/*
* RouterExecutorStart sets up the executor state and queryDesc for router
* execution.
*/
void
RouterExecutorStart(QueryDesc *queryDesc, int eflags, Task *task)
{
LOCKMODE lockMode = NoLock;
EState *executorState = NULL;
CmdType commandType = queryDesc->operation;
/* ensure that the task is not NULL */
Assert(task != NULL);
/* disallow transactions and triggers during distributed modify commands */
if (commandType != CMD_SELECT)
{
bool topLevel = true;
PreventTransactionChain(topLevel, "distributed commands");
eflags |= EXEC_FLAG_SKIP_TRIGGERS;
}
/* signal that it is a router execution */
eflags |= EXEC_FLAG_CITUS_ROUTER_EXECUTOR;
/* build empty executor state to obtain per-query memory context */
executorState = CreateExecutorState();
executorState->es_top_eflags = eflags;
executorState->es_instrument = queryDesc->instrument_options;
queryDesc->estate = executorState;
/*
* As it's similar to what we're doing, use a MaterialState node to store
* our state. This is used to store our tuplestore, so cursors etc. can
* work.
*/
queryDesc->planstate = (PlanState *) makeNode(MaterialState);
#if (PG_VERSION_NUM < 90500)
/* make sure that upsertQuery is false for versions that UPSERT is not available */
Assert(task->upsertQuery == false);
#endif
lockMode = CommutativityRuleToLockMode(commandType, task->upsertQuery);
if (lockMode != NoLock)
{
AcquireExecutorShardLock(task, lockMode);
}
}
/*
* PgPaxosExecutorStart blocks until the table is ready to read.
*/
static void
PgPaxosExecutorStart(QueryDesc *queryDesc, int eflags)
{
PlannedStmt *plannedStmt = queryDesc->plannedstmt;
List *rangeTableList = plannedStmt->rtable;
CmdType commandType = queryDesc->operation;
if (IsPgPaxosActive() && HasPaxosTable(rangeTableList))
{
char *sqlQuery = (char *) queryDesc->sourceText;
char *groupId = NULL;
bool topLevel = true;
/* disallow transactions during paxos commands */
PreventTransactionChain(topLevel, "paxos commands");
groupId = DeterminePaxosGroup(rangeTableList);
if (commandType == CMD_INSERT || commandType == CMD_UPDATE ||
commandType == CMD_DELETE)
{
PrepareConsistentWrite(groupId, sqlQuery);
}
else
{
PrepareConsistentRead(groupId);
}
queryDesc->snapshot->curcid = GetCurrentCommandId(false);
}
/* call into the standard executor start, or hook if set */
if (PreviousExecutorStartHook != NULL)
{
PreviousExecutorStartHook(queryDesc, eflags);
}
else
{
standard_ExecutorStart(queryDesc, eflags);
}
}
static void
DiscardAll(bool isTopLevel)
{
/*
* Disallow DISCARD ALL in a transaction block. This is arguably
* inconsistent (we don't make a similar check in the command sequence
* that DISCARD ALL is equivalent to), but the idea is to catch mistakes:
* DISCARD ALL inside a transaction block would leave the transaction
* still uncommitted.
*/
PreventTransactionChain(isTopLevel, "DISCARD ALL");
SetPGVariable("session_authorization", NIL, false);
ResetAllOptions();
DropAllPreparedStatements();
PortalHashTableDeleteAll();
Async_UnlistenAll();
LockReleaseAll(USER_LOCKMETHOD, true);
ResetPlanCache();
ResetTempTableNamespace();
}
/*
* Drop a table space
*
* Be careful to check that the tablespace is empty.
*/
void
DropTableSpace(DropTableSpaceStmt *stmt)
{
#ifdef HAVE_SYMLINK
char *tablespacename = stmt->tablespacename;
HeapScanDesc scandesc;
Relation rel;
HeapTuple tuple;
ScanKeyData entry[1];
Oid tablespaceoid;
/* don't call this in a transaction block */
PreventTransactionChain((void *) stmt, "DROP TABLESPACE");
/*
* Acquire ExclusiveLock on pg_tablespace to ensure that no one else is
* trying to do DROP TABLESPACE or TablespaceCreateDbspace concurrently.
*/
rel = heap_open(TableSpaceRelationId, ExclusiveLock);
/*
* Find the target tuple
*/
ScanKeyInit(&entry[0],
Anum_pg_tablespace_spcname,
BTEqualStrategyNumber, F_NAMEEQ,
CStringGetDatum(tablespacename));
scandesc = heap_beginscan(rel, SnapshotNow, 1, entry);
tuple = heap_getnext(scandesc, ForwardScanDirection);
if (!HeapTupleIsValid(tuple))
ereport(ERROR,
(errcode(ERRCODE_UNDEFINED_OBJECT),
errmsg("tablespace \"%s\" does not exist",
tablespacename)));
tablespaceoid = HeapTupleGetOid(tuple);
/* Must be tablespace owner */
if (!pg_tablespace_ownercheck(tablespaceoid, GetUserId()))
aclcheck_error(ACLCHECK_NOT_OWNER, ACL_KIND_TABLESPACE,
tablespacename);
/* Disallow drop of the standard tablespaces, even by superuser */
if (tablespaceoid == GLOBALTABLESPACE_OID ||
tablespaceoid == DEFAULTTABLESPACE_OID)
aclcheck_error(ACLCHECK_NO_PRIV, ACL_KIND_TABLESPACE,
tablespacename);
/*
* Remove the pg_tablespace tuple (this will roll back if we fail below)
*/
simple_heap_delete(rel, &tuple->t_self);
heap_endscan(scandesc);
/*
* Remove dependency on owner.
*/
deleteSharedDependencyRecordsFor(TableSpaceRelationId, tablespaceoid);
/*
* Try to remove the physical infrastructure
*/
if (!remove_tablespace_directories(tablespaceoid, false))
ereport(ERROR,
(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
errmsg("tablespace \"%s\" is not empty",
tablespacename)));
/* Record the filesystem change in XLOG */
{
xl_tblspc_drop_rec xlrec;
XLogRecData rdata[1];
xlrec.ts_id = tablespaceoid;
rdata[0].data = (char *) &xlrec;
rdata[0].len = sizeof(xl_tblspc_drop_rec);
rdata[0].buffer = InvalidBuffer;
rdata[0].next = NULL;
(void) XLogInsert(RM_TBLSPC_ID, XLOG_TBLSPC_DROP, rdata);
}
/* We keep the lock on pg_tablespace until commit */
heap_close(rel, NoLock);
#else /* !HAVE_SYMLINK */
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("tablespaces are not supported on this platform")));
#endif /* HAVE_SYMLINK */
}
/*
* Create a table space
*
* Only superusers can create a tablespace. This seems a reasonable restriction
* since we're determining the system layout and, anyway, we probably have
* root if we're doing this kind of activity
*/
void
CreateTableSpace(CreateTableSpaceStmt *stmt)
{
#ifdef HAVE_SYMLINK
Relation rel;
Datum values[Natts_pg_tablespace];
char nulls[Natts_pg_tablespace];
HeapTuple tuple;
Oid tablespaceoid;
char *location;
char *linkloc;
Oid ownerId;
/* validate */
/* don't call this in a transaction block */
PreventTransactionChain((void *) stmt, "CREATE TABLESPACE");
/* Must be super user */
if (!superuser())
ereport(ERROR,
(errcode(ERRCODE_INSUFFICIENT_PRIVILEGE),
errmsg("permission denied to create tablespace \"%s\"",
stmt->tablespacename),
errhint("Must be superuser to create a tablespace.")));
/* However, the eventual owner of the tablespace need not be */
if (stmt->owner)
ownerId = get_roleid_checked(stmt->owner);
else
ownerId = GetUserId();
/* Unix-ify the offered path, and strip any trailing slashes */
location = pstrdup(stmt->location);
canonicalize_path(location);
/* disallow quotes, else CREATE DATABASE would be at risk */
if (strchr(location, '\''))
ereport(ERROR,
(errcode(ERRCODE_INVALID_NAME),
errmsg("tablespace location may not contain single quotes")));
/*
* Allowing relative paths seems risky
*
* this also helps us ensure that location is not empty or whitespace
*/
if (!is_absolute_path(location))
ereport(ERROR,
(errcode(ERRCODE_INVALID_OBJECT_DEFINITION),
errmsg("tablespace location must be an absolute path")));
/*
* Check that location isn't too long. Remember that we're going to append
* '/<dboid>/<relid>.<nnn>' (XXX but do we ever form the whole path
* explicitly? This may be overly conservative.)
*/
if (strlen(location) >= (MAXPGPATH - 1 - 10 - 1 - 10 - 1 - 10))
ereport(ERROR,
(errcode(ERRCODE_INVALID_OBJECT_DEFINITION),
errmsg("tablespace location \"%s\" is too long",
location)));
/*
* Disallow creation of tablespaces named "pg_xxx"; we reserve this
* namespace for system purposes.
*/
if (!allowSystemTableMods && IsReservedName(stmt->tablespacename))
ereport(ERROR,
(errcode(ERRCODE_RESERVED_NAME),
errmsg("unacceptable tablespace name \"%s\"",
stmt->tablespacename),
errdetail("The prefix \"pg_\" is reserved for system tablespaces.")));
/*
* Check that there is no other tablespace by this name. (The unique
* index would catch this anyway, but might as well give a friendlier
* message.)
*/
if (OidIsValid(get_tablespace_oid(stmt->tablespacename)))
ereport(ERROR,
(errcode(ERRCODE_DUPLICATE_OBJECT),
errmsg("tablespace \"%s\" already exists",
stmt->tablespacename)));
/*
* Insert tuple into pg_tablespace. The purpose of doing this first is to
* lock the proposed tablename against other would-be creators. The
* insertion will roll back if we find problems below.
*/
rel = heap_open(TableSpaceRelationId, RowExclusiveLock);
MemSet(nulls, ' ', Natts_pg_tablespace);
values[Anum_pg_tablespace_spcname - 1] =
DirectFunctionCall1(namein, CStringGetDatum(stmt->tablespacename));
values[Anum_pg_tablespace_spcowner - 1] =
ObjectIdGetDatum(ownerId);
values[Anum_pg_tablespace_spclocation - 1] =
DirectFunctionCall1(textin, CStringGetDatum(location));
nulls[Anum_pg_tablespace_spcacl - 1] = 'n';
tuple = heap_formtuple(rel->rd_att, values, nulls);
tablespaceoid = simple_heap_insert(rel, tuple);
CatalogUpdateIndexes(rel, tuple);
heap_freetuple(tuple);
/* Record dependency on owner */
recordDependencyOnOwner(TableSpaceRelationId, tablespaceoid, ownerId);
/*
* Attempt to coerce target directory to safe permissions. If this fails,
* it doesn't exist or has the wrong owner.
*/
if (chmod(location, 0700) != 0)
ereport(ERROR,
(errcode_for_file_access(),
errmsg("could not set permissions on directory \"%s\": %m",
location)));
/*
* Check the target directory is empty.
*/
if (!directory_is_empty(location))
ereport(ERROR,
(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
errmsg("directory \"%s\" is not empty",
location)));
/*
* Create the PG_VERSION file in the target directory. This has several
* purposes: to make sure we can write in the directory, to prevent
* someone from creating another tablespace pointing at the same directory
* (the emptiness check above will fail), and to label tablespace
* directories by PG version.
*/
set_short_version(location);
/*
* All seems well, create the symlink
*/
linkloc = (char *) palloc(10 + 10 + 1);
sprintf(linkloc, "pg_tblspc/%u", tablespaceoid);
if (symlink(location, linkloc) < 0)
ereport(ERROR,
(errcode_for_file_access(),
errmsg("could not create symbolic link \"%s\": %m",
linkloc)));
/* Record the filesystem change in XLOG */
{
xl_tblspc_create_rec xlrec;
XLogRecData rdata[2];
xlrec.ts_id = tablespaceoid;
rdata[0].data = (char *) &xlrec;
rdata[0].len = offsetof(xl_tblspc_create_rec, ts_path);
rdata[0].buffer = InvalidBuffer;
rdata[0].next = &(rdata[1]);
rdata[1].data = (char *) location;
rdata[1].len = strlen(location) + 1;
rdata[1].buffer = InvalidBuffer;
rdata[1].next = NULL;
(void) XLogInsert(RM_TBLSPC_ID, XLOG_TBLSPC_CREATE, rdata);
}
pfree(linkloc);
pfree(location);
/* We keep the lock on pg_tablespace until commit */
heap_close(rel, NoLock);
#else /* !HAVE_SYMLINK */
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("tablespaces are not supported on this platform")));
#endif /* HAVE_SYMLINK */
}
/*
* Primary entry point for VACUUM and ANALYZE commands.
*
* relid is normally InvalidOid; if it is not, then it provides the relation
* OID to be processed, and vacstmt->relation is ignored. (The non-invalid
* case is currently only used by autovacuum.)
*
* do_toast is passed as FALSE by autovacuum, because it processes TOAST
* tables separately.
*
* for_wraparound is used by autovacuum to let us know when it's forcing
* a vacuum for wraparound, which should not be auto-cancelled.
*
* bstrategy is normally given as NULL, but in autovacuum it can be passed
* in to use the same buffer strategy object across multiple vacuum() calls.
*
* isTopLevel should be passed down from ProcessUtility.
*
* It is the caller's responsibility that vacstmt and bstrategy
* (if given) be allocated in a memory context that won't disappear
* at transaction commit.
*/
void
vacuum(VacuumStmt *vacstmt, Oid relid, bool do_toast,
BufferAccessStrategy bstrategy, bool for_wraparound, bool isTopLevel)
{
const char *stmttype;
volatile bool all_rels,
in_outer_xact,
use_own_xacts;
List *relations;
/* sanity checks on options */
Assert(vacstmt->options & (VACOPT_VACUUM | VACOPT_ANALYZE));
Assert((vacstmt->options & VACOPT_VACUUM) ||
!(vacstmt->options & (VACOPT_FULL | VACOPT_FREEZE)));
Assert((vacstmt->options & VACOPT_ANALYZE) || vacstmt->va_cols == NIL);
stmttype = (vacstmt->options & VACOPT_VACUUM) ? "VACUUM" : "ANALYZE";
/*
* We cannot run VACUUM inside a user transaction block; if we were inside
* a transaction, then our commit- and start-transaction-command calls
* would not have the intended effect! There are numerous other subtle
* dependencies on this, too.
*
* ANALYZE (without VACUUM) can run either way.
*/
if (vacstmt->options & VACOPT_VACUUM)
{
PreventTransactionChain(isTopLevel, stmttype);
in_outer_xact = false;
}
else
in_outer_xact = IsInTransactionChain(isTopLevel);
/*
* Send info about dead objects to the statistics collector, unless we are
* in autovacuum --- autovacuum.c does this for itself.
*/
if ((vacstmt->options & VACOPT_VACUUM) && !IsAutoVacuumWorkerProcess())
pgstat_vacuum_stat();
/*
* Create special memory context for cross-transaction storage.
*
* Since it is a child of PortalContext, it will go away eventually even
* if we suffer an error; there's no need for special abort cleanup logic.
*/
vac_context = AllocSetContextCreate(PortalContext,
"Vacuum",
ALLOCSET_DEFAULT_MINSIZE,
ALLOCSET_DEFAULT_INITSIZE,
ALLOCSET_DEFAULT_MAXSIZE);
/*
* If caller didn't give us a buffer strategy object, make one in the
* cross-transaction memory context.
*/
if (bstrategy == NULL)
{
MemoryContext old_context = MemoryContextSwitchTo(vac_context);
bstrategy = GetAccessStrategy(BAS_VACUUM);
MemoryContextSwitchTo(old_context);
}
vac_strategy = bstrategy;
/* Remember whether we are processing everything in the DB */
all_rels = (!OidIsValid(relid) && vacstmt->relation == NULL);
/*
* Build list of relations to process, unless caller gave us one. (If we
* build one, we put it in vac_context for safekeeping.)
*/
relations = get_rel_oids(relid, vacstmt->relation);
/*
* Decide whether we need to start/commit our own transactions.
*
* For VACUUM (with or without ANALYZE): always do so, so that we can
* release locks as soon as possible. (We could possibly use the outer
* transaction for a one-table VACUUM, but handling TOAST tables would be
* problematic.)
*
* For ANALYZE (no VACUUM): if inside a transaction block, we cannot
* start/commit our own transactions. Also, there's no need to do so if
* only processing one relation. For multiple relations when not within a
* transaction block, and also in an autovacuum worker, use own
* transactions so we can release locks sooner.
*/
if (vacstmt->options & VACOPT_VACUUM)
use_own_xacts = true;
else
{
Assert(vacstmt->options & VACOPT_ANALYZE);
if (IsAutoVacuumWorkerProcess())
use_own_xacts = true;
else if (in_outer_xact)
use_own_xacts = false;
else if (list_length(relations) > 1)
use_own_xacts = true;
else
use_own_xacts = false;
}
/*
* vacuum_rel expects to be entered with no transaction active; it will
* start and commit its own transaction. But we are called by an SQL
* command, and so we are executing inside a transaction already. We
* commit the transaction started in PostgresMain() here, and start
* another one before exiting to match the commit waiting for us back in
* PostgresMain().
*/
if (use_own_xacts)
{
/* ActiveSnapshot is not set by autovacuum */
if (ActiveSnapshotSet())
PopActiveSnapshot();
/* matches the StartTransaction in PostgresMain() */
CommitTransactionCommand();
}
/* Turn vacuum cost accounting on or off */
PG_TRY();
{
ListCell *cur;
VacuumCostActive = (VacuumCostDelay > 0);
VacuumCostBalance = 0;
/*
* Loop to process each selected relation.
*/
foreach(cur, relations)
{
Oid relid = lfirst_oid(cur);
bool scanned_all = false;
if (vacstmt->options & VACOPT_VACUUM)
vacuum_rel(relid, vacstmt, do_toast, for_wraparound,
&scanned_all);
if (vacstmt->options & VACOPT_ANALYZE)
{
/*
* If using separate xacts, start one for analyze. Otherwise,
* we can use the outer transaction.
*/
if (use_own_xacts)
{
StartTransactionCommand();
/* functions in indexes may want a snapshot set */
PushActiveSnapshot(GetTransactionSnapshot());
}
analyze_rel(relid, vacstmt, vac_strategy, !scanned_all);
if (use_own_xacts)
{
PopActiveSnapshot();
CommitTransactionCommand();
}
}
}
}
/*
* Primary entry point for VACUUM and ANALYZE commands.
*
* options is a bitmask of VacuumOption flags, indicating what to do.
*
* relid, if not InvalidOid, indicate the relation to process; otherwise,
* the RangeVar is used. (The latter must always be passed, because it's
* used for error messages.)
*
* params contains a set of parameters that can be used to customize the
* behavior.
*
* va_cols is a list of columns to analyze, or NIL to process them all.
*
* bstrategy is normally given as NULL, but in autovacuum it can be passed
* in to use the same buffer strategy object across multiple vacuum() calls.
*
* isTopLevel should be passed down from ProcessUtility.
*
* It is the caller's responsibility that all parameters are allocated in a
* memory context that will not disappear at transaction commit.
*/
void
vacuum(int options, RangeVar *relation, Oid relid, VacuumParams *params,
List *va_cols, BufferAccessStrategy bstrategy, bool isTopLevel)
{
const char *stmttype;
volatile bool in_outer_xact,
use_own_xacts;
List *relations;
static bool in_vacuum = false;
Assert(params != NULL);
stmttype = (options & VACOPT_VACUUM) ? "VACUUM" : "ANALYZE";
/*
* We cannot run VACUUM inside a user transaction block; if we were inside
* a transaction, then our commit- and start-transaction-command calls
* would not have the intended effect! There are numerous other subtle
* dependencies on this, too.
*
* ANALYZE (without VACUUM) can run either way.
*/
if (options & VACOPT_VACUUM)
{
PreventTransactionChain(isTopLevel, stmttype);
in_outer_xact = false;
}
else
in_outer_xact = IsInTransactionChain(isTopLevel);
/*
* Due to static variables vac_context, anl_context and vac_strategy,
* vacuum() is not reentrant. This matters when VACUUM FULL or ANALYZE
* calls a hostile index expression that itself calls ANALYZE.
*/
if (in_vacuum)
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("%s cannot be executed from VACUUM or ANALYZE",
stmttype)));
/*
* Sanity check DISABLE_PAGE_SKIPPING option.
*/
if ((options & VACOPT_FULL) != 0 &&
(options & VACOPT_DISABLE_PAGE_SKIPPING) != 0)
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("VACUUM option DISABLE_PAGE_SKIPPING cannot be used with FULL")));
/*
* Send info about dead objects to the statistics collector, unless we are
* in autovacuum --- autovacuum.c does this for itself.
*/
if ((options & VACOPT_VACUUM) && !IsAutoVacuumWorkerProcess())
pgstat_vacuum_stat();
/*
* Create special memory context for cross-transaction storage.
*
* Since it is a child of PortalContext, it will go away eventually even
* if we suffer an error; there's no need for special abort cleanup logic.
*/
vac_context = AllocSetContextCreate(PortalContext,
"Vacuum",
ALLOCSET_DEFAULT_SIZES);
/*
* If caller didn't give us a buffer strategy object, make one in the
* cross-transaction memory context.
*/
if (bstrategy == NULL)
{
MemoryContext old_context = MemoryContextSwitchTo(vac_context);
bstrategy = GetAccessStrategy(BAS_VACUUM);
MemoryContextSwitchTo(old_context);
}
vac_strategy = bstrategy;
/*
* Build list of relations to process, unless caller gave us one. (If we
* build one, we put it in vac_context for safekeeping.)
*/
relations = get_rel_oids(relid, relation);
/*
* Decide whether we need to start/commit our own transactions.
*
* For VACUUM (with or without ANALYZE): always do so, so that we can
* release locks as soon as possible. (We could possibly use the outer
* transaction for a one-table VACUUM, but handling TOAST tables would be
* problematic.)
*
* For ANALYZE (no VACUUM): if inside a transaction block, we cannot
* start/commit our own transactions. Also, there's no need to do so if
* only processing one relation. For multiple relations when not within a
* transaction block, and also in an autovacuum worker, use own
* transactions so we can release locks sooner.
*/
if (options & VACOPT_VACUUM)
use_own_xacts = true;
else
{
Assert(options & VACOPT_ANALYZE);
if (IsAutoVacuumWorkerProcess())
use_own_xacts = true;
else if (in_outer_xact)
use_own_xacts = false;
else if (list_length(relations) > 1)
use_own_xacts = true;
else
use_own_xacts = false;
}
/*
* vacuum_rel expects to be entered with no transaction active; it will
* start and commit its own transaction. But we are called by an SQL
* command, and so we are executing inside a transaction already. We
* commit the transaction started in PostgresMain() here, and start
* another one before exiting to match the commit waiting for us back in
* PostgresMain().
*/
if (use_own_xacts)
{
Assert(!in_outer_xact);
/* ActiveSnapshot is not set by autovacuum */
if (ActiveSnapshotSet())
PopActiveSnapshot();
/* matches the StartTransaction in PostgresMain() */
CommitTransactionCommand();
}
/* Turn vacuum cost accounting on or off */
PG_TRY();
{
ListCell *cur;
in_vacuum = true;
VacuumCostActive = (VacuumCostDelay > 0);
VacuumCostBalance = 0;
VacuumPageHit = 0;
VacuumPageMiss = 0;
VacuumPageDirty = 0;
/*
* Loop to process each selected relation.
*/
foreach(cur, relations)
{
Oid relid = lfirst_oid(cur);
if (options & VACOPT_VACUUM)
{
if (!vacuum_rel(relid, relation, options, params))
continue;
}
if (options & VACOPT_ANALYZE)
{
/*
* If using separate xacts, start one for analyze. Otherwise,
* we can use the outer transaction.
*/
if (use_own_xacts)
{
StartTransactionCommand();
/* functions in indexes may want a snapshot set */
PushActiveSnapshot(GetTransactionSnapshot());
}
analyze_rel(relid, relation, options, params,
va_cols, in_outer_xact, vac_strategy);
if (use_own_xacts)
{
PopActiveSnapshot();
CommitTransactionCommand();
}
}
}
}