/* Allocate and initialize walsender-related shared memory */
void
WalSndShmemInit(void)
{
bool found;
int i;
WalSndCtl = (WalSndCtlData *)
ShmemInitStruct("Wal Sender Ctl", WalSndShmemSize(), &found);
if (!found)
{
/* First time through, so initialize */
MemSet(WalSndCtl, 0, WalSndShmemSize());
SHMQueueInit(&(WalSndCtl->SyncRepQueue));
for (i = 0; i < max_wal_senders; i++)
{
WalSnd *walsnd = &WalSndCtl->walsnds[i];
SpinLockInit(&walsnd->mutex);
InitSharedLatch(&walsnd->latch);
}
}
}
/*
* ProcQueueInit -- initialize a shared memory process queue
*/
void
ProcQueueInit(PROC_QUEUE *queue)
{
SHMQueueInit(&(queue->links));
queue->size = 0;
}
/*
* InitProcGlobal -
* Initialize the global process table during postmaster or standalone
* backend startup.
*
* We also create all the per-process semaphores we will need to support
* the requested number of backends. We used to allocate semaphores
* only when backends were actually started up, but that is bad because
* it lets Postgres fail under load --- a lot of Unix systems are
* (mis)configured with small limits on the number of semaphores, and
* running out when trying to start another backend is a common failure.
* So, now we grab enough semaphores to support the desired max number
* of backends immediately at initialization --- if the sysadmin has set
* MaxConnections, max_worker_processes, or autovacuum_max_workers higher
* than his kernel will support, he'll find out sooner rather than later.
*
* Another reason for creating semaphores here is that the semaphore
* implementation typically requires us to create semaphores in the
* postmaster, not in backends.
*
* Note: this is NOT called by individual backends under a postmaster,
* not even in the EXEC_BACKEND case. The ProcGlobal and AuxiliaryProcs
* pointers must be propagated specially for EXEC_BACKEND operation.
*/
void
InitProcGlobal(void)
{
PGPROC *procs;
PGXACT *pgxacts;
int i,
j;
bool found;
uint32 TotalProcs = MaxBackends + NUM_AUXILIARY_PROCS + max_prepared_xacts;
/* Create the ProcGlobal shared structure */
ProcGlobal = (PROC_HDR *)
ShmemInitStruct("Proc Header", sizeof(PROC_HDR), &found);
Assert(!found);
/*
* Initialize the data structures.
*/
ProcGlobal->spins_per_delay = DEFAULT_SPINS_PER_DELAY;
ProcGlobal->freeProcs = NULL;
ProcGlobal->autovacFreeProcs = NULL;
ProcGlobal->bgworkerFreeProcs = NULL;
ProcGlobal->startupProc = NULL;
ProcGlobal->startupProcPid = 0;
ProcGlobal->startupBufferPinWaitBufId = -1;
ProcGlobal->walwriterLatch = NULL;
ProcGlobal->checkpointerLatch = NULL;
/*
* Create and initialize all the PGPROC structures we'll need. There are
* five separate consumers: (1) normal backends, (2) autovacuum workers
* and the autovacuum launcher, (3) background workers, (4) auxiliary
* processes, and (5) prepared transactions. Each PGPROC structure is
* dedicated to exactly one of these purposes, and they do not move
* between groups.
*/
procs = (PGPROC *) ShmemAlloc(TotalProcs * sizeof(PGPROC));
ProcGlobal->allProcs = procs;
/* XXX allProcCount isn't really all of them; it excludes prepared xacts */
ProcGlobal->allProcCount = MaxBackends + NUM_AUXILIARY_PROCS;
elog(DEBUG3, "InitProcGlobal of size %d :: TID : %d", TotalProcs, GetBackendThreadId());
if (!procs)
ereport(FATAL,
(errcode(ERRCODE_OUT_OF_MEMORY),
errmsg("out of shared memory")));
MemSet(procs, 0, TotalProcs * sizeof(PGPROC));
/*
* Also allocate a separate array of PGXACT structures. This is separate
* from the main PGPROC array so that the most heavily accessed data is
* stored contiguously in memory in as few cache lines as possible. This
* provides significant performance benefits, especially on a
* multiprocessor system. There is one PGXACT structure for every PGPROC
* structure.
*/
pgxacts = (PGXACT *) ShmemAlloc(TotalProcs * sizeof(PGXACT));
MemSet(pgxacts, 0, TotalProcs * sizeof(PGXACT));
ProcGlobal->allPgXact = pgxacts;
for (i = 0; i < TotalProcs; i++)
{
/* Common initialization for all PGPROCs, regardless of type. */
/*
* Set up per-PGPROC semaphore, latch, and backendLock. Prepared xact
* dummy PGPROCs don't need these though - they're never associated
* with a real process
*/
if (i < MaxBackends + NUM_AUXILIARY_PROCS)
{
PGSemaphoreCreate(&(procs[i].sem));
InitSharedLatch(&(procs[i].procLatch));
procs[i].backendLock = LWLockAssign();
}
procs[i].pgprocno = i;
/*
* Newly created PGPROCs for normal backends, autovacuum and bgworkers
* must be queued up on the appropriate free list. Because there can
* only ever be a small, fixed number of auxiliary processes, no free
* list is used in that case; InitAuxiliaryProcess() instead uses a
* linear search. PGPROCs for prepared transactions are added to a
* free list by TwoPhaseShmemInit().
*/
if (i < MaxConnections)
{
/* PGPROC for normal backend, add to freeProcs list */
procs[i].links.next = (SHM_QUEUE *) ProcGlobal->freeProcs;
ProcGlobal->freeProcs = &procs[i];
//elog(DEBUG3, "freeProcs %d = %p", i, ProcGlobal->freeProcs);
//Assert(ShmemAddrIsValid(ProcGlobal->freeProcs));
}
else if (i < MaxConnections + autovacuum_max_workers + 1)
{
/* PGPROC for AV launcher/worker, add to autovacFreeProcs list */
procs[i].links.next = (SHM_QUEUE *) ProcGlobal->autovacFreeProcs;
ProcGlobal->autovacFreeProcs = &procs[i];
}
else if (i < MaxBackends)
{
/* PGPROC for bgworker, add to bgworkerFreeProcs list */
procs[i].links.next = (SHM_QUEUE *) ProcGlobal->bgworkerFreeProcs;
ProcGlobal->bgworkerFreeProcs = &procs[i];
}
/* Initialize myProcLocks[] shared memory queues. */
for (j = 0; j < NUM_LOCK_PARTITIONS; j++)
SHMQueueInit(&(procs[i].myProcLocks[j]));
}
/*
* Save pointers to the blocks of PGPROC structures reserved for auxiliary
* processes and prepared transactions.
*/
AuxiliaryProcs = &procs[MaxBackends];
PreparedXactProcs = &procs[MaxBackends + NUM_AUXILIARY_PROCS];
/* Create ProcStructLock spinlock, too */
ProcStructLock = (slock_t *) ShmemAlloc(sizeof(slock_t));
SpinLockInit(ProcStructLock);
}
/*
* MarkAsPreparing
* Reserve the GID for the given transaction.
*
* Internally, this creates a gxact struct and puts it into the active array.
* NOTE: this is also used when reloading a gxact after a crash; so avoid
* assuming that we can use very much backend context.
*/
GlobalTransaction
MarkAsPreparing(TransactionId xid, const char *gid,
TimestampTz prepared_at, Oid owner, Oid databaseid)
{
GlobalTransaction gxact;
int i;
if (strlen(gid) >= GIDSIZE)
ereport(ERROR,
(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
errmsg("transaction identifier \"%s\" is too long",
gid)));
/* fail immediately if feature is disabled */
if (max_prepared_xacts == 0)
ereport(ERROR,
(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
errmsg("prepared transactions are disabled"),
errhint("Set max_prepared_transactions to a nonzero value.")));
LWLockAcquire(TwoPhaseStateLock, LW_EXCLUSIVE);
/*
* First, find and recycle any gxacts that failed during prepare. We do
* this partly to ensure we don't mistakenly say their GIDs are still
* reserved, and partly so we don't fail on out-of-slots unnecessarily.
*/
for (i = 0; i < TwoPhaseState->numPrepXacts; i++)
{
gxact = TwoPhaseState->prepXacts[i];
if (!gxact->valid && !TransactionIdIsActive(gxact->locking_xid))
{
/* It's dead Jim ... remove from the active array */
TwoPhaseState->numPrepXacts--;
TwoPhaseState->prepXacts[i] = TwoPhaseState->prepXacts[TwoPhaseState->numPrepXacts];
/* and put it back in the freelist */
gxact->proc.links.next = (SHM_QUEUE *) TwoPhaseState->freeGXacts;
TwoPhaseState->freeGXacts = gxact;
/* Back up index count too, so we don't miss scanning one */
i--;
}
}
/* Check for conflicting GID */
for (i = 0; i < TwoPhaseState->numPrepXacts; i++)
{
gxact = TwoPhaseState->prepXacts[i];
if (strcmp(gxact->gid, gid) == 0)
{
ereport(ERROR,
(errcode(ERRCODE_DUPLICATE_OBJECT),
errmsg("transaction identifier \"%s\" is already in use",
gid)));
}
}
/* Get a free gxact from the freelist */
if (TwoPhaseState->freeGXacts == NULL)
ereport(ERROR,
(errcode(ERRCODE_OUT_OF_MEMORY),
errmsg("maximum number of prepared transactions reached"),
errhint("Increase max_prepared_transactions (currently %d).",
max_prepared_xacts)));
gxact = TwoPhaseState->freeGXacts;
TwoPhaseState->freeGXacts = (GlobalTransaction) gxact->proc.links.next;
/* Initialize it */
MemSet(&gxact->proc, 0, sizeof(PGPROC));
SHMQueueElemInit(&(gxact->proc.links));
gxact->proc.waitStatus = STATUS_OK;
/* We set up the gxact's VXID as InvalidBackendId/XID */
gxact->proc.lxid = (LocalTransactionId) xid;
gxact->proc.xid = xid;
gxact->proc.xmin = InvalidTransactionId;
gxact->proc.pid = 0;
gxact->proc.backendId = InvalidBackendId;
gxact->proc.databaseId = databaseid;
gxact->proc.roleId = owner;
gxact->proc.inCommit = false;
gxact->proc.vacuumFlags = 0;
gxact->proc.lwWaiting = false;
gxact->proc.lwExclusive = false;
gxact->proc.lwWaitLink = NULL;
gxact->proc.waitLock = NULL;
gxact->proc.waitProcLock = NULL;
for (i = 0; i < NUM_LOCK_PARTITIONS; i++)
SHMQueueInit(&(gxact->proc.myProcLocks[i]));
/* subxid data must be filled later by GXactLoadSubxactData */
gxact->proc.subxids.overflowed = false;
gxact->proc.subxids.nxids = 0;
gxact->prepared_at = prepared_at;
/* initialize LSN to 0 (start of WAL) */
gxact->prepare_lsn.xlogid = 0;
gxact->prepare_lsn.xrecoff = 0;
gxact->owner = owner;
gxact->locking_xid = xid;
gxact->valid = false;
strcpy(gxact->gid, gid);
/* And insert it into the active array */
Assert(TwoPhaseState->numPrepXacts < max_prepared_xacts);
TwoPhaseState->prepXacts[TwoPhaseState->numPrepXacts++] = gxact;
LWLockRelease(TwoPhaseStateLock);
return gxact;
}
/* ------------------------
* InitProc -- create a per-process data structure for this process
* used by the lock manager on semaphore queues.
* ------------------------
*/
void
InitProcess(IPCKey key)
{
bool found = false;
int pid;
int semstat;
unsigned long location, myOffset;
/* ------------------
* Routine called if deadlock timer goes off. See ProcSleep()
* ------------------
*/
#ifndef WIN32
signal(SIGALRM, HandleDeadLock);
#endif /* WIN32 we'll have to figure out how to handle this later */
SpinAcquire(ProcStructLock);
/* attach to the free list */
ProcGlobal = (PROC_HDR *)
ShmemInitStruct("Proc Header",(unsigned)sizeof(PROC_HDR),&found);
if (!found) {
/* this should not happen. InitProcGlobal() is called before this. */
elog(WARN, "InitProcess: Proc Header uninitialized");
}
if (MyProc != NULL)
{
SpinRelease(ProcStructLock);
elog(WARN,"ProcInit: you already exist");
return;
}
/* try to get a proc from the free list first */
myOffset = ProcGlobal->freeProcs;
if (myOffset != INVALID_OFFSET)
{
MyProc = (PROC *) MAKE_PTR(myOffset);
ProcGlobal->freeProcs = MyProc->links.next;
}
else
{
/* have to allocate one. We can't use the normal binding
* table mechanism because the proc structure is stored
* by PID instead of by a global name (need to look it
* up by PID when we cleanup dead processes).
*/
MyProc = (PROC *) ShmemAlloc((unsigned)sizeof(PROC));
if (! MyProc)
{
SpinRelease(ProcStructLock);
elog (FATAL,"cannot create new proc: out of memory");
}
/* this cannot be initialized until after the buffer pool */
SHMQueueInit(&(MyProc->lockQueue));
MyProc->procId = ProcGlobal->numProcs;
ProcGlobal->numProcs++;
}
/*
* zero out the spin lock counts and set the sLocks field for
* ProcStructLock to 1 as we have acquired this spinlock above but
* didn't record it since we didn't have MyProc until now.
*/
memset(MyProc->sLocks, 0, sizeof(MyProc->sLocks));
MyProc->sLocks[ProcStructLock] = 1;
if (IsUnderPostmaster) {
IPCKey semKey;
int semNum;
int semId;
union semun semun;
ProcGetNewSemKeyAndNum(&semKey, &semNum);
semId = IpcSemaphoreCreate(semKey,
PROC_NSEMS_PER_SET,
IPCProtection,
IpcSemaphoreDefaultStartValue,
0,
&semstat);
/*
* we might be reusing a semaphore that belongs to a dead
* backend. So be careful and reinitialize its value here.
*/
semun.val = IpcSemaphoreDefaultStartValue;
semctl(semId, semNum, SETVAL, semun);
IpcSemaphoreLock(semId, semNum, IpcExclusiveLock);
MyProc->sem.semId = semId;
MyProc->sem.semNum = semNum;
MyProc->sem.semKey = semKey;
} else {
MyProc->sem.semId = -1;
}
/* ----------------------
* Release the lock.
* ----------------------
*/
SpinRelease(ProcStructLock);
MyProc->pid = 0;
#if 0
MyProc->pid = MyPid;
#endif
/* ----------------
* Start keeping spin lock stats from here on. Any botch before
* this initialization is forever botched
* ----------------
*/
memset(MyProc->sLocks, 0, MAX_SPINS*sizeof(*MyProc->sLocks));
/* -------------------------
* Install ourselves in the binding table. The name to
* use is determined by the OS-assigned process id. That
* allows the cleanup process to find us after any untimely
* exit.
* -------------------------
*/
pid = getpid();
location = MAKE_OFFSET(MyProc);
if ((! ShmemPIDLookup(pid,&location)) || (location != MAKE_OFFSET(MyProc)))
{
elog(FATAL,"InitProc: ShmemPID table broken");
}
MyProc->errType = NO_ERROR;
SHMQueueElemInit(&(MyProc->links));
on_exitpg(ProcKill, (caddr_t)pid);
ProcInitialized = TRUE;
}
/*
* InitAuxiliaryProcess -- create a per-auxiliary-process data structure
*
* This is called by bgwriter and similar processes so that they will have a
* MyProc value that's real enough to let them wait for LWLocks. The PGPROC
* and sema that are assigned are one of the extra ones created during
* InitProcGlobal.
*
* Auxiliary processes are presently not expected to wait for real (lockmgr)
* locks, so we need not set up the deadlock checker. They are never added
* to the ProcArray or the sinval messaging mechanism, either. They also
* don't get a VXID assigned, since this is only useful when we actually
* hold lockmgr locks.
*/
void
InitAuxiliaryProcess(void)
{
PGPROC *auxproc;
int proctype;
int i;
/*
* ProcGlobal should be set up already (if we are a backend, we inherit
* this by fork() or EXEC_BACKEND mechanism from the postmaster).
*/
if (ProcGlobal == NULL || AuxiliaryProcs == NULL)
elog(PANIC, "proc header uninitialized");
if (MyProc != NULL)
elog(ERROR, "you already exist");
/*
* Find a free auxproc entry. Use compare_and_swap to avoid locking.
*/
for (proctype = 0; proctype < NUM_AUXILIARY_PROCS; proctype++)
{
auxproc = &AuxiliaryProcs[proctype];
if (compare_and_swap_32((uint32*)(&(auxproc->pid)),
0,
MyProcPid))
{
/* Find a free entry, break here. */
break;
}
}
if (proctype >= NUM_AUXILIARY_PROCS)
{
elog(FATAL, "all AuxiliaryProcs are in use");
}
set_spins_per_delay(ProcGlobal->spins_per_delay);
MyProc = auxproc;
lockHolderProcPtr = auxproc;
/*
* Initialize all fields of MyProc, except for the semaphore which was
* prepared for us by InitProcGlobal.
*/
SHMQueueElemInit(&(MyProc->links));
MyProc->waitStatus = STATUS_OK;
MyProc->xid = InvalidTransactionId;
MyProc->xmin = InvalidTransactionId;
MyProc->databaseId = InvalidOid;
MyProc->roleId = InvalidOid;
MyProc->mppLocalProcessSerial = 0;
MyProc->mppSessionId = 0;
MyProc->mppIsWriter = false;
MyProc->inVacuum = false;
MyProc->postmasterResetRequired = true;
MyProc->lwWaiting = false;
MyProc->lwExclusive = false;
MyProc->lwWaitLink = NULL;
MyProc->waitLock = NULL;
MyProc->waitProcLock = NULL;
for (i = 0; i < NUM_LOCK_PARTITIONS; i++)
SHMQueueInit(&(MyProc->myProcLocks[i]));
/*
* We might be reusing a semaphore that belonged to a failed process. So
* be careful and reinitialize its value here. (This is not strictly
* necessary anymore, but seems like a good idea for cleanliness.)
*/
PGSemaphoreReset(&MyProc->sem);
MyProc->queryCommandId = -1;
/*
* Arrange to clean up at process exit.
*/
on_shmem_exit(AuxiliaryProcKill, Int32GetDatum(proctype));
}
/*
* InitProcess -- initialize a per-process data structure for this backend
*/
void
InitProcess(void)
{
/* use volatile pointer to prevent code rearrangement */
volatile PROC_HDR *procglobal = ProcGlobal;
int i;
/*
* ProcGlobal should be set up already (if we are a backend, we inherit
* this by fork() or EXEC_BACKEND mechanism from the postmaster).
*/
if (procglobal == NULL)
elog(PANIC, "proc header uninitialized");
if (MyProc != NULL)
elog(ERROR, "you already exist");
MyProc = RemoveFirst();
if (MyProc == NULL)
{
ereport(FATAL,
(errcode(ERRCODE_TOO_MANY_CONNECTIONS),
errmsg("sorry, too many clients already")));
}
if (gp_debug_pgproc)
{
elog(LOG, "allocating PGPROC entry for pid %d, freeProcs (prev offset, new offset): (%ld, %ld)",
MyProcPid, MAKE_OFFSET(MyProc), MyProc->links.next);
}
set_spins_per_delay(procglobal->spins_per_delay);
int mppLocalProcessSerial = gp_atomic_add_32(&procglobal->mppLocalProcessCounter, 1);
lockHolderProcPtr = MyProc;
/* Set the next pointer to INVALID_OFFSET */
MyProc->links.next = INVALID_OFFSET;
/*
* Initialize all fields of MyProc, except for the semaphore which was
* prepared for us by InitProcGlobal.
*/
SHMQueueElemInit(&(MyProc->links));
MyProc->waitStatus = STATUS_OK;
MyProc->xid = InvalidTransactionId;
MyProc->xmin = InvalidTransactionId;
MyProc->pid = MyProcPid;
/* databaseId and roleId will be filled in later */
MyProc->databaseId = InvalidOid;
MyProc->roleId = InvalidOid;
MyProc->inVacuum = false;
MyProc->postmasterResetRequired = true;
MyProc->lwWaiting = false;
MyProc->lwExclusive = false;
MyProc->lwWaitLink = NULL;
MyProc->waitLock = NULL;
MyProc->waitProcLock = NULL;
for (i = 0; i < NUM_LOCK_PARTITIONS; i++)
SHMQueueInit(&(MyProc->myProcLocks[i]));
/*
* mppLocalProcessSerial uniquely identifies this backend process among
* all those that our parent postmaster process creates over its lifetime.
*
* Since we use the process serial number to decide if we should
* deliver a response from a server under this spin, we need to
* assign it under the spin lock.
*/
MyProc->mppLocalProcessSerial = mppLocalProcessSerial;
/*
* A nonzero gp_session_id uniquely identifies an MPP client session
* over the lifetime of the entry postmaster process. A qDisp passes
* its gp_session_id down to all of its qExecs. If this is a qExec,
* we have already received the gp_session_id from the qDisp.
*/
elog(DEBUG1,"InitProcess(): gp_session_id %d", gp_session_id);
if (Gp_role == GP_ROLE_DISPATCH && gp_session_id == -1)
gp_session_id = mppLocalProcessSerial;
MyProc->mppSessionId = gp_session_id;
MyProc->mppIsWriter = Gp_is_writer;
/*
* We might be reusing a semaphore that belonged to a failed process. So
* be careful and reinitialize its value here. (This is not strictly
* necessary anymore, but seems like a good idea for cleanliness.)
*/
PGSemaphoreReset(&MyProc->sem);
/* Set wait portal (do not check if resource scheduling is enabled) */
MyProc->waitPortalId = INVALID_PORTALID;
MyProc->queryCommandId = -1;
/*
* Arrange to clean up at backend exit.
*/
on_shmem_exit(ProcKill, 0);
/*
* Now that we have a PGPROC, we could try to acquire locks, so initialize
* the deadlock checker.
*/
InitDeadLockChecking();
}