodes; i++)
{
SpGistNodeTuple node = nodes[i];
memcpy(ptr, node, IndexTupleSize(node));
ptr += IndexTupleSize(node);
}
return tup;
}
static Datum
gp_aovisimap_entry_internal(PG_FUNCTION_ARGS, Oid aoRelOid)
{
Datum values[4];
bool nulls[4];
HeapTuple tuple;
Datum result;
typedef struct Context
{
AppendOnlyVisimap visiMap;
Relation parentRelation;
IndexScanDesc indexScan;
text *bitmapBuffer;
} Context;
FuncCallContext *funcctx;
Context *context;
if (SRF_IS_FIRSTCALL())
{
TupleDesc tupdesc;
MemoryContext oldcontext;
/* create a function context for cross-call persistence */
funcctx = SRF_FIRSTCALL_INIT();
/*
* switch to memory context appropriate for multiple function
* calls
*/
oldcontext = MemoryContextSwitchTo(funcctx->multi_call_memory_ctx);
/* build tupdesc for result tuples */
tupdesc = CreateTemplateTupleDesc(4, false);
TupleDescInitEntry(tupdesc, (AttrNumber) 1, "segno",
INT4OID, -1, 0);
TupleDescInitEntry(tupdesc, (AttrNumber) 2, "first_row_num",
INT8OID, -1, 0);
TupleDescInitEntry(tupdesc, (AttrNumber) 3, "hidden_tupcount",
INT4OID, -1, 0);
TupleDescInitEntry(tupdesc, (AttrNumber) 4, "bitmap",
TEXTOID, -1, 0);
funcctx->tuple_desc = BlessTupleDesc(tupdesc);
/*
* Collect all the locking information that we will format and send
* out as a result set.
*/
context = (Context *) palloc0(sizeof(Context));
context->parentRelation = heap_open(aoRelOid, AccessShareLock);
if (!(RelationIsAoRows(context->parentRelation) || RelationIsAoCols(context->parentRelation)))
{
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("Function not supported on relation")));
}
AppendOnlyVisimap_Init(&context->visiMap,
context->parentRelation->rd_appendonly->visimaprelid,
context->parentRelation->rd_appendonly->visimapidxid,
AccessShareLock,
SnapshotNow);
context->indexScan = AppendOnlyVisimapStore_BeginScan(&
context->visiMap.visimapStore, 0, NULL);
context->bitmapBuffer = palloc0(VARHDRSZ + APPENDONLY_VISIMAP_MAX_RANGE + 1);
funcctx->user_fctx = (void *) context;
MemoryContextSwitchTo(oldcontext);
}
funcctx = SRF_PERCALL_SETUP();
context = (Context *) funcctx->user_fctx;
if (AppendOnlyVisimapStore_GetNext(&context->visiMap.visimapStore,
context->indexScan,
ForwardScanDirection,
&context->visiMap.visimapEntry,
NULL))
{
AppendOnlyVisimapEntry *visimapEntry = &context->visiMap.visimapEntry;
MemSet(values, 0, sizeof(values));
MemSet(nulls, false, sizeof(nulls));
values[0] = Int32GetDatum(visimapEntry->segmentFileNum);
values[1] = Int64GetDatum(visimapEntry->firstRowNum);
values[2] = Int32GetDatum(
(int32)AppendOnlyVisimapEntry_GetHiddenTupleCount(visimapEntry));
gp_aovisimap_encode_bitmap(VARDATA(context->bitmapBuffer),
visimapEntry->bitmap);
SET_VARSIZE(context->bitmapBuffer, APPENDONLY_VISIMAP_MAX_RANGE);
values[3] = PointerGetDatum(context->bitmapBuffer);
tuple = heap_form_tuple(funcctx->tuple_desc, values, nulls);
result = HeapTupleGetDatum(tuple);
SRF_RETURN_NEXT(funcctx, result);
}
AppendOnlyVisimapStore_EndScan(&context->visiMap.visimapStore,
context->indexScan);
AppendOnlyVisimap_Finish(&context->visiMap, AccessShareLock);
heap_close(context->parentRelation, AccessShareLock);
pfree(context->bitmapBuffer);
pfree(context);
funcctx->user_fctx = NULL;
SRF_RETURN_DONE(funcctx);
}
static Datum
gp_aovisimap_hidden_info_internal(PG_FUNCTION_ARGS, Oid aoRelOid)
{
Datum values[3];
bool nulls[3];
HeapTuple tuple;
Datum result;
typedef struct Context
{
AppendOnlyVisimap visiMap;
Relation parentRelation;
FileSegInfo **appendonlySegfileInfo;
AOCSFileSegInfo **aocsSegfileInfo;
int segfile_info_total;
int i;
} Context;
FuncCallContext *funcctx;
Context *context;
if (SRF_IS_FIRSTCALL())
{
TupleDesc tupdesc;
MemoryContext oldcontext;
/* create a function context for cross-call persistence */
funcctx = SRF_FIRSTCALL_INIT();
/*
* switch to memory context appropriate for multiple function
* calls
*/
oldcontext = MemoryContextSwitchTo(funcctx->multi_call_memory_ctx);
/* build tupdesc for result tuples */
tupdesc = CreateTemplateTupleDesc(3, false);
TupleDescInitEntry(tupdesc, (AttrNumber) 1, "segno",
INT4OID, -1, 0);
TupleDescInitEntry(tupdesc, (AttrNumber) 2, "hidden_tupcount",
INT8OID, -1, 0);
TupleDescInitEntry(tupdesc, (AttrNumber) 3, "total_tupcount",
INT8OID, -1, 0);
funcctx->tuple_desc = BlessTupleDesc(tupdesc);
/*
* Collect all the locking information that we will format and send
* out as a result set.
*/
context = (Context *) palloc0(sizeof(Context));
context->parentRelation = heap_open(aoRelOid, AccessShareLock);
if (!(RelationIsAoRows(context->parentRelation) || RelationIsAoCols(context->parentRelation)))
{
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("Function not supported on relation")));
}
if (RelationIsAoRows(context->parentRelation))
{
context->appendonlySegfileInfo = GetAllFileSegInfo(
context->parentRelation,
SnapshotNow,
&context->segfile_info_total);
}
else
{
Assert(RelationIsAoCols(context->parentRelation));
context->aocsSegfileInfo = GetAllAOCSFileSegInfo(context->parentRelation,
SnapshotNow, &context->segfile_info_total);
}
context->i = 0;
AppendOnlyVisimap_Init(&context->visiMap,
context->parentRelation->rd_appendonly->visimaprelid,
context->parentRelation->rd_appendonly->visimapidxid,
AccessShareLock,
SnapshotNow);
funcctx->user_fctx = (void *) context;
MemoryContextSwitchTo(oldcontext);
}
funcctx = SRF_PERCALL_SETUP();
context = (Context *) funcctx->user_fctx;
while (context->i < context->segfile_info_total)
{
int64 tupcount;
int segno;
if (context->appendonlySegfileInfo)
{
FileSegInfo *fsinfo = context->appendonlySegfileInfo[context->i];
tupcount = fsinfo->total_tupcount;
segno = fsinfo->segno;
}
else if (context->aocsSegfileInfo)
{
AOCSFileSegInfo *fsinfo = context->aocsSegfileInfo[context->i];
tupcount = fsinfo->total_tupcount;
segno = fsinfo->segno;
}
else
{
Insist(false);
}
MemSet(values, 0, sizeof(values));
MemSet(nulls, false, sizeof(nulls));
values[0] = Int32GetDatum(segno);
values[1] = Int64GetDatum(AppendOnlyVisimap_GetSegmentFileHiddenTupleCount(
&context->visiMap, segno));
values[2] = Int64GetDatum(tupcount);
tuple = heap_form_tuple(funcctx->tuple_desc, values, nulls);
result = HeapTupleGetDatum(tuple);
context->i++;
SRF_RETURN_NEXT(funcctx, result);
}
AppendOnlyVisimap_Finish(&context->visiMap, AccessShareLock);
if (context->appendonlySegfileInfo)
{
FreeAllSegFileInfo(context->appendonlySegfileInfo, context->segfile_info_total);
pfree(context->appendonlySegfileInfo);
context->appendonlySegfileInfo = NULL;
}
if (context->aocsSegfileInfo)
{
FreeAllAOCSSegFileInfo(context->aocsSegfileInfo, context->segfile_info_total);
pfree(context->aocsSegfileInfo);
context->aocsSegfileInfo = NULL;
}
heap_close(context->parentRelation, AccessShareLock);
pfree(context);
funcctx->user_fctx = NULL;
SRF_RETURN_DONE(funcctx);
}
Datum
ginbulkdelete(PG_FUNCTION_ARGS)
{
IndexVacuumInfo *info = (IndexVacuumInfo *) PG_GETARG_POINTER(0);
IndexBulkDeleteResult *stats = (IndexBulkDeleteResult *) PG_GETARG_POINTER(1);
IndexBulkDeleteCallback callback = (IndexBulkDeleteCallback) PG_GETARG_POINTER(2);
void *callback_state = (void *) PG_GETARG_POINTER(3);
Relation index = info->index;
BlockNumber blkno = GIN_ROOT_BLKNO;
GinVacuumState gvs;
Buffer buffer;
BlockNumber rootOfPostingTree[BLCKSZ / (sizeof(IndexTupleData) + sizeof(ItemId))];
uint32 nRoot;
gvs.tmpCxt = AllocSetContextCreate(CurrentMemoryContext,
"Gin vacuum temporary context",
ALLOCSET_DEFAULT_MINSIZE,
ALLOCSET_DEFAULT_INITSIZE,
ALLOCSET_DEFAULT_MAXSIZE);
gvs.index = index;
gvs.callback = callback;
gvs.callback_state = callback_state;
gvs.strategy = info->strategy;
initGinState(&gvs.ginstate, index);
/* first time through? */
if (stats == NULL)
{
/* Yes, so initialize stats to zeroes */
stats = (IndexBulkDeleteResult *) palloc0(sizeof(IndexBulkDeleteResult));
/* and cleanup any pending inserts */
ginInsertCleanup(&gvs.ginstate, true, stats);
}
/* we'll re-count the tuples each time */
stats->num_index_tuples = 0;
gvs.result = stats;
buffer = ReadBufferExtended(index, MAIN_FORKNUM, blkno,
RBM_NORMAL, info->strategy);
/* find leaf page */
for (;;)
{
Page page = BufferGetPage(buffer);
IndexTuple itup;
LockBuffer(buffer, GIN_SHARE);
Assert(!GinPageIsData(page));
if (GinPageIsLeaf(page))
{
LockBuffer(buffer, GIN_UNLOCK);
LockBuffer(buffer, GIN_EXCLUSIVE);
if (blkno == GIN_ROOT_BLKNO && !GinPageIsLeaf(page))
{
LockBuffer(buffer, GIN_UNLOCK);
continue; /* check it one more */
}
break;
}
Assert(PageGetMaxOffsetNumber(page) >= FirstOffsetNumber);
itup = (IndexTuple) PageGetItem(page, PageGetItemId(page, FirstOffsetNumber));
blkno = GinGetDownlink(itup);
Assert(blkno != InvalidBlockNumber);
UnlockReleaseBuffer(buffer);
buffer = ReadBufferExtended(index, MAIN_FORKNUM, blkno,
RBM_NORMAL, info->strategy);
}
/* right now we found leftmost page in entry's BTree */
for (;;)
{
Page page = BufferGetPage(buffer);
Page resPage;
uint32 i;
Assert(!GinPageIsData(page));
resPage = ginVacuumEntryPage(&gvs, buffer, rootOfPostingTree, &nRoot);
blkno = GinPageGetOpaque(page)->rightlink;
if (resPage)
{
START_CRIT_SECTION();
PageRestoreTempPage(resPage, page);
MarkBufferDirty(buffer);
xlogVacuumPage(gvs.index, buffer);
UnlockReleaseBuffer(buffer);
END_CRIT_SECTION();
}
else
{
UnlockReleaseBuffer(buffer);
}
vacuum_delay_point();
for (i = 0; i < nRoot; i++)
{
ginVacuumPostingTree(&gvs, rootOfPostingTree[i]);
vacuum_delay_point();
}
if (blkno == InvalidBlockNumber) /* rightmost page */
break;
buffer = ReadBufferExtended(index, MAIN_FORKNUM, blkno,
RBM_NORMAL, info->strategy);
LockBuffer(buffer, GIN_EXCLUSIVE);
}
MemoryContextDelete(gvs.tmpCxt);
PG_RETURN_POINTER(gvs.result);
}
/*
* 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;
bool isempty;
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 = GinDataPageGetPostingItem(page, i);
if (ginScanToDelete(gvs, PostingItemGetBlockNumber(pitem), FALSE, me, i))
i--;
}
}
if (GinPageIsLeaf(page))
isempty = GinDataLeafPageIsEmpty(page);
else
isempty = GinPageGetOpaque(page)->maxoff < FirstOffsetNumber;
if (isempty)
{
/* we never delete the left- or rightmost branch */
if (me->leftBlkno != InvalidBlockNumber && !GinPageRightMost(page))
{
Assert(!isRoot);
ginDeletePage(gvs, blkno, me->leftBlkno, me->parent->blkno, myoff, me->parent->isRoot);
meDelete = TRUE;
}
}
ReleaseBuffer(buffer);
if (!meDelete)
me->leftBlkno = blkno;
return meDelete;
}
/*
* Actually do a base backup for the specified tablespaces.
*
* This is split out mainly to avoid complaints about "variable might be
* clobbered by longjmp" from stupider versions of gcc.
*/
static void
perform_base_backup(basebackup_options *opt, DIR *tblspcdir)
{
XLogRecPtr startptr;
TimeLineID starttli;
XLogRecPtr endptr;
TimeLineID endtli;
char *labelfile;
int datadirpathlen;
datadirpathlen = strlen(DataDir);
backup_started_in_recovery = RecoveryInProgress();
startptr = do_pg_start_backup(opt->label, opt->fastcheckpoint, &starttli,
&labelfile);
SendXlogRecPtrResult(startptr, starttli);
/*
* Calculate the relative path of temporary statistics directory
* in order to skip the files which are located in that directory later.
*/
if (is_absolute_path(pgstat_stat_directory) &&
strncmp(pgstat_stat_directory, DataDir, datadirpathlen) == 0)
statrelpath = psprintf("./%s", pgstat_stat_directory + datadirpathlen + 1);
else if (strncmp(pgstat_stat_directory, "./", 2) != 0)
statrelpath = psprintf("./%s", pgstat_stat_directory);
else
statrelpath = pgstat_stat_directory;
PG_ENSURE_ERROR_CLEANUP(base_backup_cleanup, (Datum) 0);
{
List *tablespaces = NIL;
ListCell *lc;
struct dirent *de;
tablespaceinfo *ti;
/* Collect information about all tablespaces */
while ((de = ReadDir(tblspcdir, "pg_tblspc")) != NULL)
{
char fullpath[MAXPGPATH];
char linkpath[MAXPGPATH];
char *relpath = NULL;
int rllen;
/* Skip special stuff */
if (strcmp(de->d_name, ".") == 0 || strcmp(de->d_name, "..") == 0)
continue;
snprintf(fullpath, sizeof(fullpath), "pg_tblspc/%s", de->d_name);
#if defined(HAVE_READLINK) || defined(WIN32)
rllen = readlink(fullpath, linkpath, sizeof(linkpath));
if (rllen < 0)
{
ereport(WARNING,
(errmsg("could not read symbolic link \"%s\": %m",
fullpath)));
continue;
}
else if (rllen >= sizeof(linkpath))
{
ereport(WARNING,
(errmsg("symbolic link \"%s\" target is too long",
fullpath)));
continue;
}
linkpath[rllen] = '\0';
/*
* Relpath holds the relative path of the tablespace directory
* when it's located within PGDATA, or NULL if it's located
* elsewhere.
*/
if (rllen > datadirpathlen &&
strncmp(linkpath, DataDir, datadirpathlen) == 0 &&
IS_DIR_SEP(linkpath[datadirpathlen]))
relpath = linkpath + datadirpathlen + 1;
ti = palloc(sizeof(tablespaceinfo));
ti->oid = pstrdup(de->d_name);
ti->path = pstrdup(linkpath);
ti->rpath = relpath ? pstrdup(relpath) : NULL;
ti->size = opt->progress ? sendTablespace(fullpath, true) : -1;
tablespaces = lappend(tablespaces, ti);
#else
/*
* If the platform does not have symbolic links, it should not be
* possible to have tablespaces - clearly somebody else created
* them. Warn about it and ignore.
*/
ereport(WARNING,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("tablespaces are not supported on this platform")));
#endif
}
/* Add a node for the base directory at the end */
ti = palloc0(sizeof(tablespaceinfo));
ti->size = opt->progress ? sendDir(".", 1, true, tablespaces) : -1;
tablespaces = lappend(tablespaces, ti);
/* Send tablespace header */
SendBackupHeader(tablespaces);
/* Send off our tablespaces one by one */
foreach(lc, tablespaces)
{
tablespaceinfo *ti = (tablespaceinfo *) lfirst(lc);
StringInfoData buf;
/* Send CopyOutResponse message */
pq_beginmessage(&buf, 'H');
pq_sendbyte(&buf, 0); /* overall format */
pq_sendint(&buf, 0, 2); /* natts */
pq_endmessage(&buf);
if (ti->path == NULL)
{
struct stat statbuf;
/* In the main tar, include the backup_label first... */
sendFileWithContent(BACKUP_LABEL_FILE, labelfile);
/* ... then the bulk of the files ... */
sendDir(".", 1, false, tablespaces);
/* ... and pg_control after everything else. */
if (lstat(XLOG_CONTROL_FILE, &statbuf) != 0)
ereport(ERROR,
(errcode_for_file_access(),
errmsg("could not stat control file \"%s\": %m",
XLOG_CONTROL_FILE)));
sendFile(XLOG_CONTROL_FILE, XLOG_CONTROL_FILE, &statbuf, false);
}
else
sendTablespace(ti->path, false);
/*
* If we're including WAL, and this is the main data directory we
* don't terminate the tar stream here. Instead, we will append
* the xlog files below and terminate it then. This is safe since
* the main data directory is always sent *last*.
*/
if (opt->includewal && ti->path == NULL)
{
Assert(lnext(lc) == NULL);
}
else
pq_putemptymessage('c'); /* CopyDone */
}
}
/* ----------------------------------------------------------------
* ExecHashTableCreate
*
* create an empty hashtable data structure for hashjoin.
* ----------------------------------------------------------------
*/
HashJoinTable
ExecHashTableCreate(Hash *node, List *hashOperators, bool keepNulls)
{
HashJoinTable hashtable;
Plan *outerNode;
int nbuckets;
int nbatch;
int num_skew_mcvs;
int log2_nbuckets;
int nkeys;
int i;
ListCell *ho;
MemoryContext oldcxt;
/*
* Get information about the size of the relation to be hashed (it's the
* "outer" subtree of this node, but the inner relation of the hashjoin).
* Compute the appropriate size of the hash table.
*/
outerNode = outerPlan(node);
ExecChooseHashTableSize(outerNode->plan_rows, outerNode->plan_width,
OidIsValid(node->skewTable),
&nbuckets, &nbatch, &num_skew_mcvs);
#ifdef HJDEBUG
printf("nbatch = %d, nbuckets = %d\n", nbatch, nbuckets);
#endif
/* nbuckets must be a power of 2 */
log2_nbuckets = my_log2(nbuckets);
Assert(nbuckets == (1 << log2_nbuckets));
/*
* Initialize the hash table control block.
*
* The hashtable control block is just palloc'd from the executor's
* per-query memory context.
*/
hashtable = (HashJoinTable) palloc(sizeof(HashJoinTableData));
hashtable->nbuckets = nbuckets;
hashtable->log2_nbuckets = log2_nbuckets;
hashtable->buckets = NULL;
hashtable->keepNulls = keepNulls;
hashtable->skewEnabled = false;
hashtable->skewBucket = NULL;
hashtable->skewBucketLen = 0;
hashtable->nSkewBuckets = 0;
hashtable->skewBucketNums = NULL;
hashtable->nbatch = nbatch;
hashtable->curbatch = 0;
hashtable->nbatch_original = nbatch;
hashtable->nbatch_outstart = nbatch;
hashtable->growEnabled = true;
hashtable->totalTuples = 0;
hashtable->innerBatchFile = NULL;
hashtable->outerBatchFile = NULL;
hashtable->spaceUsed = 0;
hashtable->spacePeak = 0;
hashtable->spaceAllowed = work_mem * 1024L;
hashtable->spaceUsedSkew = 0;
hashtable->spaceAllowedSkew =
hashtable->spaceAllowed * SKEW_WORK_MEM_PERCENT / 100;
// cs3223, allocate memory for bitvector
//printf(" Allocating memory for bitvector \n");
hashtable->bitvector = (int*) palloc0(bitvector_size*1024); // sizeof(int)/32 = 1/8
hashtable->numBVfilter = 0;
hashtable->numProbNotJoin = 0;
hashtable->firstCheck = 0;
//printf("zero element: %d\n",hashtable->bitvector[0]);
/*
* Get info about the hash functions to be used for each hash key. Also
* remember whether the join operators are strict.
*/
nkeys = list_length(hashOperators);
hashtable->outer_hashfunctions =
(FmgrInfo *) palloc(nkeys * sizeof(FmgrInfo));
hashtable->inner_hashfunctions =
(FmgrInfo *) palloc(nkeys * sizeof(FmgrInfo));
hashtable->hashStrict = (bool *) palloc(nkeys * sizeof(bool));
i = 0;
foreach(ho, hashOperators)
{
Oid hashop = lfirst_oid(ho);
Oid left_hashfn;
Oid right_hashfn;
if (!get_op_hash_functions(hashop, &left_hashfn, &right_hashfn))
elog(ERROR, "could not find hash function for hash operator %u",
hashop);
fmgr_info(left_hashfn, &hashtable->outer_hashfunctions[i]);
fmgr_info(right_hashfn, &hashtable->inner_hashfunctions[i]);
hashtable->hashStrict[i] = op_strict(hashop);
i++;
}
/*
* setup_regexp_matches --- do the initial matching for regexp_matches()
* or regexp_split()
*
* To avoid having to re-find the compiled pattern on each call, we do
* all the matching in one swoop. The returned regexp_matches_ctx contains
* the locations of all the substrings matching the pattern.
*
* The four bool parameters have only two patterns (one for matching, one for
* splitting) but it seems clearer to distinguish the functionality this way
* than to key it all off one "is_split" flag. We don't currently assume that
* fetching_unmatched is exclusive of fetching the matched text too; if it's
* set, the conversion buffer is large enough to fetch any single matched or
* unmatched string, but not any larger substring. (In practice, when splitting
* the matches are usually small anyway, and it didn't seem worth complicating
* the code further.)
*/
static regexp_matches_ctx *
setup_regexp_matches(text *orig_str, text *pattern, text *flags,
Oid collation,
bool force_glob,
bool use_subpatterns,
bool ignore_degenerate,
bool fetching_unmatched)
{
regexp_matches_ctx *matchctx = palloc0(sizeof(regexp_matches_ctx));
int eml = pg_database_encoding_max_length();
int orig_len;
pg_wchar *wide_str;
int wide_len;
pg_re_flags re_flags;
regex_t *cpattern;
regmatch_t *pmatch;
int pmatch_len;
int array_len;
int array_idx;
int prev_match_end;
int prev_valid_match_end;
int start_search;
int maxlen = 0; /* largest fetch length in characters */
/* save original string --- we'll extract result substrings from it */
matchctx->orig_str = orig_str;
/* convert string to pg_wchar form for matching */
orig_len = VARSIZE_ANY_EXHDR(orig_str);
wide_str = (pg_wchar *) palloc(sizeof(pg_wchar) * (orig_len + 1));
wide_len = pg_mb2wchar_with_len(VARDATA_ANY(orig_str), wide_str, orig_len);
/* determine options */
parse_re_flags(&re_flags, flags);
if (force_glob)
{
/* user mustn't specify 'g' for regexp_split */
if (re_flags.glob)
ereport(ERROR,
(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
errmsg("regexp_split does not support the global option")));
/* but we find all the matches anyway */
re_flags.glob = true;
}
/* set up the compiled pattern */
cpattern = RE_compile_and_cache(pattern, re_flags.cflags, collation);
/* do we want to remember subpatterns? */
if (use_subpatterns && cpattern->re_nsub > 0)
{
matchctx->npatterns = cpattern->re_nsub;
pmatch_len = cpattern->re_nsub + 1;
}
else
{
use_subpatterns = false;
matchctx->npatterns = 1;
pmatch_len = 1;
}
/* temporary output space for RE package */
pmatch = palloc(sizeof(regmatch_t) * pmatch_len);
/*
* the real output space (grown dynamically if needed)
*
* use values 2^n-1, not 2^n, so that we hit the limit at 2^28-1 rather
* than at 2^27
*/
array_len = re_flags.glob ? 255 : 31;
matchctx->match_locs = (int *) palloc(sizeof(int) * array_len);
array_idx = 0;
/* search for the pattern, perhaps repeatedly */
prev_match_end = 0;
prev_valid_match_end = 0;
start_search = 0;
while (RE_wchar_execute(cpattern, wide_str, wide_len, start_search,
pmatch_len, pmatch))
{
/*
* If requested, ignore degenerate matches, which are zero-length
* matches occurring at the start or end of a string or just after a
* previous match.
*/
if (!ignore_degenerate ||
(pmatch[0].rm_so < wide_len &&
pmatch[0].rm_eo > prev_match_end))
{
/* enlarge output space if needed */
while (array_idx + matchctx->npatterns * 2 + 1 > array_len)
{
array_len += array_len + 1; /* 2^n-1 => 2^(n+1)-1 */
if (array_len > MaxAllocSize/sizeof(int))
ereport(ERROR,
(errcode(ERRCODE_PROGRAM_LIMIT_EXCEEDED),
errmsg("too many regular expression matches")));
matchctx->match_locs = (int *) repalloc(matchctx->match_locs,
sizeof(int) * array_len);
}
/* save this match's locations */
if (use_subpatterns)
{
int i;
for (i = 1; i <= matchctx->npatterns; i++)
{
int so = pmatch[i].rm_so;
int eo = pmatch[i].rm_eo;
matchctx->match_locs[array_idx++] = so;
matchctx->match_locs[array_idx++] = eo;
if (so >= 0 && eo >= 0 && (eo - so) > maxlen)
maxlen = (eo - so);
}
}
else
{
int so = pmatch[0].rm_so;
int eo = pmatch[0].rm_eo;
matchctx->match_locs[array_idx++] = so;
matchctx->match_locs[array_idx++] = eo;
if (so >= 0 && eo >= 0 && (eo - so) > maxlen)
maxlen = (eo - so);
}
matchctx->nmatches++;
/*
* check length of unmatched portion between end of previous valid
* (nondegenerate, or degenerate but not ignored) match and start
* of current one
*/
if (fetching_unmatched &&
pmatch[0].rm_so >= 0 &&
(pmatch[0].rm_so - prev_valid_match_end) > maxlen)
maxlen = (pmatch[0].rm_so - prev_valid_match_end);
prev_valid_match_end = pmatch[0].rm_eo;
}
prev_match_end = pmatch[0].rm_eo;
/* if not glob, stop after one match */
if (!re_flags.glob)
break;
/*
* Advance search position. Normally we start the next search at the
* end of the previous match; but if the match was of zero length, we
* have to advance by one character, or we'd just find the same match
* again.
*/
start_search = prev_match_end;
if (pmatch[0].rm_so == pmatch[0].rm_eo)
start_search++;
if (start_search > wide_len)
break;
}
/*
* check length of unmatched portion between end of last match and end of
* input string
*/
if (fetching_unmatched &&
(wide_len - prev_valid_match_end) > maxlen)
maxlen = (wide_len - prev_valid_match_end);
/*
* Keep a note of the end position of the string for the benefit of
* splitting code.
*/
matchctx->match_locs[array_idx] = wide_len;
if (eml > 1)
{
int64 maxsiz = eml * (int64) maxlen;
int conv_bufsiz;
/*
* Make the conversion buffer large enough for any substring of
* interest.
*
* Worst case: assume we need the maximum size (maxlen*eml), but take
* advantage of the fact that the original string length in bytes is an
* upper bound on the byte length of any fetched substring (and we know
* that len+1 is safe to allocate because the varlena header is longer
* than 1 byte).
*/
if (maxsiz > orig_len)
conv_bufsiz = orig_len + 1;
else
conv_bufsiz = maxsiz + 1; /* safe since maxsiz < 2^30 */
matchctx->conv_buf = palloc(conv_bufsiz);
matchctx->conv_bufsiz = conv_bufsiz;
matchctx->wide_str = wide_str;
}
else
{
/* No need to keep the wide string if we're in a single-byte charset. */
pfree(wide_str);
matchctx->wide_str = NULL;
matchctx->conv_buf = NULL;
matchctx->conv_bufsiz = 0;
}
/* Clean up temp storage */
pfree(pmatch);
return matchctx;
}
/*
* lazy_vacuum_rel() -- perform LAZY VACUUM for one heap relation
*
* This routine vacuums a single heap, cleans out its indexes, and
* updates its relpages and reltuples statistics.
*
* At entry, we have already established a transaction and opened
* and locked the relation.
*
* The return value indicates whether this function has held off
* interrupts -- caller must RESUME_INTERRUPTS() after commit if true.
*/
bool
lazy_vacuum_rel(Relation onerel, VacuumStmt *vacstmt,
BufferAccessStrategy bstrategy, List *updated_stats)
{
LVRelStats *vacrelstats;
Relation *Irel;
int nindexes;
BlockNumber possibly_freeable;
PGRUsage ru0;
TimestampTz starttime = 0;
bool heldoff = false;
pg_rusage_init(&ru0);
/* measure elapsed time iff autovacuum logging requires it */
if (IsAutoVacuumWorkerProcess() && Log_autovacuum_min_duration > 0)
starttime = GetCurrentTimestamp();
if (vacstmt->verbose)
elevel = INFO;
else
elevel = DEBUG2;
if (Gp_role == GP_ROLE_DISPATCH)
elevel = DEBUG2; /* vacuum and analyze messages aren't interesting from the QD */
#ifdef FAULT_INJECTOR
if (vacuumStatement_IsInAppendOnlyDropPhase(vacstmt))
{
FaultInjector_InjectFaultIfSet(
CompactionBeforeSegmentFileDropPhase,
DDLNotSpecified,
"", // databaseName
""); // tableName
}
if (vacummStatement_IsInAppendOnlyCleanupPhase(vacstmt))
{
FaultInjector_InjectFaultIfSet(
CompactionBeforeCleanupPhase,
DDLNotSpecified,
"", // databaseName
""); // tableName
}
#endif
/*
* MPP-23647. Update xid limits for heap as well as appendonly
* relations. This allows setting relfrozenxid to correct value
* for an appendonly (AO/CO) table.
*/
vac_strategy = bstrategy;
vacuum_set_xid_limits(vacstmt->freeze_min_age, onerel->rd_rel->relisshared,
&OldestXmin, &FreezeLimit);
/*
* Execute the various vacuum operations. Appendonly tables are treated
* differently.
*/
if (RelationIsAoRows(onerel) || RelationIsAoCols(onerel))
{
lazy_vacuum_aorel(onerel, vacstmt, updated_stats);
return false;
}
vacrelstats = (LVRelStats *) palloc0(sizeof(LVRelStats));
/* heap relation */
/* Set threshold for interesting free space = average request size */
/* XXX should we scale it up or down? Adjust vacuum.c too, if so */
vacrelstats->threshold = GetAvgFSMRequestSize(&onerel->rd_node);
vacrelstats->num_index_scans = 0;
/* Open all indexes of the relation */
vac_open_indexes(onerel, RowExclusiveLock, &nindexes, &Irel);
vacrelstats->hasindex = (nindexes > 0);
/* Do the vacuuming */
lazy_scan_heap(onerel, vacrelstats, Irel, nindexes, updated_stats, vacstmt->extra_oids);
/* Done with indexes */
vac_close_indexes(nindexes, Irel, NoLock);
/*
* Optionally truncate the relation.
*
* Don't even think about it unless we have a shot at releasing a goodly
* number of pages. Otherwise, the time taken isn't worth it.
*
* Note that after we've truncated the heap, it's too late to abort the
* transaction; doing so would lose the sinval messages needed to tell
* the other backends about the table being shrunk. We prevent interrupts
* in that case; caller is responsible for re-enabling them after
* committing the transaction.
*/
possibly_freeable = vacrelstats->rel_pages - vacrelstats->nonempty_pages;
if (possibly_freeable > 0 &&
(possibly_freeable >= REL_TRUNCATE_MINIMUM ||
possibly_freeable >= vacrelstats->rel_pages / REL_TRUNCATE_FRACTION))
{
HOLD_INTERRUPTS();
heldoff = true;
lazy_truncate_heap(onerel, vacrelstats);
}
/* Update shared free space map with final free space info */
lazy_update_fsm(onerel, vacrelstats);
if (vacrelstats->tot_free_pages > MaxFSMPages)
ereport(WARNING,
(errmsg("relation \"%s.%s\" contains more than \"max_fsm_pages\" pages with useful free space",
get_namespace_name(RelationGetNamespace(onerel)),
RelationGetRelationName(onerel)),
/* Only suggest VACUUM FULL if > 20% free */
(vacrelstats->tot_free_pages > vacrelstats->rel_pages * 0.20) ?
errhint("Consider using VACUUM FULL on this relation or increasing the configuration parameter \"max_fsm_pages\".") :
errhint("Consider increasing the configuration parameter \"max_fsm_pages\".")));
/* Update statistics in pg_class */
vac_update_relstats_from_list(onerel,
vacrelstats->rel_pages,
vacrelstats->rel_tuples,
vacrelstats->hasindex,
FreezeLimit,
updated_stats);
/* report results to the stats collector, too */
pgstat_report_vacuum(RelationGetRelid(onerel), onerel->rd_rel->relisshared,
true /*vacrelstats->scanned_all*/,
vacstmt->analyze, vacrelstats->rel_tuples);
if (gp_indexcheck_vacuum == INDEX_CHECK_ALL ||
(gp_indexcheck_vacuum == INDEX_CHECK_SYSTEM &&
PG_CATALOG_NAMESPACE == RelationGetNamespace(onerel)))
{
int i;
for (i = 0; i < nindexes; i++)
{
if (Irel[i]->rd_rel->relam == BTREE_AM_OID)
_bt_validate_vacuum(Irel[i], onerel, OldestXmin);
}
}
/* and log the action if appropriate */
if (IsAutoVacuumWorkerProcess() && Log_autovacuum_min_duration >= 0)
{
if (Log_autovacuum_min_duration == 0 ||
TimestampDifferenceExceeds(starttime, GetCurrentTimestamp(),
Log_autovacuum_min_duration))
ereport(LOG,
(errmsg("automatic vacuum of table \"%s.%s.%s\": index scans: %d\n"
"pages: %d removed, %d remain\n"
"tuples: %.0f removed, %.0f remain\n"
"system usage: %s",
get_database_name(MyDatabaseId),
get_namespace_name(RelationGetNamespace(onerel)),
RelationGetRelationName(onerel),
vacrelstats->num_index_scans,
vacrelstats->pages_removed, vacrelstats->rel_pages,
vacrelstats->tuples_deleted, vacrelstats->rel_tuples,
pg_rusage_show(&ru0))));
}
return heldoff;
}
/*
* Initializes the dictionary for use in backends - checks whether such dictionary
* and list of stopwords is already used, and if not then parses it and loads it into
* the shared segment.
*
* This is called through dispell_init() which is responsible for proper locking
* of the shared memory (using SegmentInfo->lock).
*/
static
void init_shared_dict(DictInfo * info, char * dictFile, char * affFile, char * stopFile) {
int size;
SharedIspellDict * shdict = NULL;
SharedStopList * shstop = NULL;
IspellDict * dict;
StopList stoplist;
/* DICTIONARY + AFFIXES */
/* TODO This should probably check that the filenames are not NULL, and maybe that
* it exists. Or maybe that's handled by the NIImport* functions. */
/* lookup if the dictionary (words and affixes) is already loaded in the shared segment */
shdict = get_shared_dict(dictFile, affFile);
/* load the dictionary / affixes if not yet defined */
if (shdict == NULL) {
dict = (IspellDict *)palloc0(sizeof(IspellDict));
NIStartBuild(dict);
NIImportDictionary(dict,
get_tsearch_config_filename(dictFile, "dict"));
NIImportAffixes(dict,
get_tsearch_config_filename(affFile, "affix"));
NISortDictionary(dict);
NISortAffixes(dict);
NIFinishBuild(dict);
/* check available space in shared segment */
size = sizeIspellDict(dict, dictFile, affFile);
if (size > segment_info->available)
elog(ERROR, "shared dictionary %s.dict / %s.affix needs %d B, only %ld B available",
dictFile, affFile, size, segment_info->available);
/* fine, there's enough space - copy the dictionary */
shdict = copyIspellDict(dict, dictFile, affFile, size, dict->nspell);
elog(INFO, "shared dictionary %s.dict / %s.affix loaded, used %d B, %ld B remaining",
dictFile, affFile, size, segment_info->available);
/* add the new dictionary to the linked list (of SharedIspellDict structures) */
shdict->next = segment_info->dict;
segment_info->dict = shdict;
}
/* STOP WORDS */
/* lookup if the stop words are already loaded in the shared segment, but only if there
* actually is a list */
if (stopFile != NULL) {
shstop = get_shared_stop_list(stopFile);
/* load the stopwords if not yet defined */
if (shstop == NULL) {
readstoplist(stopFile, &stoplist, lowerstr);
size = sizeStopList(&stoplist, stopFile);
if (size > segment_info->available) {
elog(ERROR, "shared stoplist %s.stop needs %d B, only %ld B available",
stopFile, size, segment_info->available);
}
/* fine, there's enough space - copy the stoplist */
shstop = copyStopList(&stoplist, stopFile, size);
elog(INFO, "shared stoplist %s.stop loaded, used %d B, %ld B remaining",
affFile, size, segment_info->available);
/* add the new stopword list to the linked list (of SharedStopList structures) */
shstop->next = segment_info->stop;
segment_info->stop = shstop;
}
}
/* Now, fill the DictInfo structure for the backend (references to dictionary,
* stopwords and the filenames). */
info->dict = shdict;
info->stop = shstop;
info->lookup = GetCurrentTimestamp();
memcpy(info->dictFile, dictFile, strlen(dictFile) + 1);
memcpy(info->affixFile, dictFile, strlen(affFile)+ 1);
memcpy(info->stopFile, dictFile, strlen(stopFile) + 1);
}
/*
* lazy_vacuum_aorel -- perform LAZY VACUUM for one Append-only relation.
*/
static void
lazy_vacuum_aorel(Relation onerel, VacuumStmt *vacstmt, List *updated_stats)
{
LVRelStats *vacrelstats;
bool update_relstats = true;
vacrelstats = (LVRelStats *) palloc0(sizeof(LVRelStats));
if (vacuumStatement_IsInAppendOnlyPreparePhase(vacstmt))
{
elogif(Debug_appendonly_print_compaction, LOG,
"Vacuum prepare phase %s", RelationGetRelationName(onerel));
vacuum_appendonly_indexes(onerel, vacstmt, updated_stats);
if (RelationIsAoRows(onerel))
AppendOnlyTruncateToEOF(onerel);
else
AOCSTruncateToEOF(onerel);
/*
* MPP-23647. For empty tables, we skip compaction phase
* and cleanup phase. Therefore, we update the stats
* (specifically, relfrozenxid) in prepare phase if the
* table is empty. Otherwise, the stats will be updated in
* the cleanup phase, when we would have computed the
* correct values for stats.
*/
if (vacstmt->appendonly_relation_empty)
{
update_relstats = true;
/*
* For an empty relation, the only stats we care about
* is relfrozenxid and relhasindex. We need to be
* mindful of correctly setting relhasindex here.
* relfrozenxid is already taken care of above by
* calling vacuum_set_xid_limits().
*/
vacrelstats->hasindex = onerel->rd_rel->relhasindex;
}
else
{
/*
* For a non-empty relation, follow the usual
* compaction phases and do not update stats in
* prepare phase.
*/
update_relstats = false;
}
}
else if (!vacummStatement_IsInAppendOnlyCleanupPhase(vacstmt))
{
vacuum_appendonly_rel(onerel, vacstmt);
update_relstats = false;
}
else
{
elogif(Debug_appendonly_print_compaction, LOG,
"Vacuum cleanup phase %s", RelationGetRelationName(onerel));
vacuum_appendonly_fill_stats(onerel, ActiveSnapshot,
&vacrelstats->rel_pages,
&vacrelstats->rel_tuples,
&vacrelstats->hasindex);
/* reset the remaining LVRelStats values */
vacrelstats->nonempty_pages = 0;
vacrelstats->num_dead_tuples = 0;
vacrelstats->max_dead_tuples = 0;
vacrelstats->tuples_deleted = 0;
vacrelstats->tot_free_pages = 0;
vacrelstats->fs_is_heap = false;
vacrelstats->num_free_pages = 0;
vacrelstats->max_free_pages = 0;
vacrelstats->pages_removed = 0;
}
if (update_relstats)
{
/* Update statistics in pg_class */
vac_update_relstats(onerel,
vacrelstats->rel_pages,
vacrelstats->rel_tuples,
vacrelstats->hasindex,
FreezeLimit,
updated_stats);
/* report results to the stats collector, too */
pgstat_report_vacuum(RelationGetRelid(onerel),
onerel->rd_rel->relisshared,
true /*vacrelstats->scanned_all*/,
vacstmt->analyze, vacrelstats->rel_tuples);
}
}
/*
* lazy_scan_heap() -- scan an open heap relation
*
* This routine sets commit status bits, builds lists of dead tuples
* and pages with free space, and calculates statistics on the number
* of live tuples in the heap. When done, or when we run low on space
* for dead-tuple TIDs, invoke vacuuming of indexes and heap.
*
* If there are no indexes then we just vacuum each dirty page as we
* process it, since there's no point in gathering many tuples.
*/
static void
lazy_scan_heap(Relation onerel, LVRelStats *vacrelstats,
Relation *Irel, int nindexes, List *updated_stats, List *all_extra_oids)
{
MIRROREDLOCK_BUFMGR_DECLARE;
BlockNumber nblocks,
blkno;
HeapTupleData tuple;
char *relname;
BlockNumber empty_pages,
vacuumed_pages;
double num_tuples,
tups_vacuumed,
nkeep,
nunused;
IndexBulkDeleteResult **indstats;
int i;
int reindex_count = 1;
PGRUsage ru0;
/* Fetch gp_persistent_relation_node information that will be added to XLOG record. */
RelationFetchGpRelationNodeForXLog(onerel);
pg_rusage_init(&ru0);
relname = RelationGetRelationName(onerel);
ereport(elevel,
(errmsg("vacuuming \"%s.%s\"",
get_namespace_name(RelationGetNamespace(onerel)),
relname)));
empty_pages = vacuumed_pages = 0;
num_tuples = tups_vacuumed = nkeep = nunused = 0;
indstats = (IndexBulkDeleteResult **)
palloc0(nindexes * sizeof(IndexBulkDeleteResult *));
nblocks = RelationGetNumberOfBlocks(onerel);
vacrelstats->rel_pages = nblocks;
vacrelstats->nonempty_pages = 0;
lazy_space_alloc(vacrelstats, nblocks);
for (blkno = 0; blkno < nblocks; blkno++)
{
Buffer buf;
Page page;
OffsetNumber offnum,
maxoff;
bool tupgone,
hastup;
int prev_dead_count;
OffsetNumber frozen[MaxOffsetNumber];
int nfrozen;
vacuum_delay_point();
/*
* If we are close to overrunning the available space for dead-tuple
* TIDs, pause and do a cycle of vacuuming before we tackle this page.
*/
if ((vacrelstats->max_dead_tuples - vacrelstats->num_dead_tuples) < MaxHeapTuplesPerPage &&
vacrelstats->num_dead_tuples > 0)
{
/* Remove index entries */
for (i = 0; i < nindexes; i++)
{
List *extra_oids = get_oids_for_bitmap(all_extra_oids, Irel[i],
onerel, reindex_count);
lazy_vacuum_index(Irel[i],
&indstats[i],
vacrelstats,
extra_oids);
list_free(extra_oids);
}
reindex_count++;
/* Remove tuples from heap */
lazy_vacuum_heap(onerel, vacrelstats);
/* Forget the now-vacuumed tuples, and press on */
vacrelstats->num_dead_tuples = 0;
}
/* -------- MirroredLock ---------- */
MIRROREDLOCK_BUFMGR_LOCK;
buf = ReadBuffer(onerel, blkno);
/* Initially, we only need shared access to the buffer */
LockBuffer(buf, BUFFER_LOCK_SHARE);
page = BufferGetPage(buf);
if (PageIsNew(page))
{
/*
* An all-zeroes page could be left over if a backend extends the
* relation but crashes before initializing the page. Reclaim such
* pages for use.
*
* We have to be careful here because we could be looking at a
* page that someone has just added to the relation and not yet
* been able to initialize (see RelationGetBufferForTuple). To
* protect against that, release the buffer lock, grab the
* relation extension lock momentarily, and re-lock the buffer. If
* the page is still uninitialized by then, it must be left over
* from a crashed backend, and we can initialize it.
*
* We don't really need the relation lock when this is a new or
* temp relation, but it's probably not worth the code space to
* check that, since this surely isn't a critical path.
*
* Note: the comparable code in vacuum.c need not worry because
* it's got exclusive lock on the whole relation.
*/
LockBuffer(buf, BUFFER_LOCK_UNLOCK);
MIRROREDLOCK_BUFMGR_UNLOCK;
/* -------- MirroredLock ---------- */
LockRelationForExtension(onerel, ExclusiveLock);
UnlockRelationForExtension(onerel, ExclusiveLock);
/* -------- MirroredLock ---------- */
MIRROREDLOCK_BUFMGR_LOCK;
LockBuffer(buf, BUFFER_LOCK_EXCLUSIVE); /* LockBufferForCleanup(buf)? */
if (PageIsNew(page))
{
ereport(WARNING,
(errmsg("relation \"%s\" page %u is uninitialized --- fixing",
relname, blkno)));
PageInit(page, BufferGetPageSize(buf), 0);
/* must record in xlog so that changetracking will know about this change */
log_heap_newpage(onerel, page, blkno);
empty_pages++;
lazy_record_free_space(vacrelstats, blkno,
PageGetFreeSpace(page));
}
MarkBufferDirty(buf);
UnlockReleaseBuffer(buf);
MIRROREDLOCK_BUFMGR_UNLOCK;
/* -------- MirroredLock ---------- */
continue;
}
if (PageIsEmpty(page))
{
empty_pages++;
lazy_record_free_space(vacrelstats, blkno,
PageGetFreeSpace(page));
UnlockReleaseBuffer(buf);
MIRROREDLOCK_BUFMGR_UNLOCK;
/* -------- MirroredLock ---------- */
continue;
}
nfrozen = 0;
hastup = false;
prev_dead_count = vacrelstats->num_dead_tuples;
maxoff = PageGetMaxOffsetNumber(page);
for (offnum = FirstOffsetNumber;
offnum <= maxoff;
offnum = OffsetNumberNext(offnum))
{
ItemId itemid;
itemid = PageGetItemId(page, offnum);
if (!ItemIdIsUsed(itemid))
{
nunused += 1;
continue;
}
tuple.t_data = (HeapTupleHeader) PageGetItem(page, itemid);
tuple.t_len = ItemIdGetLength(itemid);
ItemPointerSet(&(tuple.t_self), blkno, offnum);
tupgone = false;
switch (HeapTupleSatisfiesVacuum(tuple.t_data, OldestXmin, buf, false))
{
case HEAPTUPLE_DEAD:
tupgone = true; /* we can delete the tuple */
break;
case HEAPTUPLE_LIVE:
/* Tuple is good --- but let's do some validity checks */
if (onerel->rd_rel->relhasoids &&
!OidIsValid(HeapTupleGetOid(&tuple)))
elog(WARNING, "relation \"%s\" TID %u/%u: OID is invalid",
relname, blkno, offnum);
break;
case HEAPTUPLE_RECENTLY_DEAD:
/*
* If tuple is recently deleted then we must not remove it
* from relation.
*/
nkeep += 1;
break;
case HEAPTUPLE_INSERT_IN_PROGRESS:
/* This is an expected case during concurrent vacuum */
break;
case HEAPTUPLE_DELETE_IN_PROGRESS:
/* This is an expected case during concurrent vacuum */
break;
default:
elog(ERROR, "unexpected HeapTupleSatisfiesVacuum result");
break;
}
if (tupgone)
{
lazy_record_dead_tuple(vacrelstats, &(tuple.t_self));
tups_vacuumed += 1;
}
else
{
num_tuples += 1;
hastup = true;
/*
* Each non-removable tuple must be checked to see if it
* needs freezing. If we already froze anything, then
* we've already switched the buffer lock to exclusive.
*/
if (heap_freeze_tuple(tuple.t_data, FreezeLimit,
(nfrozen > 0) ? InvalidBuffer : buf))
frozen[nfrozen++] = offnum;
}
} /* scan along page */
/*
* If we froze any tuples, mark the buffer dirty, and write a WAL
* record recording the changes. We must log the changes to be
* crash-safe against future truncation of CLOG.
*/
if (nfrozen > 0)
{
MarkBufferDirty(buf);
/* no XLOG for temp tables, though */
if (!onerel->rd_istemp)
{
XLogRecPtr recptr;
recptr = log_heap_freeze(onerel, buf, FreezeLimit,
frozen, nfrozen);
PageSetLSN(page, recptr);
PageSetTLI(page, ThisTimeLineID);
}
}
/*
* If there are no indexes then we can vacuum the page right now
* instead of doing a second scan.
*/
if (nindexes == 0 &&
vacrelstats->num_dead_tuples > 0)
{
/* Trade in buffer share lock for super-exclusive lock */
LockBuffer(buf, BUFFER_LOCK_UNLOCK);
LockBufferForCleanup(buf);
/* Remove tuples from heap */
lazy_vacuum_page(onerel, blkno, buf, 0, vacrelstats);
/* Forget the now-vacuumed tuples, and press on */
vacrelstats->num_dead_tuples = 0;
vacuumed_pages++;
}
/*
* If we remembered any tuples for deletion, then the page will be
* visited again by lazy_vacuum_heap, which will compute and record
* its post-compaction free space. If not, then we're done with this
* page, so remember its free space as-is. (This path will always be
* taken if there are no indexes.)
*/
if (vacrelstats->num_dead_tuples == prev_dead_count)
{
lazy_record_free_space(vacrelstats, blkno,
PageGetFreeSpace(page));
}
/* Remember the location of the last page with nonremovable tuples */
if (hastup)
vacrelstats->nonempty_pages = blkno + 1;
UnlockReleaseBuffer(buf);
MIRROREDLOCK_BUFMGR_UNLOCK;
/* -------- MirroredLock ---------- */
}
/* save stats for use later */
vacrelstats->rel_tuples = num_tuples;
vacrelstats->tuples_deleted = tups_vacuumed;
/* If any tuples need to be deleted, perform final vacuum cycle */
/* XXX put a threshold on min number of tuples here? */
if (vacrelstats->num_dead_tuples > 0)
{
/* Remove index entries */
for (i = 0; i < nindexes; i++)
{
List *extra_oids = get_oids_for_bitmap(all_extra_oids, Irel[i],
onerel, reindex_count);
lazy_vacuum_index(Irel[i],
&indstats[i],
vacrelstats,
extra_oids);
list_free(extra_oids);
}
reindex_count++;
/* Remove tuples from heap */
lazy_vacuum_heap(onerel, vacrelstats);
}
/* Do post-vacuum cleanup and statistics update for each index */
for (i = 0; i < nindexes; i++)
lazy_cleanup_index(Irel[i], indstats[i], vacrelstats, updated_stats);
/* If no indexes, make log report that lazy_vacuum_heap would've made */
if (vacuumed_pages)
ereport(elevel,
(errmsg("\"%s\": removed %.0f row versions in %u pages",
RelationGetRelationName(onerel),
tups_vacuumed, vacuumed_pages)));
ereport(elevel,
(errmsg("\"%s\": found %.0f removable, %.0f nonremovable row versions in %u pages",
RelationGetRelationName(onerel),
tups_vacuumed, num_tuples, nblocks),
errdetail("%.0f dead row versions cannot be removed yet.\n"
"There were %.0f unused item pointers.\n"
"%u pages contain useful free space.\n"
"%u pages are entirely empty.\n"
"%s.",
nkeep,
nunused,
vacrelstats->tot_free_pages,
empty_pages,
pg_rusage_show(&ru0))));
if (vacrelstats->tot_free_pages > MaxFSMPages)
ereport(WARNING,
(errmsg("relation \"%s.%s\" contains more than \"max_fsm_pages\" pages with useful free space",
get_namespace_name(RelationGetNamespace(onerel)),
relname),
errhint("Consider compacting this relation or increasing the configuration parameter \"max_fsm_pages\".")));
}
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);
}
static Datum
gp_aovisimap_internal(PG_FUNCTION_ARGS, Oid aoRelOid)
{
Datum values[3];
bool nulls[3];
HeapTuple tuple;
Datum result;
typedef struct Context
{
Relation aorel;
AppendOnlyVisimapScan visiMapScan;
AOTupleId aoTupleId;
} Context;
FuncCallContext *funcctx;
Context *context;
if (SRF_IS_FIRSTCALL())
{
TupleDesc tupdesc;
MemoryContext oldcontext;
/* create a function context for cross-call persistence */
funcctx = SRF_FIRSTCALL_INIT();
/*
* switch to memory context appropriate for multiple function
* calls
*/
oldcontext = MemoryContextSwitchTo(funcctx->multi_call_memory_ctx);
/* build tupdesc for result tuples */
tupdesc = CreateTemplateTupleDesc(3, false);
TupleDescInitEntry(tupdesc, (AttrNumber) 1, "tid",
TIDOID, -1, 0);
TupleDescInitEntry(tupdesc, (AttrNumber) 2, "segno",
INT4OID, -1, 0);
TupleDescInitEntry(tupdesc, (AttrNumber) 3, "row_num",
INT8OID, -1, 0);
funcctx->tuple_desc = BlessTupleDesc(tupdesc);
/*
* Collect all the locking information that we will format and send
* out as a result set.
*/
context = (Context *) palloc0(sizeof(Context));
context->aorel = heap_open(aoRelOid, AccessShareLock);
if (!(RelationIsAoRows(context->aorel) || RelationIsAoCols(context->aorel)))
{
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("Function not supported on relation")));
}
AppendOnlyVisimapScan_Init(&context->visiMapScan,
context->aorel->rd_appendonly->visimaprelid,
context->aorel->rd_appendonly->visimapidxid,
AccessShareLock,
SnapshotNow);
AOTupleIdInit_Init(&context->aoTupleId);
funcctx->user_fctx = (void *) context;
MemoryContextSwitchTo(oldcontext);
}
funcctx = SRF_PERCALL_SETUP();
context = (Context *) funcctx->user_fctx;
while (true)
{
if (!AppendOnlyVisimapScan_GetNextInvisible(
&context->visiMapScan,
&context->aoTupleId))
{
break;
}
MemSet(values, 0, sizeof(values));
MemSet(nulls, false, sizeof(nulls));
values[0] = ItemPointerGetDatum((ItemPointer)&context->aoTupleId);
values[1] = Int32GetDatum(AOTupleIdGet_segmentFileNum(&context->aoTupleId));
values[2] = Int64GetDatum(AOTupleIdGet_rowNum(&context->aoTupleId));
tuple = heap_form_tuple(funcctx->tuple_desc, values, nulls);
result = HeapTupleGetDatum(tuple);
SRF_RETURN_NEXT(funcctx, result);
}
AppendOnlyVisimapScan_Finish(&context->visiMapScan, AccessShareLock);
heap_close(context->aorel, AccessShareLock);
pfree(context);
funcctx->user_fctx = NULL;
SRF_RETURN_DONE(funcctx);
}
/*
* get_relation_info -
* Retrieves catalog information for a given relation.
*
* Given the Oid of the relation, return the following info into fields
* of the RelOptInfo struct:
*
* min_attr lowest valid AttrNumber
* max_attr highest valid AttrNumber
* indexlist list of IndexOptInfos for relation's indexes
* pages number of pages
* tuples number of tuples
*
* Also, initialize the attr_needed[] and attr_widths[] arrays. In most
* cases these are left as zeroes, but sometimes we need to compute attr
* widths here, and we may as well cache the results for costsize.c.
*
* If inhparent is true, all we need to do is set up the attr arrays:
* the RelOptInfo actually represents the appendrel formed by an inheritance
* tree, and so the parent rel's physical size and index information isn't
* important for it.
*/
void
get_relation_info(PlannerInfo *root, Oid relationObjectId, bool inhparent,
RelOptInfo *rel)
{
Index varno = rel->relid;
Relation relation;
bool hasindex;
List *indexinfos = NIL;
bool needs_longlock;
/*
* We need not lock the relation since it was already locked, either by
* the rewriter or when expand_inherited_rtentry() added it to the query's
* rangetable.
*/
relation = heap_open(relationObjectId, NoLock);
needs_longlock = rel_needs_long_lock(relationObjectId);
rel->min_attr = FirstLowInvalidHeapAttributeNumber + 1;
rel->max_attr = RelationGetNumberOfAttributes(relation);
Assert(rel->max_attr >= rel->min_attr);
rel->attr_needed = (Relids *)
palloc0((rel->max_attr - rel->min_attr + 1) * sizeof(Relids));
rel->attr_widths = (int32 *)
palloc0((rel->max_attr - rel->min_attr + 1) * sizeof(int32));
/*
* CDB: Get partitioning key info for distributed relation.
*/
rel->cdbpolicy = RelationGetPartitioningKey(relation);
/*
* Estimate relation size --- unless it's an inheritance parent, in which
* case the size will be computed later in set_append_rel_pathlist, and we
* must leave it zero for now to avoid bollixing the total_table_pages
* calculation.
*/
if (!inhparent)
{
cdb_estimate_rel_size(
rel,
relation,
relation,
rel->attr_widths - rel->min_attr,
&rel->pages,
&rel->tuples,
&rel->cdb_default_stats_used
);
}
/*
* Make list of indexes. Ignore indexes on system catalogs if told to.
* Don't bother with indexes for an inheritance parent, either.
*/
if (inhparent ||
(IgnoreSystemIndexes && IsSystemClass(relation->rd_rel)))
hasindex = false;
else
hasindex = relation->rd_rel->relhasindex;
if (hasindex)
{
List *indexoidlist;
ListCell *l;
LOCKMODE lmode;
/* Warn if indexed table needs ANALYZE. */
if (rel->cdb_default_stats_used)
cdb_default_stats_warning_for_table(relation->rd_id);
indexoidlist = RelationGetIndexList(relation);
/*
* For each index, we get the same type of lock that the executor will
* need, and do not release it. This saves a couple of trips to the
* shared lock manager while not creating any real loss of
* concurrency, because no schema changes could be happening on the
* index while we hold lock on the parent rel, and neither lock type
* blocks any other kind of index operation.
*/
if (rel->relid == root->parse->resultRelation)
lmode = RowExclusiveLock;
else
lmode = AccessShareLock;
foreach(l, indexoidlist)
{
Oid indexoid = lfirst_oid(l);
Relation indexRelation;
Form_pg_index index;
IndexOptInfo *info;
int ncolumns;
int i;
/*
* Extract info from the relation descriptor for the index.
*/
indexRelation = index_open(indexoid, lmode);
index = indexRelation->rd_index;
/*
* Ignore invalid indexes, since they can't safely be used for
* queries. Note that this is OK because the data structure we
* are constructing is only used by the planner --- the executor
* still needs to insert into "invalid" indexes, if they're marked
* IndexIsReady.
*/
if (!IndexIsValid(index))
{
index_close(indexRelation, NoLock);
continue;
}
/*
* If the index is valid, but cannot yet be used, ignore it; but
* mark the plan we are generating as transient. See
* src/backend/access/heap/README.HOT for discussion.
*/
if (index->indcheckxmin &&
!TransactionIdPrecedes(HeapTupleHeaderGetXmin(indexRelation->rd_indextuple->t_data),
TransactionXmin))
{
root->glob->transientPlan = true;
index_close(indexRelation, NoLock);
continue;
}
info = makeNode(IndexOptInfo);
info->indexoid = index->indexrelid;
info->rel = rel;
info->ncolumns = ncolumns = index->indnatts;
/*
* Allocate per-column info arrays. To save a few palloc cycles
* we allocate all the Oid-type arrays in one request. Note that
* the opfamily array needs an extra, terminating zero at the end.
* We pre-zero the ordering info in case the index is unordered.
*/
info->indexkeys = (int *) palloc(sizeof(int) * ncolumns);
info->opfamily = (Oid *) palloc0(sizeof(Oid) * (4 * ncolumns + 1));
info->opcintype = info->opfamily + (ncolumns + 1);
info->fwdsortop = info->opcintype + ncolumns;
info->revsortop = info->fwdsortop + ncolumns;
info->nulls_first = (bool *) palloc0(sizeof(bool) * ncolumns);
for (i = 0; i < ncolumns; i++)
{
info->indexkeys[i] = index->indkey.values[i];
info->opfamily[i] = indexRelation->rd_opfamily[i];
info->opcintype[i] = indexRelation->rd_opcintype[i];
}
info->relam = indexRelation->rd_rel->relam;
info->amcostestimate = indexRelation->rd_am->amcostestimate;
info->amoptionalkey = indexRelation->rd_am->amoptionalkey;
info->amsearchnulls = indexRelation->rd_am->amsearchnulls;
/*
* Fetch the ordering operators associated with the index, if any.
* We expect that all ordering-capable indexes use btree's
* strategy numbers for the ordering operators.
*/
if (indexRelation->rd_am->amcanorder)
{
int nstrat = indexRelation->rd_am->amstrategies;
for (i = 0; i < ncolumns; i++)
{
int16 opt = indexRelation->rd_indoption[i];
int fwdstrat;
int revstrat;
if (opt & INDOPTION_DESC)
{
fwdstrat = BTGreaterStrategyNumber;
revstrat = BTLessStrategyNumber;
}
else
{
fwdstrat = BTLessStrategyNumber;
revstrat = BTGreaterStrategyNumber;
}
/*
* Index AM must have a fixed set of strategies for it to
* make sense to specify amcanorder, so we need not allow
* the case amstrategies == 0.
*/
if (fwdstrat > 0)
{
Assert(fwdstrat <= nstrat);
info->fwdsortop[i] = indexRelation->rd_operator[i * nstrat + fwdstrat - 1];
}
if (revstrat > 0)
{
Assert(revstrat <= nstrat);
info->revsortop[i] = indexRelation->rd_operator[i * nstrat + revstrat - 1];
}
info->nulls_first[i] = (opt & INDOPTION_NULLS_FIRST) != 0;
}
}
/*
* Fetch the index expressions and predicate, if any. We must
* modify the copies we obtain from the relcache to have the
* correct varno for the parent relation, so that they match up
* correctly against qual clauses.
*/
info->indexprs = RelationGetIndexExpressions(indexRelation);
info->indpred = RelationGetIndexPredicate(indexRelation);
if (info->indexprs && varno != 1)
ChangeVarNodes((Node *) info->indexprs, 1, varno, 0);
if (info->indpred && varno != 1)
ChangeVarNodes((Node *) info->indpred, 1, varno, 0);
info->predOK = false; /* set later in indxpath.c */
info->unique = index->indisunique;
/*
* Estimate the index size. If it's not a partial index, we lock
* the number-of-tuples estimate to equal the parent table; if it
* is partial then we have to use the same methods as we would for
* a table, except we can be sure that the index is not larger
* than the table.
*/
cdb_estimate_rel_size(rel,
relation,
indexRelation,
NULL,
&info->pages,
&info->tuples,
&info->cdb_default_stats_used);
if (!info->indpred ||
info->tuples > rel->tuples)
info->tuples = rel->tuples;
if (info->cdb_default_stats_used &&
!rel->cdb_default_stats_used)
cdb_default_stats_warning_for_index(relation->rd_id, indexoid);
index_close(indexRelation, needs_longlock ? NoLock : lmode);
indexinfos = lcons(info, indexinfos);
}
list_free(indexoidlist);
}
Datum
crosstab(PG_FUNCTION_ARGS)
{
char *sql = text_to_cstring(PG_GETARG_TEXT_PP(0));
ReturnSetInfo *rsinfo = (ReturnSetInfo *) fcinfo->resultinfo;
Tuplestorestate *tupstore;
TupleDesc tupdesc;
int call_cntr;
int max_calls;
AttInMetadata *attinmeta;
SPITupleTable *spi_tuptable;
TupleDesc spi_tupdesc;
bool firstpass;
char *lastrowid;
int i;
int num_categories;
MemoryContext per_query_ctx;
MemoryContext oldcontext;
int ret;
int proc;
/* check to see if caller supports us returning a tuplestore */
if (rsinfo == NULL || !IsA(rsinfo, ReturnSetInfo))
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("set-valued function called in context that cannot accept a set")));
if (!(rsinfo->allowedModes & SFRM_Materialize))
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("materialize mode required, but it is not " \
"allowed in this context")));
per_query_ctx = rsinfo->econtext->ecxt_per_query_memory;
/* Connect to SPI manager */
if ((ret = SPI_connect()) < 0)
/* internal error */
elog(ERROR, "crosstab: SPI_connect returned %d", ret);
/* Retrieve the desired rows */
ret = SPI_execute(sql, true, 0);
proc = SPI_processed;
/* If no qualifying tuples, fall out early */
if (ret != SPI_OK_SELECT || proc <= 0)
{
SPI_finish();
rsinfo->isDone = ExprEndResult;
PG_RETURN_NULL();
}
spi_tuptable = SPI_tuptable;
spi_tupdesc = spi_tuptable->tupdesc;
/*----------
* The provided SQL query must always return three columns.
*
* 1. rowname
* the label or identifier for each row in the final result
* 2. category
* the label or identifier for each column in the final result
* 3. values
* the value for each column in the final result
*----------
*/
if (spi_tupdesc->natts != 3)
ereport(ERROR,
(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
errmsg("invalid source data SQL statement"),
errdetail("The provided SQL must return 3 "
"columns: rowid, category, and values.")));
/* get a tuple descriptor for our result type */
switch (get_call_result_type(fcinfo, NULL, &tupdesc))
{
case TYPEFUNC_COMPOSITE:
/* success */
break;
case TYPEFUNC_RECORD:
/* failed to determine actual type of RECORD */
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("function returning record called in context "
"that cannot accept type record")));
break;
default:
/* result type isn't composite */
elog(ERROR, "return type must be a row type");
break;
}
/*
* Check that return tupdesc is compatible with the data we got from SPI,
* at least based on number and type of attributes
*/
if (!compatCrosstabTupleDescs(tupdesc, spi_tupdesc))
ereport(ERROR,
(errcode(ERRCODE_SYNTAX_ERROR),
errmsg("return and sql tuple descriptions are " \
"incompatible")));
/*
* switch to long-lived memory context
*/
oldcontext = MemoryContextSwitchTo(per_query_ctx);
/* make sure we have a persistent copy of the result tupdesc */
tupdesc = CreateTupleDescCopy(tupdesc);
/* initialize our tuplestore in long-lived context */
tupstore =
tuplestore_begin_heap(rsinfo->allowedModes & SFRM_Materialize_Random,
false, work_mem);
MemoryContextSwitchTo(oldcontext);
/*
* Generate attribute metadata needed later to produce tuples from raw C
* strings
*/
attinmeta = TupleDescGetAttInMetadata(tupdesc);
/* total number of tuples to be examined */
max_calls = proc;
/* the return tuple always must have 1 rowid + num_categories columns */
num_categories = tupdesc->natts - 1;
firstpass = true;
lastrowid = NULL;
for (call_cntr = 0; call_cntr < max_calls; call_cntr++)
{
bool skip_tuple = false;
char **values;
/* allocate and zero space */
values = (char **) palloc0((1 + num_categories) * sizeof(char *));
/*
* now loop through the sql results and assign each value in sequence
* to the next category
*/
for (i = 0; i < num_categories; i++)
{
HeapTuple spi_tuple;
char *rowid;
/* see if we've gone too far already */
if (call_cntr >= max_calls)
break;
/* get the next sql result tuple */
spi_tuple = spi_tuptable->vals[call_cntr];
/* get the rowid from the current sql result tuple */
rowid = SPI_getvalue(spi_tuple, spi_tupdesc, 1);
/*
* If this is the first pass through the values for this rowid,
* set the first column to rowid
*/
if (i == 0)
{
xpstrdup(values[0], rowid);
/*
* Check to see if the rowid is the same as that of the last
* tuple sent -- if so, skip this tuple entirely
*/
if (!firstpass && xstreq(lastrowid, rowid))
{
xpfree(rowid);
skip_tuple = true;
break;
}
}
/*
* If rowid hasn't changed on us, continue building the output
* tuple.
*/
if (xstreq(rowid, values[0]))
{
/*
* Get the next category item value, which is always attribute
* number three.
*
* Be careful to assign the value to the array index based on
* which category we are presently processing.
*/
values[1 + i] = SPI_getvalue(spi_tuple, spi_tupdesc, 3);
/*
* increment the counter since we consume a row for each
* category, but not for last pass because the outer loop will
* do that for us
*/
if (i < (num_categories - 1))
call_cntr++;
xpfree(rowid);
}
else
{
/*
* We'll fill in NULLs for the missing values, but we need to
* decrement the counter since this sql result row doesn't
* belong to the current output tuple.
*/
call_cntr--;
xpfree(rowid);
break;
}
}
if (!skip_tuple)
{
HeapTuple tuple;
/* build the tuple and store it */
tuple = BuildTupleFromCStrings(attinmeta, values);
tuplestore_puttuple(tupstore, tuple);
heap_freetuple(tuple);
}
/* Remember current rowid */
xpfree(lastrowid);
xpstrdup(lastrowid, values[0]);
firstpass = false;
/* Clean up */
for (i = 0; i < num_categories + 1; i++)
if (values[i] != NULL)
pfree(values[i]);
pfree(values);
}
/* let the caller know we're sending back a tuplestore */
rsinfo->returnMode = SFRM_Materialize;
rsinfo->setResult = tupstore;
rsinfo->setDesc = tupdesc;
/* release SPI related resources (and return to caller's context) */
SPI_finish();
return (Datum) 0;
}
/*
* ConstructTupleDescriptor
*
* Build an index tuple descriptor for a new index
*/
static TupleDesc
ConstructTupleDescriptor(Relation heapRelation,
IndexInfo *indexInfo,
Oid *classObjectId)
{
int numatts = indexInfo->ii_NumIndexAttrs;
List *indexprs = indexInfo->ii_Expressions;
TupleDesc heapTupDesc;
TupleDesc indexTupDesc;
int natts; /* #atts in heap rel --- for error checks */
int i;
heapTupDesc = RelationGetDescr(heapRelation);
natts = RelationGetForm(heapRelation)->relnatts;
/*
* allocate the new tuple descriptor
*/
indexTupDesc = CreateTemplateTupleDesc(numatts, false);
/*
* For simple index columns, we copy the pg_attribute row from the
* parent relation and modify it as necessary. For expressions we
* have to cons up a pg_attribute row the hard way.
*/
for (i = 0; i < numatts; i++)
{
AttrNumber atnum = indexInfo->ii_KeyAttrNumbers[i];
Form_pg_attribute to;
HeapTuple tuple;
Form_pg_type typeTup;
Oid keyType;
indexTupDesc->attrs[i] = to =
(Form_pg_attribute) palloc0(ATTRIBUTE_TUPLE_SIZE);
if (atnum != 0)
{
/* Simple index column */
Form_pg_attribute from;
if (atnum < 0)
{
/*
* here we are indexing on a system attribute (-1...-n)
*/
from = SystemAttributeDefinition(atnum,
heapRelation->rd_rel->relhasoids);
}
else
{
/*
* here we are indexing on a normal attribute (1...n)
*/
if (atnum > natts) /* safety check */
elog(ERROR, "invalid column number %d", atnum);
from = heapTupDesc->attrs[AttrNumberGetAttrOffset(atnum)];
}
/*
* now that we've determined the "from", let's copy the tuple
* desc data...
*/
memcpy(to, from, ATTRIBUTE_TUPLE_SIZE);
/*
* Fix the stuff that should not be the same as the underlying
* attr
*/
to->attnum = i + 1;
to->attstattarget = 0;
to->attcacheoff = -1;
to->attnotnull = false;
to->atthasdef = false;
to->attislocal = true;
to->attinhcount = 0;
}
else
{
/* Expressional index */
Node *indexkey;
if (indexprs == NIL) /* shouldn't happen */
elog(ERROR, "too few entries in indexprs list");
indexkey = (Node *) lfirst(indexprs);
indexprs = lnext(indexprs);
/*
* Make the attribute's name "pg_expresssion_nnn" (maybe think
* of something better later)
*/
sprintf(NameStr(to->attname), "pg_expression_%d", i + 1);
/*
* Lookup the expression type in pg_type for the type length
* etc.
*/
keyType = exprType(indexkey);
tuple = SearchSysCache(TYPEOID,
ObjectIdGetDatum(keyType),
0, 0, 0);
if (!HeapTupleIsValid(tuple))
elog(ERROR, "cache lookup failed for type %u", keyType);
typeTup = (Form_pg_type) GETSTRUCT(tuple);
/*
* Assign some of the attributes values. Leave the rest as 0.
*/
to->attnum = i + 1;
to->atttypid = keyType;
to->attlen = typeTup->typlen;
to->attbyval = typeTup->typbyval;
to->attstorage = typeTup->typstorage;
to->attalign = typeTup->typalign;
to->attcacheoff = -1;
to->atttypmod = -1;
to->attislocal = true;
ReleaseSysCache(tuple);
}
/*
* We do not yet have the correct relation OID for the index, so
* just set it invalid for now. InitializeAttributeOids() will
* fix it later.
*/
to->attrelid = InvalidOid;
/*
* Check the opclass to see if it provides a keytype (overriding
* the attribute type).
*/
tuple = SearchSysCache(CLAOID,
ObjectIdGetDatum(classObjectId[i]),
0, 0, 0);
if (!HeapTupleIsValid(tuple))
elog(ERROR, "cache lookup failed for opclass %u",
classObjectId[i]);
keyType = ((Form_pg_opclass) GETSTRUCT(tuple))->opckeytype;
ReleaseSysCache(tuple);
if (OidIsValid(keyType) && keyType != to->atttypid)
{
/* index value and heap value have different types */
tuple = SearchSysCache(TYPEOID,
ObjectIdGetDatum(keyType),
0, 0, 0);
if (!HeapTupleIsValid(tuple))
elog(ERROR, "cache lookup failed for type %u", keyType);
typeTup = (Form_pg_type) GETSTRUCT(tuple);
to->atttypid = keyType;
to->atttypmod = -1;
to->attlen = typeTup->typlen;
to->attbyval = typeTup->typbyval;
to->attalign = typeTup->typalign;
to->attstorage = typeTup->typstorage;
ReleaseSysCache(tuple);
}
}
return indexTupDesc;
}
/*
* Look to see if we have template information for the given language name.
*/
static PLTemplate *
find_language_template(const char *languageName)
{
PLTemplate *result;
Relation rel;
SysScanDesc scan;
ScanKeyData key;
HeapTuple tup;
rel = heap_open(PLTemplateRelationId, AccessShareLock);
ScanKeyInit(&key,
Anum_pg_pltemplate_tmplname,
BTEqualStrategyNumber, F_NAMEEQ,
CStringGetDatum(languageName));
scan = systable_beginscan(rel, PLTemplateNameIndexId, true,
SnapshotNow, 1, &key);
tup = systable_getnext(scan);
if (HeapTupleIsValid(tup))
{
Form_pg_pltemplate tmpl = (Form_pg_pltemplate) GETSTRUCT(tup);
Datum datum;
bool isnull;
result = (PLTemplate *) palloc0(sizeof(PLTemplate));
result->tmpltrusted = tmpl->tmpltrusted;
result->tmpldbacreate = tmpl->tmpldbacreate;
/* Remaining fields are variable-width so we need heap_getattr */
datum = heap_getattr(tup, Anum_pg_pltemplate_tmplhandler,
RelationGetDescr(rel), &isnull);
if (!isnull)
result->tmplhandler = TextDatumGetCString(datum);
datum = heap_getattr(tup, Anum_pg_pltemplate_tmplinline,
RelationGetDescr(rel), &isnull);
if (!isnull)
result->tmplinline = TextDatumGetCString(datum);
datum = heap_getattr(tup, Anum_pg_pltemplate_tmplvalidator,
RelationGetDescr(rel), &isnull);
if (!isnull)
result->tmplvalidator = TextDatumGetCString(datum);
datum = heap_getattr(tup, Anum_pg_pltemplate_tmpllibrary,
RelationGetDescr(rel), &isnull);
if (!isnull)
result->tmpllibrary = TextDatumGetCString(datum);
/* Ignore template if handler or library info is missing */
if (!result->tmplhandler || !result->tmpllibrary)
result = NULL;
}
else
result = NULL;
systable_endscan(scan);
heap_close(rel, AccessShareLock);
return result;
}
/*
* Bulk deletion of all index entries pointing to a set of heap tuples.
* The set of target tuples is specified via a callback routine that tells
* whether any given heap tuple (identified by ItemPointer) is being deleted.
*
* Result: a palloc'd struct containing statistical info for VACUUM displays.
*/
Datum
hashbulkdelete(PG_FUNCTION_ARGS)
{
IndexVacuumInfo *info = (IndexVacuumInfo *) PG_GETARG_POINTER(0);
IndexBulkDeleteResult *stats = (IndexBulkDeleteResult *) PG_GETARG_POINTER(1);
IndexBulkDeleteCallback callback = (IndexBulkDeleteCallback) PG_GETARG_POINTER(2);
void *callback_state = (void *) PG_GETARG_POINTER(3);
Relation rel = info->index;
double tuples_removed;
double num_index_tuples;
double orig_ntuples;
Bucket orig_maxbucket;
Bucket cur_maxbucket;
Bucket cur_bucket;
Buffer metabuf;
HashMetaPage metap;
HashMetaPageData local_metapage;
tuples_removed = 0;
num_index_tuples = 0;
/*
* Read the metapage to fetch original bucket and tuple counts. Also, we
* keep a copy of the last-seen metapage so that we can use its
* hashm_spares[] values to compute bucket page addresses. This is a bit
* hokey but perfectly safe, since the interesting entries in the spares
* array cannot change under us; and it beats rereading the metapage for
* each bucket.
*/
metabuf = _hash_getbuf(rel, HASH_METAPAGE, HASH_READ, LH_META_PAGE);
metap = HashPageGetMeta(BufferGetPage(metabuf));
orig_maxbucket = metap->hashm_maxbucket;
orig_ntuples = metap->hashm_ntuples;
memcpy(&local_metapage, metap, sizeof(local_metapage));
_hash_relbuf(rel, metabuf);
/* Scan the buckets that we know exist */
cur_bucket = 0;
cur_maxbucket = orig_maxbucket;
loop_top:
while (cur_bucket <= cur_maxbucket)
{
BlockNumber bucket_blkno;
BlockNumber blkno;
bool bucket_dirty = false;
/* Get address of bucket's start page */
bucket_blkno = BUCKET_TO_BLKNO(&local_metapage, cur_bucket);
/* Exclusive-lock the bucket so we can shrink it */
_hash_getlock(rel, bucket_blkno, HASH_EXCLUSIVE);
/* Shouldn't have any active scans locally, either */
if (_hash_has_active_scan(rel, cur_bucket))
elog(ERROR, "hash index has active scan during VACUUM");
/* Scan each page in bucket */
blkno = bucket_blkno;
while (BlockNumberIsValid(blkno))
{
Buffer buf;
Page page;
HashPageOpaque opaque;
OffsetNumber offno;
OffsetNumber maxoffno;
OffsetNumber deletable[MaxOffsetNumber];
int ndeletable = 0;
vacuum_delay_point();
buf = _hash_getbuf_with_strategy(rel, blkno, HASH_WRITE,
LH_BUCKET_PAGE | LH_OVERFLOW_PAGE,
info->strategy);
page = BufferGetPage(buf);
opaque = (HashPageOpaque) PageGetSpecialPointer(page);
Assert(opaque->hasho_bucket == cur_bucket);
/* Scan each tuple in page */
maxoffno = PageGetMaxOffsetNumber(page);
for (offno = FirstOffsetNumber;
offno <= maxoffno;
offno = OffsetNumberNext(offno))
{
IndexTuple itup;
ItemPointer htup;
itup = (IndexTuple) PageGetItem(page,
PageGetItemId(page, offno));
htup = &(itup->t_tid);
if (callback(htup, callback_state))
{
/* mark the item for deletion */
deletable[ndeletable++] = offno;
tuples_removed += 1;
}
else
num_index_tuples += 1;
}
/*
* Apply deletions and write page if needed, advance to next page.
*/
blkno = opaque->hasho_nextblkno;
if (ndeletable > 0)
{
PageIndexMultiDelete(page, deletable, ndeletable);
_hash_wrtbuf(rel, buf);
bucket_dirty = true;
}
else
_hash_relbuf(rel, buf);
}
/* If we deleted anything, try to compact free space */
if (bucket_dirty)
_hash_squeezebucket(rel, cur_bucket, bucket_blkno,
info->strategy);
/* Release bucket lock */
_hash_droplock(rel, bucket_blkno, HASH_EXCLUSIVE);
/* Advance to next bucket */
cur_bucket++;
}
/* Write-lock metapage and check for split since we started */
metabuf = _hash_getbuf(rel, HASH_METAPAGE, HASH_WRITE, LH_META_PAGE);
metap = HashPageGetMeta(BufferGetPage(metabuf));
if (cur_maxbucket != metap->hashm_maxbucket)
{
/* There's been a split, so process the additional bucket(s) */
cur_maxbucket = metap->hashm_maxbucket;
memcpy(&local_metapage, metap, sizeof(local_metapage));
_hash_relbuf(rel, metabuf);
goto loop_top;
}
/* Okay, we're really done. Update tuple count in metapage. */
if (orig_maxbucket == metap->hashm_maxbucket &&
orig_ntuples == metap->hashm_ntuples)
{
/*
* No one has split or inserted anything since start of scan, so
* believe our count as gospel.
*/
metap->hashm_ntuples = num_index_tuples;
}
else
{
/*
* Otherwise, our count is untrustworthy since we may have
* double-scanned tuples in split buckets. Proceed by dead-reckoning.
* (Note: we still return estimated_count = false, because using this
* count is better than not updating reltuples at all.)
*/
if (metap->hashm_ntuples > tuples_removed)
metap->hashm_ntuples -= tuples_removed;
else
metap->hashm_ntuples = 0;
num_index_tuples = metap->hashm_ntuples;
}
_hash_wrtbuf(rel, metabuf);
/* return statistics */
if (stats == NULL)
stats = (IndexBulkDeleteResult *) palloc0(sizeof(IndexBulkDeleteResult));
stats->estimated_count = false;
stats->num_index_tuples = num_index_tuples;
stats->tuples_removed += tuples_removed;
/* hashvacuumcleanup will fill in num_pages */
PG_RETURN_POINTER(stats);
}
/*
* CompactCheckpointerRequestQueue
* Remove duplicates from the request queue to avoid backend fsyncs.
* Returns "true" if any entries were removed.
*
* Although a full fsync request queue is not common, it can lead to severe
* performance problems when it does happen. So far, this situation has
* only been observed to occur when the system is under heavy write load,
* and especially during the "sync" phase of a checkpoint. Without this
* logic, each backend begins doing an fsync for every block written, which
* gets very expensive and can slow down the whole system.
*
* Trying to do this every time the queue is full could lose if there
* aren't any removable entries. But that should be vanishingly rare in
* practice: there's one queue entry per shared buffer.
*/
static bool
CompactCheckpointerRequestQueue(void)
{
struct CheckpointerSlotMapping
{
CheckpointerRequest request;
int slot;
};
int n,
preserve_count;
int num_skipped = 0;
HASHCTL ctl;
HTAB *htab;
bool *skip_slot;
/* must hold CheckpointerCommLock in exclusive mode */
Assert(LWLockHeldByMe(CheckpointerCommLock));
/* Initialize skip_slot array */
skip_slot = palloc0(sizeof(bool) * CheckpointerShmem->num_requests);
/* Initialize temporary hash table */
MemSet(&ctl, 0, sizeof(ctl));
ctl.keysize = sizeof(CheckpointerRequest);
ctl.entrysize = sizeof(struct CheckpointerSlotMapping);
ctl.hcxt = CurrentMemoryContext;
htab = hash_create("CompactCheckpointerRequestQueue",
CheckpointerShmem->num_requests,
&ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
/*
* The basic idea here is that a request can be skipped if it's followed
* by a later, identical request. It might seem more sensible to work
* backwards from the end of the queue and check whether a request is
* *preceded* by an earlier, identical request, in the hopes of doing less
* copying. But that might change the semantics, if there's an
* intervening FORGET_RELATION_FSYNC or FORGET_DATABASE_FSYNC request, so
* we do it this way. It would be possible to be even smarter if we made
* the code below understand the specific semantics of such requests (it
* could blow away preceding entries that would end up being canceled
* anyhow), but it's not clear that the extra complexity would buy us
* anything.
*/
for (n = 0; n < CheckpointerShmem->num_requests; n++)
{
CheckpointerRequest *request;
struct CheckpointerSlotMapping *slotmap;
bool found;
/*
* We use the request struct directly as a hashtable key. This
* assumes that any padding bytes in the structs are consistently the
* same, which should be okay because we zeroed them in
* CheckpointerShmemInit. Note also that RelFileNode had better
* contain no pad bytes.
*/
request = &CheckpointerShmem->requests[n];
slotmap = hash_search(htab, request, HASH_ENTER, &found);
if (found)
{
/* Duplicate, so mark the previous occurrence as skippable */
skip_slot[slotmap->slot] = true;
num_skipped++;
}
/* Remember slot containing latest occurrence of this request value */
slotmap->slot = n;
}
/* Done with the hash table. */
hash_destroy(htab);
/* If no duplicates, we're out of luck. */
if (!num_skipped)
{
pfree(skip_slot);
return false;
}
/* We found some duplicates; remove them. */
preserve_count = 0;
for (n = 0; n < CheckpointerShmem->num_requests; n++)
{
if (skip_slot[n])
continue;
CheckpointerShmem->requests[preserve_count++] = CheckpointerShmem->requests[n];
}
ereport(DEBUG1,
(errmsg("compacted fsync request queue from %d entries to %d entries",
CheckpointerShmem->num_requests, preserve_count)));
CheckpointerShmem->num_requests = preserve_count;
/* Cleanup. */
pfree(skip_slot);
return true;
}
/*
* gp_read_error_log
*
* Returns set of error log tuples.
*/
Datum
gp_read_error_log(PG_FUNCTION_ARGS)
{
FuncCallContext *funcctx;
ReadErrorLogContext *context;
HeapTuple tuple;
Datum result;
/*
* First call setup
*/
if (SRF_IS_FIRSTCALL())
{
MemoryContext oldcontext;
FILE *fp;
text *relname;
funcctx = SRF_FIRSTCALL_INIT();
relname = PG_GETARG_TEXT_P(0);
oldcontext = MemoryContextSwitchTo(funcctx->multi_call_memory_ctx);
context = palloc0(sizeof(ReadErrorLogContext));
funcctx->user_fctx = (void *) context;
funcctx->tuple_desc = BlessTupleDesc(GetErrorTupleDesc());
/*
* Though this function is usually executed on segment, we dispatch
* the execution if it happens to be on QD, and combine the results
* into one set.
*/
if (Gp_role == GP_ROLE_DISPATCH)
{
int resultCount = 0;
PGresult **results = NULL;
StringInfoData sql;
StringInfoData errbuf;
int i;
initStringInfo(&sql);
initStringInfo(&errbuf);
/*
* construct SQL
*/
appendStringInfo(&sql,
"SELECT * FROM pg_catalog.gp_read_error_log(%s) ",
quote_literal_internal(text_to_cstring(relname)));
results = cdbdisp_dispatchRMCommand(sql.data, true, &errbuf,
&resultCount);
if (errbuf.len > 0)
elog(ERROR, "%s", errbuf.data);
Assert(resultCount > 0);
for (i = 0; i < resultCount; i++)
{
if (PQresultStatus(results[i]) != PGRES_TUPLES_OK)
elog(ERROR, "unexpected result from segment: %d",
PQresultStatus(results[i]));
context->numTuples += PQntuples(results[i]);
}
pfree(errbuf.data);
pfree(sql.data);
context->segResults = results;
context->numSegResults = resultCount;
}
else
{
/*
* In QE, read the error log.
*/
RangeVar *relrv;
Oid relid;
relrv = makeRangeVarFromNameList(textToQualifiedNameList(relname));
relid = RangeVarGetRelid(relrv, true);
/*
* If the relation has gone, silently return no tuples.
*/
if (OidIsValid(relid))
{
AclResult aclresult;
/*
* Requires SELECT priv to read error log.
*/
aclresult = pg_class_aclcheck(relid, GetUserId(), ACL_SELECT);
if (aclresult != ACLCHECK_OK)
aclcheck_error(aclresult, ACL_KIND_CLASS, relrv->relname);
ErrorLogFileName(context->filename, MyDatabaseId, relid);
fp = AllocateFile(context->filename, "r");
context->fp = fp;
}
}
MemoryContextSwitchTo(oldcontext);
if (Gp_role != GP_ROLE_DISPATCH && !context->fp)
{
pfree(context);
SRF_RETURN_DONE(funcctx);
}
}
funcctx = SRF_PERCALL_SETUP();
context = (ReadErrorLogContext *) funcctx->user_fctx;
/*
* Read error log, probably on segments. We don't check Gp_role, however,
* in case master also wants to read the file.
*/
if (context->fp)
{
pg_crc32 crc, written_crc;
tuple = ErrorLogRead(context->fp, &written_crc);
/*
* CRC check.
*/
if (HeapTupleIsValid(tuple))
{
INIT_CRC32C(crc);
COMP_CRC32C(crc, tuple->t_data, tuple->t_len);
FIN_CRC32C(crc);
if (!EQ_CRC32C(crc, written_crc))
{
elog(LOG, "incorrect checksum in error log %s",
context->filename);
tuple = NULL;
}
}
/*
* If we found a valid tuple, return it. Otherwise, fall through
* in the DONE routine.
*/
if (HeapTupleIsValid(tuple))
{
/*
* We need to set typmod for the executor to understand
* its type we just blessed.
*/
HeapTupleHeaderSetTypMod(tuple->t_data,
funcctx->tuple_desc->tdtypmod);
result = HeapTupleGetDatum(tuple);
SRF_RETURN_NEXT(funcctx, result);
}
}
/*
* If we got results from dispatch, return all the tuples.
*/
while (context->currentResult < context->numSegResults)
{
Datum values[NUM_ERRORTABLE_ATTR];
bool isnull[NUM_ERRORTABLE_ATTR];
PGresult *segres = context->segResults[context->currentResult];
int row = context->currentRow;
if (row >= PQntuples(segres))
{
context->currentRow = 0;
context->currentResult++;
continue;
}
context->currentRow++;
MemSet(isnull, false, sizeof(isnull));
values[0] = ResultToDatum(segres, row, 0, timestamptz_in, &isnull[0]);
values[1] = ResultToDatum(segres, row, 1, textin, &isnull[1]);
values[2] = ResultToDatum(segres, row, 2, textin, &isnull[2]);
values[3] = ResultToDatum(segres, row, 3, int4in, &isnull[3]);
values[4] = ResultToDatum(segres, row, 4, int4in, &isnull[4]);
values[5] = ResultToDatum(segres, row, 5, textin, &isnull[5]);
values[6] = ResultToDatum(segres, row, 6, textin, &isnull[6]);
values[7] = ResultToDatum(segres, row, 7, byteain, &isnull[7]);
tuple = heap_form_tuple(funcctx->tuple_desc, values, isnull);
result = HeapTupleGetDatum(tuple);
SRF_RETURN_NEXT(funcctx, result);
}
if (context->segResults != NULL)
{
int i;
for (i = 0; i < context->numSegResults; i++)
PQclear(context->segResults[i]);
/* XXX: better to copy to palloc'ed area */
free(context->segResults);
}
/*
* Close the file, if we have opened it.
*/
if (context->fp != NULL)
{
FreeFile(context->fp);
context->fp = NULL;
}
SRF_RETURN_DONE(funcctx);
}
Datum
ltree_in(PG_FUNCTION_ARGS)
{
char *buf = (char *) PG_GETARG_POINTER(0);
char *ptr;
nodeitem *list,
*lptr;
int num = 0,
totallen = 0;
int state = LTPRS_WAITNAME;
ltree *result;
ltree_level *curlevel;
int charlen;
int pos = 0;
ptr = buf;
while (*ptr)
{
charlen = pg_mblen(ptr);
if (charlen == 1 && t_iseq(ptr, '.'))
num++;
ptr += charlen;
}
if (num + 1 > MaxAllocSize / sizeof(nodeitem))
ereport(ERROR,
(errcode(ERRCODE_PROGRAM_LIMIT_EXCEEDED),
errmsg("number of levels (%d) exceeds the maximum allowed (%d)",
num + 1, (int) (MaxAllocSize / sizeof(nodeitem)))));
list = lptr = (nodeitem *) palloc(sizeof(nodeitem) * (num + 1));
ptr = buf;
while (*ptr)
{
charlen = pg_mblen(ptr);
if (state == LTPRS_WAITNAME)
{
if (ISALNUM(ptr))
{
lptr->start = ptr;
lptr->wlen = 0;
state = LTPRS_WAITDELIM;
}
else
UNCHAR;
}
else if (state == LTPRS_WAITDELIM)
{
if (charlen == 1 && t_iseq(ptr, '.'))
{
lptr->len = ptr - lptr->start;
if (lptr->wlen > 255)
ereport(ERROR,
(errcode(ERRCODE_NAME_TOO_LONG),
errmsg("name of level is too long"),
errdetail("Name length is %d, must "
"be < 256, in position %d.",
lptr->wlen, pos)));
totallen += MAXALIGN(lptr->len + LEVEL_HDRSIZE);
lptr++;
state = LTPRS_WAITNAME;
}
else if (!ISALNUM(ptr))
UNCHAR;
}
else
/* internal error */
elog(ERROR, "internal error in parser");
ptr += charlen;
lptr->wlen++;
pos++;
}
if (state == LTPRS_WAITDELIM)
{
lptr->len = ptr - lptr->start;
if (lptr->wlen > 255)
ereport(ERROR,
(errcode(ERRCODE_NAME_TOO_LONG),
errmsg("name of level is too long"),
errdetail("Name length is %d, must "
"be < 256, in position %d.",
lptr->wlen, pos)));
totallen += MAXALIGN(lptr->len + LEVEL_HDRSIZE);
lptr++;
}
else if (!(state == LTPRS_WAITNAME && lptr == list))
ereport(ERROR,
(errcode(ERRCODE_SYNTAX_ERROR),
errmsg("syntax error"),
errdetail("Unexpected end of line.")));
result = (ltree *) palloc0(LTREE_HDRSIZE + totallen);
SET_VARSIZE(result, LTREE_HDRSIZE + totallen);
result->numlevel = lptr - list;
curlevel = LTREE_FIRST(result);
lptr = list;
while (lptr - list < result->numlevel)
{
curlevel->len = (uint16) lptr->len;
memcpy(curlevel->name, lptr->start, lptr->len);
curlevel = LEVEL_NEXT(curlevel);
lptr++;
}
pfree(list);
PG_RETURN_POINTER(result);
}
/*
* Collects info about fields of a composite type.
*
* Based on TupleDescGetAttInMetadata.
*/
ProxyComposite *
plproxy_composite_info(ProxyFunction *func, TupleDesc tupdesc)
{
int i,
natts = tupdesc->natts;
ProxyComposite *ret;
MemoryContext old_ctx;
Form_pg_attribute a;
ProxyType *type;
const char *name;
Oid oid = tupdesc->tdtypeid;
old_ctx = MemoryContextSwitchTo(func->ctx);
ret = palloc(sizeof(*ret));
ret->type_list = palloc(sizeof(ProxyType *) * natts);
ret->name_list = palloc0(sizeof(char *) * natts);
ret->tupdesc = BlessTupleDesc(tupdesc);
ret->use_binary = 1;
ret->alterable = 0;
if (oid != RECORDOID)
{
HeapTuple type_tuple;
HeapTuple rel_tuple;
Form_pg_type pg_type;
type_tuple = SearchSysCache(TYPEOID, ObjectIdGetDatum(oid), 0, 0, 0);
if (!HeapTupleIsValid(type_tuple))
elog(ERROR, "cache lookup failed for type %u", oid);
pg_type = (Form_pg_type) GETSTRUCT(type_tuple);
rel_tuple = SearchSysCache(RELOID, ObjectIdGetDatum(pg_type->typrelid), 0, 0, 0);
if (!HeapTupleIsValid(rel_tuple))
elog(ERROR, "cache lookup failed for type relation %u", pg_type->typrelid);
plproxy_set_stamp(&ret->stamp, rel_tuple);
ReleaseSysCache(rel_tuple);
ReleaseSysCache(type_tuple);
ret->alterable = 1;
if (ret->tupdesc->tdtypeid != oid)
elog(ERROR, "lost oid");
}
MemoryContextSwitchTo(old_ctx);
ret->nfields = 0;
for (i = 0; i < natts; i++)
{
a = TupleDescAttr(tupdesc, i);
if (a->attisdropped)
{
ret->name_list[i] = NULL;
ret->type_list[i] = NULL;
continue;
}
ret->nfields++;
name = quote_identifier(NameStr(a->attname));
ret->name_list[i] = plproxy_func_strdup(func, name);
type = plproxy_find_type_info(func, a->atttypid, 0);
ret->type_list[i] = type;
if (!type->has_recv)
ret->use_binary = 0;
}
return ret;
}
/*
* lazy_vacuum_rel() -- perform LAZY VACUUM for one heap relation
*
* This routine vacuums a single heap, cleans out its indexes, and
* updates its relpages and reltuples statistics.
*
* At entry, we have already established a transaction and opened
* and locked the relation.
*/
void
lazy_vacuum_rel(Relation onerel, VacuumStmt *vacstmt, List *updated_stats)
{
LVRelStats *vacrelstats;
Relation *Irel;
int nindexes;
BlockNumber possibly_freeable;
if (vacstmt->verbose)
elevel = INFO;
else
elevel = DEBUG2;
if (Gp_role == GP_ROLE_DISPATCH)
elevel = DEBUG2; /* vacuum and analyze messages aren't interesting from the QD */
#ifdef FAULT_INJECTOR
if (vacuumStatement_IsInAppendOnlyDropPhase(vacstmt))
{
FaultInjector_InjectFaultIfSet(
CompactionBeforeSegmentFileDropPhase,
DDLNotSpecified,
"", // databaseName
""); // tableName
}
if (vacummStatement_IsInAppendOnlyCleanupPhase(vacstmt))
{
FaultInjector_InjectFaultIfSet(
CompactionBeforeCleanupPhase,
DDLNotSpecified,
"", // databaseName
""); // tableName
}
#endif
/*
* MPP-23647. Update xid limits for heap as well as appendonly
* relations. This allows setting relfrozenxid to correct value
* for an appendonly (AO/CO) table.
*/
vacuum_set_xid_limits(vacstmt, onerel->rd_rel->relisshared,
&OldestXmin, &FreezeLimit);
/*
* Execute the various vacuum operations. Appendonly tables are treated
* differently.
*/
if (RelationIsAoRows(onerel) || RelationIsAoCols(onerel))
{
lazy_vacuum_aorel(onerel, vacstmt, updated_stats);
return;
}
vacrelstats = (LVRelStats *) palloc0(sizeof(LVRelStats));
/* heap relation */
/* Set threshold for interesting free space = average request size */
/* XXX should we scale it up or down? Adjust vacuum.c too, if so */
vacrelstats->threshold = GetAvgFSMRequestSize(&onerel->rd_node);
/* Open all indexes of the relation */
vac_open_indexes(onerel, RowExclusiveLock, &nindexes, &Irel);
vacrelstats->hasindex = (nindexes > 0);
/* Do the vacuuming */
lazy_scan_heap(onerel, vacrelstats, Irel, nindexes, updated_stats, vacstmt->extra_oids);
/* Done with indexes */
vac_close_indexes(nindexes, Irel, NoLock);
/*
* Optionally truncate the relation.
*
* Don't even think about it unless we have a shot at releasing a goodly
* number of pages. Otherwise, the time taken isn't worth it.
*/
possibly_freeable = vacrelstats->rel_pages - vacrelstats->nonempty_pages;
if (possibly_freeable >= REL_TRUNCATE_MINIMUM ||
possibly_freeable >= vacrelstats->rel_pages / REL_TRUNCATE_FRACTION)
lazy_truncate_heap(onerel, vacrelstats);
/* Update shared free space map with final free space info */
lazy_update_fsm(onerel, vacrelstats);
/* Update statistics in pg_class */
vac_update_relstats(onerel,
vacrelstats->rel_pages,
vacrelstats->rel_tuples,
vacrelstats->hasindex,
FreezeLimit,
updated_stats);
/* report results to the stats collector, too */
pgstat_report_vacuum(RelationGetRelid(onerel), onerel->rd_rel->relisshared,
true /*vacrelstats->scanned_all*/,
vacstmt->analyze, vacrelstats->rel_tuples);
}
/* ----------------------------------------------------------------
* ExecInitGather
* ----------------------------------------------------------------
*/
GatherMergeState *
ExecInitGatherMerge(GatherMerge *node, EState *estate, int eflags)
{
GatherMergeState *gm_state;
Plan *outerNode;
bool hasoid;
TupleDesc tupDesc;
/* Gather merge node doesn't have innerPlan node. */
Assert(innerPlan(node) == NULL);
/*
* create state structure
*/
gm_state = makeNode(GatherMergeState);
gm_state->ps.plan = (Plan *) node;
gm_state->ps.state = estate;
gm_state->ps.ExecProcNode = ExecGatherMerge;
/*
* Miscellaneous initialization
*
* create expression context for node
*/
ExecAssignExprContext(estate, &gm_state->ps);
/*
* initialize child expressions
*/
gm_state->ps.qual =
ExecInitQual(node->plan.qual, &gm_state->ps);
/*
* tuple table initialization
*/
ExecInitResultTupleSlot(estate, &gm_state->ps);
/*
* now initialize outer plan
*/
outerNode = outerPlan(node);
outerPlanState(gm_state) = ExecInitNode(outerNode, estate, eflags);
/*
* Initialize result tuple type and projection info.
*/
ExecAssignResultTypeFromTL(&gm_state->ps);
ExecAssignProjectionInfo(&gm_state->ps, NULL);
gm_state->gm_initialized = false;
/*
* initialize sort-key information
*/
if (node->numCols)
{
int i;
gm_state->gm_nkeys = node->numCols;
gm_state->gm_sortkeys =
palloc0(sizeof(SortSupportData) * node->numCols);
for (i = 0; i < node->numCols; i++)
{
SortSupport sortKey = gm_state->gm_sortkeys + i;
sortKey->ssup_cxt = CurrentMemoryContext;
sortKey->ssup_collation = node->collations[i];
sortKey->ssup_nulls_first = node->nullsFirst[i];
sortKey->ssup_attno = node->sortColIdx[i];
/*
* We don't perform abbreviated key conversion here, for the same
* reasons that it isn't used in MergeAppend
*/
sortKey->abbreviate = false;
PrepareSortSupportFromOrderingOp(node->sortOperators[i], sortKey);
}
}
/*
* store the tuple descriptor into gather merge state, so we can use it
* later while initializing the gather merge slots.
*/
if (!ExecContextForcesOids(&gm_state->ps, &hasoid))
hasoid = false;
tupDesc = ExecTypeFromTL(outerNode->targetlist, hasoid);
gm_state->tupDesc = tupDesc;
return gm_state;
}