static void BgwPoolMainLoop(Datum arg)
{
BgwPoolExecutorCtx* ctx = (BgwPoolExecutorCtx*)arg;
int id = ctx->id;
BgwPool* pool = ctx->constructor();
int size;
void* work;
BackgroundWorkerUnblockSignals();
BackgroundWorkerInitializeConnection(pool->dbname, NULL);
while(true) {
PGSemaphoreLock(&pool->available);
SpinLockAcquire(&pool->lock);
size = *(int*)&pool->queue[pool->head];
Assert(size < pool->size);
work = malloc(size);
pool->pending -= 1;
pool->active += 1;
if (pool->lastPeakTime == 0 && pool->active == pool->nWorkers && pool->pending != 0) {
pool->lastPeakTime = MtmGetSystemTime();
}
if (pool->head + size + 4 > pool->size) {
memcpy(work, pool->queue, size);
pool->head = INTALIGN(size);
} else {
memcpy(work, &pool->queue[pool->head+4], size);
pool->head += 4 + INTALIGN(size);
}
if (pool->size == pool->head) {
pool->head = 0;
}
if (pool->producerBlocked) {
pool->producerBlocked = false;
PGSemaphoreUnlock(&pool->overflow);
pool->lastPeakTime = 0;
}
SpinLockRelease(&pool->lock);
pool->executor(id, work, size);
free(work);
SpinLockAcquire(&pool->lock);
pool->active -= 1;
pool->lastPeakTime = 0;
SpinLockRelease(&pool->lock);
}
}
void BgwPoolExecute(BgwPool* pool, void* work, size_t size)
{
if (size+4 > pool->size) {
/*
* Size of work is larger than size of shared buffer:
* run it immediately
*/
pool->executor(0, work, size);
return;
}
SpinLockAcquire(&pool->lock);
while (true) {
if ((pool->head <= pool->tail && pool->size - pool->tail < size + 4 && pool->head < size)
|| (pool->head > pool->tail && pool->head - pool->tail < size + 4))
{
if (pool->lastPeakTime == 0) {
pool->lastPeakTime = MtmGetSystemTime();
}
pool->producerBlocked = true;
SpinLockRelease(&pool->lock);
PGSemaphoreLock(&pool->overflow);
SpinLockAcquire(&pool->lock);
} else {
pool->pending += 1;
if (pool->lastPeakTime == 0 && pool->active == pool->nWorkers && pool->pending != 0) {
pool->lastPeakTime = MtmGetSystemTime();
}
*(int*)&pool->queue[pool->tail] = size;
if (pool->size - pool->tail >= size + 4) {
memcpy(&pool->queue[pool->tail+4], work, size);
pool->tail += 4 + INTALIGN(size);
} else {
memcpy(pool->queue, work, size);
pool->tail = INTALIGN(size);
}
if (pool->tail == pool->size) {
pool->tail = 0;
}
PGSemaphoreUnlock(&pool->available);
break;
}
}
SpinLockRelease(&pool->lock);
}
int
main(int argc, char **argv)
{
MyStorage *storage;
int cpid;
printf("Creating shared memory ... ");
fflush(stdout);
storage = (MyStorage *) PGSharedMemoryCreate(8192, false, 5433);
storage->flag = 1234;
printf("OK\n");
printf("Creating semaphores ... ");
fflush(stdout);
PGReserveSemaphores(2, 5433);
PGSemaphoreCreate(&storage->sem);
printf("OK\n");
/* sema initial value is 1, so lock should work */
printf("Testing Lock ... ");
fflush(stdout);
PGSemaphoreLock(&storage->sem, false);
printf("OK\n");
/* now sema value is 0, so trylock should fail */
printf("Testing TryLock ... ");
fflush(stdout);
if (PGSemaphoreTryLock(&storage->sem))
printf("unexpected result!\n");
else
printf("OK\n");
/* unlocking twice and then locking twice should work... */
printf("Testing Multiple Lock ... ");
fflush(stdout);
PGSemaphoreUnlock(&storage->sem);
PGSemaphoreUnlock(&storage->sem);
PGSemaphoreLock(&storage->sem, false);
PGSemaphoreLock(&storage->sem, false);
printf("OK\n");
/* check Reset too */
printf("Testing Reset ... ");
fflush(stdout);
PGSemaphoreUnlock(&storage->sem);
PGSemaphoreReset(&storage->sem);
if (PGSemaphoreTryLock(&storage->sem))
printf("unexpected result!\n");
else
printf("OK\n");
/* Fork a child process and see if it can communicate */
printf("Forking child process ... ");
fflush(stdout);
cpid = fork();
if (cpid == 0)
{
/* In child */
on_exit_reset();
sleep(3);
storage->flag++;
PGSemaphoreUnlock(&storage->sem);
proc_exit(0);
}
if (cpid < 0)
{
/* Fork failed */
printf("failed: %s\n", strerror(errno));
proc_exit(1);
}
printf("forked child pid %d OK\n", cpid);
if (storage->flag != 1234)
printf("Wrong value found in shared memory!\n");
printf("Waiting for child (should wait 3 sec here) ... ");
fflush(stdout);
PGSemaphoreLock(&storage->sem, false);
printf("OK\n");
if (storage->flag != 1235)
printf("Wrong value found in shared memory!\n");
/* Test shutdown */
printf("Running shmem_exit processing ... ");
fflush(stdout);
shmem_exit(0);
printf("OK\n");
printf("Tests complete.\n");
proc_exit(0);
return 0; /* not reached */
}
/*
* LWLockAcquire - acquire a lightweight lock in the specified mode
*
* If the lock is not available, sleep until it is.
*
* Side effect: cancel/die interrupts are held off until lock release.
*/
void
LWLockAcquire(LWLockId lockid, LWLockMode mode)
{
volatile LWLock *lock = &(LWLockArray[lockid].lock);
#if LWLOCK_LOCK_PARTS > 1
volatile LWLockPart *part = LWLOCK_PART(lock, lockid, MyBackendId);
#endif
PGPROC *proc = MyProc;
bool retry = false;
int extraWaits = 0;
PRINT_LWDEBUG("LWLockAcquire", lockid, lock);
#ifdef LWLOCK_STATS
/* Set up local count state first time through in a given process */
if (counts_for_pid != MyProcPid)
{
int *LWLockCounter = (int *) ((char *) LWLockArray - 2 * sizeof(int));
int numLocks = LWLockCounter[1];
sh_acquire_counts = calloc(numLocks, sizeof(int));
ex_acquire_counts = calloc(numLocks, sizeof(int));
block_counts = calloc(numLocks, sizeof(int));
counts_for_pid = MyProcPid;
on_shmem_exit(print_lwlock_stats, 0);
}
/* Count lock acquisition attempts */
if (mode == LW_EXCLUSIVE)
ex_acquire_counts[lockid]++;
else
sh_acquire_counts[lockid]++;
#endif /* LWLOCK_STATS */
/*
* We can't wait if we haven't got a PGPROC. This should only occur
* during bootstrap or shared memory initialization. Put an Assert here
* to catch unsafe coding practices.
*/
Assert(!(proc == NULL && IsUnderPostmaster));
/* Ensure we will have room to remember the lock */
if (num_held_lwlocks >= MAX_SIMUL_LWLOCKS)
elog(ERROR, "too many LWLocks taken");
/*
* Lock out cancel/die interrupts until we exit the code section protected
* by the LWLock. This ensures that interrupts will not interfere with
* manipulations of data structures in shared memory.
*/
HOLD_INTERRUPTS();
/*
* Loop here to try to acquire lock after each time we are signaled by
* LWLockRelease.
*
* NOTE: it might seem better to have LWLockRelease actually grant us the
* lock, rather than retrying and possibly having to go back to sleep. But
* in practice that is no good because it means a process swap for every
* lock acquisition when two or more processes are contending for the same
* lock. Since LWLocks are normally used to protect not-very-long
* sections of computation, a process needs to be able to acquire and
* release the same lock many times during a single CPU time slice, even
* in the presence of contention. The efficiency of being able to do that
* outweighs the inefficiency of sometimes wasting a process dispatch
* cycle because the lock is not free when a released waiter finally gets
* to run. See pgsql-hackers archives for 29-Dec-01.
*/
for (;;)
{
bool mustwait;
if (mode == LW_SHARED)
{
#ifdef LWLOCK_PART_SHARED_OPS_ATOMIC
/* Increment shared counter partition. If there's no contention,
* this is sufficient to take the lock
*/
LWLOCK_PART_SHARED_POSTINC_ATOMIC(lock, lockid, part, MyBackendId);
LWLOCK_PART_SHARED_FENCE();
/* A concurrent exclusive locking attempt does the following
* three steps
* 1) Acquire mutex
* 2) Check shared counter partitions for readers.
* 3a) If found add proc to wait queue, block, restart at (1)
* 3b) If not found, set exclusive flag, continue with (4)
* 4) Enter protected section
* The fence after the atomic add above ensures that no further
* such attempt can proceed to (3b) or beyond. There may be
* pre-existing exclusive locking attempts at step (3b) or beyond,
* but we can recognize those by either the mutex being taken, or
* the exclusive flag being set. Conversely, if we see neither, we
* may proceed and enter the protected section.
*
* FIXME: This doesn't work if slock_t is a struct or doesn't
* use 0 for state "unlocked".
*/
if ((lock->mutex == 0) && (lock->exclusive == 0)) {
/* If retrying, allow LWLockRelease to release waiters again.
* Usually this happens after we acquired the mutex, but if
* we skip that, we still need to set releaseOK.
*
* Acquiring the mutex here is not really an option - if many
* reader are awoken simultaneously by an exclusive unlock,
* that would be a source of considerable contention.
*
* Fotunately, this is safe even without the mutex. First,
* there actually cannot be any non-fast path unlocking
* attempt in progress, because we'd then either still see
* the exclusive flag set or the mutex being taken. And
* even if there was, and such an attempt cleared the flag
* immediately after we set it, it'd also wake up some waiter
* who'd then re-set the flag.
*
* The only reason to do this here, and not directly
* after returning from PGSemaphoreLock(), is that it seems
* benefical to make SpinLockAcquire() the first thing to
* touch the lock if possible, in case we acquire the spin
* lock at all. That way, the cache line doesn't go through
* a possible shared state, but instead directly to exclusive.
* On Opterons at least, there seems to be a difference, c.f.
* the comment above tas() for x86_64 in s_lock.h
*/
if (retry && !lock->releaseOK)
lock->releaseOK = true;
goto lock_acquired;
}
/* At this point, we don't know if the concurrent exclusive locker
* has proceeded to (3b) or blocked. We must take the mutex and
* re-check
*/
#endif /* LWLOCK_PART_SHARED_OPS_ATOMIC */
/* Acquire mutex. Time spent holding mutex should be short! */
SpinLockAcquire(&lock->mutex);
if (lock->exclusive == 0)
{
#ifdef LWLOCK_PART_SHARED_OPS_ATOMIC
/* Already incremented the shared counter partition above */
#else
lock->shared++;
#endif
mustwait = false;
}
else
{
#ifdef LWLOCK_PART_SHARED_OPS_ATOMIC
/* Must undo shared counter partition increment. Note that
* we *need* to do that while holding the mutex. Otherwise,
* the exclusive lock could be released and attempted to be
* re-acquired before we undo the increment. That attempt
* would then block, even though there'd be no lock holder
* left
*/
LWLOCK_PART_SHARED_POSTDEC_ATOMIC(lock, lockid, part, MyBackendId);
#endif
mustwait = true;
}
}
else
{
/* Step (1). Acquire mutex. Time spent holding mutex should be
* short!
*/
SpinLockAcquire(&lock->mutex);
if (lock->exclusive == 0)
{
/* Step (2). Check for shared lockers. This surely happens
* after (1), otherwise SpinLockAcquire() is broken. Lock
* acquire semantics demand that no load must be re-ordered
* from after a lock acquisition to before, for obvious
* reasons.
*/
LWLOCK_IS_SHARED(mustwait, lock, lockid);
if (!mustwait) {
/* Step (3a). Set exclusive flag. This surely happens
* after (2) because it depends on the result of (2),
* no matter how much reordering is going on here.
*/
lock->exclusive++;
}
}
else
mustwait = true;
}
/* If retrying, allow LWLockRelease to release waiters again.
* This is also separately done in the LW_SHARED early exit case
* above, and in contrast to there we don't hold the mutex there.
* See the comment there for why this is safe
*/
if (retry)
lock->releaseOK = true;
if (!mustwait)
break; /* got the lock */
/*
* Step (3b). Add myself to wait queue.
*
* If we don't have a PGPROC structure, there's no way to wait. This
* should never occur, since MyProc should only be null during shared
* memory initialization.
*/
if (proc == NULL)
elog(PANIC, "cannot wait without a PGPROC structure");
proc->lwWaiting = true;
proc->lwExclusive = (mode == LW_EXCLUSIVE);
proc->lwWaitLink = NULL;
if (lock->head == NULL)
lock->head = proc;
else
lock->tail->lwWaitLink = proc;
lock->tail = proc;
/* Can release the mutex now */
SpinLockRelease(&lock->mutex);
/*
* Wait until awakened.
*
* Since we share the process wait semaphore with the regular lock
* manager and ProcWaitForSignal, and we may need to acquire an LWLock
* while one of those is pending, it is possible that we get awakened
* for a reason other than being signaled by LWLockRelease. If so,
* loop back and wait again. Once we've gotten the LWLock,
* re-increment the sema by the number of additional signals received,
* so that the lock manager or signal manager will see the received
* signal when it next waits.
*/
LOG_LWDEBUG("LWLockAcquire", lockid, "waiting");
#ifdef LWLOCK_STATS
block_counts[lockid]++;
#endif
TRACE_POSTGRESQL_LWLOCK_WAIT_START(lockid, mode);
for (;;)
{
/* "false" means cannot accept cancel/die interrupt here. */
PGSemaphoreLock(&proc->sem, false);
if (!proc->lwWaiting)
break;
extraWaits++;
}
TRACE_POSTGRESQL_LWLOCK_WAIT_DONE(lockid, mode);
LOG_LWDEBUG("LWLockAcquire", lockid, "awakened");
/* Now loop back and try to acquire lock again. */
retry = true;
}
/* We are done updating shared state of the lock itself. */
SpinLockRelease(&lock->mutex);
/* Step 4. Enter protected section. This surely happens after (3),
* this time because lock release semantics demand that no store
* must be moved from before a lock release to after the release,
* again for obvious reasons
*/
#ifdef LWLOCK_PART_SHARED_OPS_ATOMIC
lock_acquired:
#endif
TRACE_POSTGRESQL_LWLOCK_ACQUIRE(lockid, mode);
/* Add lock to list of locks held by this backend */
held_lwlocks[num_held_lwlocks] = lockid;
held_lwlocks_mode[num_held_lwlocks] = mode;
++num_held_lwlocks;
/*
* Fix the process wait semaphore's count for any absorbed wakeups.
*/
while (extraWaits-- > 0)
PGSemaphoreUnlock(&proc->sem);
}
/*
* LWLockAcquire - acquire a lightweight lock in the specified mode
*
* If the lock is not available, sleep until it is.
*
* Side effect: cancel/die interrupts are held off until lock release.
*/
void
LWLockAcquire(LWLockId lockid, LWLockMode mode)
{
volatile LWLock *lock = &(LWLockArray[lockid].lock);
PGPROC *proc = MyProc;
bool retry = false;
int extraWaits = 0;
PRINT_LWDEBUG("LWLockAcquire", lockid, lock);
#ifdef LWLOCK_STATS
/* Set up local count state first time through in a given process */
if (counts_for_pid != MyProcPid)
{
int *LWLockCounter = (int *) ((char *) LWLockArray - 2 * sizeof(int));
int numLocks = LWLockCounter[1];
sh_acquire_counts = calloc(numLocks, sizeof(int));
ex_acquire_counts = calloc(numLocks, sizeof(int));
block_counts = calloc(numLocks, sizeof(int));
counts_for_pid = MyProcPid;
on_shmem_exit(print_lwlock_stats, 0);
}
/* Count lock acquisition attempts */
if (mode == LW_EXCLUSIVE)
ex_acquire_counts[lockid]++;
else
sh_acquire_counts[lockid]++;
#endif /* LWLOCK_STATS */
/*
* We can't wait if we haven't got a PGPROC. This should only occur
* during bootstrap or shared memory initialization. Put an Assert here
* to catch unsafe coding practices.
*/
Assert(!(proc == NULL && IsUnderPostmaster));
/* Ensure we will have room to remember the lock */
if (num_held_lwlocks >= MAX_SIMUL_LWLOCKS)
elog(ERROR, "too many LWLocks taken");
/*
* Lock out cancel/die interrupts until we exit the code section protected
* by the LWLock. This ensures that interrupts will not interfere with
* manipulations of data structures in shared memory.
*/
HOLD_INTERRUPTS();
/*
* Loop here to try to acquire lock after each time we are signaled by
* LWLockRelease.
*
* NOTE: it might seem better to have LWLockRelease actually grant us the
* lock, rather than retrying and possibly having to go back to sleep. But
* in practice that is no good because it means a process swap for every
* lock acquisition when two or more processes are contending for the same
* lock. Since LWLocks are normally used to protect not-very-long
* sections of computation, a process needs to be able to acquire and
* release the same lock many times during a single CPU time slice, even
* in the presence of contention. The efficiency of being able to do that
* outweighs the inefficiency of sometimes wasting a process dispatch
* cycle because the lock is not free when a released waiter finally gets
* to run. See pgsql-hackers archives for 29-Dec-01.
*/
for (;;)
{
bool mustwait;
/* Acquire mutex. Time spent holding mutex should be short! */
SpinLockAcquire(&lock->mutex);
/* If retrying, allow LWLockRelease to release waiters again */
if (retry)
lock->releaseOK = true;
/* If I can get the lock, do so quickly. */
if (mode == LW_EXCLUSIVE)
{
if (lock->exclusive == 0 && lock->shared == 0)
{
lock->exclusive++;
mustwait = false;
}
else
mustwait = true;
}
else
{
if (lock->exclusive == 0)
{
lock->shared++;
mustwait = false;
}
else
mustwait = true;
}
if (!mustwait)
break; /* got the lock */
/*
* Add myself to wait queue.
*
* If we don't have a PGPROC structure, there's no way to wait. This
* should never occur, since MyProc should only be null during shared
* memory initialization.
*/
if (proc == NULL)
elog(PANIC, "cannot wait without a PGPROC structure");
proc->lwWaiting = true;
proc->lwExclusive = (mode == LW_EXCLUSIVE);
proc->lwWaitLink = NULL;
if (lock->head == NULL)
lock->head = proc;
else
lock->tail->lwWaitLink = proc;
lock->tail = proc;
/* Can release the mutex now */
SpinLockRelease(&lock->mutex);
/*
* Wait until awakened.
*
* Since we share the process wait semaphore with the regular lock
* manager and ProcWaitForSignal, and we may need to acquire an LWLock
* while one of those is pending, it is possible that we get awakened
* for a reason other than being signaled by LWLockRelease. If so,
* loop back and wait again. Once we've gotten the LWLock,
* re-increment the sema by the number of additional signals received,
* so that the lock manager or signal manager will see the received
* signal when it next waits.
*/
LOG_LWDEBUG("LWLockAcquire", lockid, "waiting");
#ifdef LWLOCK_STATS
block_counts[lockid]++;
#endif
TRACE_POSTGRESQL_LWLOCK_WAIT_START(lockid, mode);
for (;;)
{
/* "false" means cannot accept cancel/die interrupt here. */
PGSemaphoreLock(&proc->sem, false);
if (!proc->lwWaiting)
break;
extraWaits++;
}
TRACE_POSTGRESQL_LWLOCK_WAIT_DONE(lockid, mode);
LOG_LWDEBUG("LWLockAcquire", lockid, "awakened");
/* Now loop back and try to acquire lock again. */
retry = true;
}
/* We are done updating shared state of the lock itself. */
SpinLockRelease(&lock->mutex);
TRACE_POSTGRESQL_LWLOCK_ACQUIRE(lockid, mode);
/* Add lock to list of locks held by this backend */
held_lwlocks[num_held_lwlocks++] = lockid;
/*
* Fix the process wait semaphore's count for any absorbed wakeups.
*/
while (extraWaits-- > 0)
PGSemaphoreUnlock(&proc->sem);
}
/*
* LWLockWaitForVar - Wait until lock is free, or a variable is updated.
*
* If the lock is held and *valptr equals oldval, waits until the lock is
* either freed, or the lock holder updates *valptr by calling
* LWLockUpdateVar. If the lock is free on exit (immediately or after
* waiting), returns true. If the lock is still held, but *valptr no longer
* matches oldval, returns false and sets *newval to the current value in
* *valptr.
*
* It's possible that the lock holder releases the lock, but another backend
* acquires it again before we get a chance to observe that the lock was
* momentarily released. We wouldn't need to wait for the new lock holder,
* but we cannot distinguish that case, so we will have to wait.
*
* Note: this function ignores shared lock holders; if the lock is held
* in shared mode, returns 'true'.
*/
bool
LWLockWaitForVar(LWLock *lock, uint64 *valptr, uint64 oldval, uint64 *newval)
{
PGPROC *proc = MyProc;
int extraWaits = 0;
bool result = false;
#ifdef LWLOCK_STATS
lwlock_stats *lwstats;
#endif
PRINT_LWDEBUG("LWLockWaitForVar", lock);
#ifdef LWLOCK_STATS
lwstats = get_lwlock_stats_entry(lock);
#endif /* LWLOCK_STATS */
/*
* Quick test first to see if it the slot is free right now.
*
* XXX: the caller uses a spinlock before this, so we don't need a memory
* barrier here as far as the current usage is concerned. But that might
* not be safe in general.
*/
if (lock->exclusive == 0)
return true;
/*
* Lock out cancel/die interrupts while we sleep on the lock. There is no
* cleanup mechanism to remove us from the wait queue if we got
* interrupted.
*/
HOLD_INTERRUPTS();
/*
* Loop here to check the lock's status after each time we are signaled.
*/
for (;;)
{
bool mustwait;
uint64 value;
/* Acquire mutex. Time spent holding mutex should be short! */
#ifdef LWLOCK_STATS
lwstats->spin_delay_count += SpinLockAcquire(&lock->mutex);
#else
SpinLockAcquire(&lock->mutex);
#endif
/* Is the lock now free, and if not, does the value match? */
if (lock->exclusive == 0)
{
result = true;
mustwait = false;
}
else
{
value = *valptr;
if (value != oldval)
{
result = false;
mustwait = false;
*newval = value;
}
else
mustwait = true;
}
if (!mustwait)
break; /* the lock was free or value didn't match */
/*
* Add myself to wait queue.
*/
proc->lwWaiting = true;
proc->lwWaitMode = LW_WAIT_UNTIL_FREE;
/* waiters are added to the front of the queue */
proc->lwWaitLink = lock->head;
if (lock->head == NULL)
lock->tail = proc;
lock->head = proc;
/*
* Set releaseOK, to make sure we get woken up as soon as the lock is
* released.
*/
lock->releaseOK = true;
/* Can release the mutex now */
SpinLockRelease(&lock->mutex);
/*
* Wait until awakened.
*
* Since we share the process wait semaphore with the regular lock
* manager and ProcWaitForSignal, and we may need to acquire an LWLock
* while one of those is pending, it is possible that we get awakened
* for a reason other than being signaled by LWLockRelease. If so,
* loop back and wait again. Once we've gotten the LWLock,
* re-increment the sema by the number of additional signals received,
* so that the lock manager or signal manager will see the received
* signal when it next waits.
*/
LOG_LWDEBUG("LWLockWaitForVar", T_NAME(lock), T_ID(lock), "waiting");
#ifdef LWLOCK_STATS
lwstats->block_count++;
#endif
TRACE_POSTGRESQL_LWLOCK_WAIT_START(T_NAME(lock), T_ID(lock),
LW_EXCLUSIVE);
for (;;)
{
/* "false" means cannot accept cancel/die interrupt here. */
PGSemaphoreLock(&proc->sem, false);
if (!proc->lwWaiting)
break;
extraWaits++;
}
TRACE_POSTGRESQL_LWLOCK_WAIT_DONE(T_NAME(lock), T_ID(lock),
LW_EXCLUSIVE);
LOG_LWDEBUG("LWLockWaitForVar", T_NAME(lock), T_ID(lock), "awakened");
/* Now loop back and check the status of the lock again. */
}
/* We are done updating shared state of the lock itself. */
SpinLockRelease(&lock->mutex);
TRACE_POSTGRESQL_LWLOCK_ACQUIRE(T_NAME(lock), T_ID(lock), LW_EXCLUSIVE);
/*
* Fix the process wait semaphore's count for any absorbed wakeups.
*/
while (extraWaits-- > 0)
PGSemaphoreUnlock(&proc->sem);
/*
* Now okay to allow cancel/die interrupts.
*/
RESUME_INTERRUPTS();
return result;
}
/*
* LWLockAcquireOrWait - Acquire lock, or wait until it's free
*
* The semantics of this function are a bit funky. If the lock is currently
* free, it is acquired in the given mode, and the function returns true. If
* the lock isn't immediately free, the function waits until it is released
* and returns false, but does not acquire the lock.
*
* This is currently used for WALWriteLock: when a backend flushes the WAL,
* holding WALWriteLock, it can flush the commit records of many other
* backends as a side-effect. Those other backends need to wait until the
* flush finishes, but don't need to acquire the lock anymore. They can just
* wake up, observe that their records have already been flushed, and return.
*/
bool
LWLockAcquireOrWait(LWLock *lock, LWLockMode mode)
{
PGPROC *proc = MyProc;
bool mustwait;
int extraWaits = 0;
#ifdef LWLOCK_STATS
lwlock_stats *lwstats;
#endif
PRINT_LWDEBUG("LWLockAcquireOrWait", lock);
#ifdef LWLOCK_STATS
lwstats = get_lwlock_stats_entry(lock);
#endif
/* Ensure we will have room to remember the lock */
if (num_held_lwlocks >= MAX_SIMUL_LWLOCKS)
elog(ERROR, "too many LWLocks taken");
/*
* Lock out cancel/die interrupts until we exit the code section protected
* by the LWLock. This ensures that interrupts will not interfere with
* manipulations of data structures in shared memory.
*/
HOLD_INTERRUPTS();
/* Acquire mutex. Time spent holding mutex should be short! */
SpinLockAcquire(&lock->mutex);
/* If I can get the lock, do so quickly. */
if (mode == LW_EXCLUSIVE)
{
if (lock->exclusive == 0 && lock->shared == 0)
{
lock->exclusive++;
mustwait = false;
}
else
mustwait = true;
}
else
{
if (lock->exclusive == 0)
{
lock->shared++;
mustwait = false;
}
else
mustwait = true;
}
if (mustwait)
{
/*
* Add myself to wait queue.
*
* If we don't have a PGPROC structure, there's no way to wait. This
* should never occur, since MyProc should only be null during shared
* memory initialization.
*/
if (proc == NULL)
elog(PANIC, "cannot wait without a PGPROC structure");
proc->lwWaiting = true;
proc->lwWaitMode = LW_WAIT_UNTIL_FREE;
proc->lwWaitLink = NULL;
if (lock->head == NULL)
lock->head = proc;
else
lock->tail->lwWaitLink = proc;
lock->tail = proc;
/* Can release the mutex now */
SpinLockRelease(&lock->mutex);
/*
* Wait until awakened. Like in LWLockAcquire, be prepared for bogus
* wakups, because we share the semaphore with ProcWaitForSignal.
*/
LOG_LWDEBUG("LWLockAcquireOrWait", T_NAME(lock), T_ID(lock),
"waiting");
#ifdef LWLOCK_STATS
lwstats->block_count++;
#endif
TRACE_POSTGRESQL_LWLOCK_WAIT_START(T_NAME(lock), T_ID(lock), mode);
for (;;)
{
/* "false" means cannot accept cancel/die interrupt here. */
PGSemaphoreLock(&proc->sem, false);
if (!proc->lwWaiting)
break;
extraWaits++;
}
TRACE_POSTGRESQL_LWLOCK_WAIT_DONE(T_NAME(lock), T_ID(lock), mode);
LOG_LWDEBUG("LWLockAcquireOrWait", T_NAME(lock), T_ID(lock),
"awakened");
}
else
{
/* We are done updating shared state of the lock itself. */
SpinLockRelease(&lock->mutex);
}
/*
* Fix the process wait semaphore's count for any absorbed wakeups.
*/
while (extraWaits-- > 0)
PGSemaphoreUnlock(&proc->sem);
if (mustwait)
{
/* Failed to get lock, so release interrupt holdoff */
RESUME_INTERRUPTS();
LOG_LWDEBUG("LWLockAcquireOrWait", T_NAME(lock), T_ID(lock), "failed");
TRACE_POSTGRESQL_LWLOCK_ACQUIRE_OR_WAIT_FAIL(T_NAME(lock), T_ID(lock),
mode);
}
else
{
/* Add lock to list of locks held by this backend */
held_lwlocks[num_held_lwlocks++] = lock;
TRACE_POSTGRESQL_LWLOCK_ACQUIRE_OR_WAIT(T_NAME(lock), T_ID(lock),
mode);
}
return !mustwait;
}
/*
* ProcSleep -- put a process to sleep on the specified lock
*
* Caller must have set MyProc->heldLocks to reflect locks already held
* on the lockable object by this process (under all XIDs).
*
* The lock table's partition lock must be held at entry, and will be held
* at exit.
*
* Result: STATUS_OK if we acquired the lock, STATUS_ERROR if not (deadlock).
*
* ASSUME: that no one will fiddle with the queue until after
* we release the partition lock.
*
* NOTES: The process queue is now a priority queue for locking.
*
* P() on the semaphore should put us to sleep. The process
* semaphore is normally zero, so when we try to acquire it, we sleep.
*/
int
ProcSleep(LOCALLOCK *locallock, LockMethod lockMethodTable)
{
LOCKMODE lockmode = locallock->tag.mode;
LOCK *lock = locallock->lock;
PROCLOCK *proclock = locallock->proclock;
uint32 hashcode = locallock->hashcode;
LWLockId partitionLock = LockHashPartitionLock(hashcode);
PROC_QUEUE *waitQueue = &(lock->waitProcs);
LOCKMASK myHeldLocks = MyProc->heldLocks;
bool early_deadlock = false;
bool allow_autovacuum_cancel = true;
int myWaitStatus;
PGPROC *proc;
int i;
/*
* Determine where to add myself in the wait queue.
*
* Normally I should go at the end of the queue. However, if I already
* hold locks that conflict with the request of any previous waiter, put
* myself in the queue just in front of the first such waiter. This is not
* a necessary step, since deadlock detection would move me to before that
* waiter anyway; but it's relatively cheap to detect such a conflict
* immediately, and avoid delaying till deadlock timeout.
*
* Special case: if I find I should go in front of some waiter, check to
* see if I conflict with already-held locks or the requests before that
* waiter. If not, then just grant myself the requested lock immediately.
* This is the same as the test for immediate grant in LockAcquire, except
* we are only considering the part of the wait queue before my insertion
* point.
*/
if (myHeldLocks != 0)
{
LOCKMASK aheadRequests = 0;
proc = (PGPROC *) waitQueue->links.next;
for (i = 0; i < waitQueue->size; i++)
{
/* Must he wait for me? */
if (lockMethodTable->conflictTab[proc->waitLockMode] & myHeldLocks)
{
/* Must I wait for him ? */
if (lockMethodTable->conflictTab[lockmode] & proc->heldLocks)
{
/*
* Yes, so we have a deadlock. Easiest way to clean up
* correctly is to call RemoveFromWaitQueue(), but we
* can't do that until we are *on* the wait queue. So, set
* a flag to check below, and break out of loop. Also,
* record deadlock info for later message.
*/
RememberSimpleDeadLock(MyProc, lockmode, lock, proc);
early_deadlock = true;
break;
}
/* I must go before this waiter. Check special case. */
if ((lockMethodTable->conflictTab[lockmode] & aheadRequests) == 0 &&
LockCheckConflicts(lockMethodTable,
lockmode,
lock,
proclock,
MyProc) == STATUS_OK)
{
/* Skip the wait and just grant myself the lock. */
GrantLock(lock, proclock, lockmode);
GrantAwaitedLock();
return STATUS_OK;
}
/* Break out of loop to put myself before him */
break;
}
/* Nope, so advance to next waiter */
aheadRequests |= LOCKBIT_ON(proc->waitLockMode);
proc = (PGPROC *) proc->links.next;
}
/*
* If we fall out of loop normally, proc points to waitQueue head, so
* we will insert at tail of queue as desired.
*/
}
else
{
/* I hold no locks, so I can't push in front of anyone. */
proc = (PGPROC *) &(waitQueue->links);
}
/*
* Insert self into queue, ahead of the given proc (or at tail of queue).
*/
SHMQueueInsertBefore(&(proc->links), &(MyProc->links));
waitQueue->size++;
lock->waitMask |= LOCKBIT_ON(lockmode);
/* Set up wait information in PGPROC object, too */
MyProc->waitLock = lock;
MyProc->waitProcLock = proclock;
MyProc->waitLockMode = lockmode;
MyProc->waitStatus = STATUS_WAITING;
/*
* If we detected deadlock, give up without waiting. This must agree with
* CheckDeadLock's recovery code, except that we shouldn't release the
* semaphore since we haven't tried to lock it yet.
*/
if (early_deadlock)
{
RemoveFromWaitQueue(MyProc, hashcode);
return STATUS_ERROR;
}
/* mark that we are waiting for a lock */
lockAwaited = locallock;
/*
* Release the lock table's partition lock.
*
* NOTE: this may also cause us to exit critical-section state, possibly
* allowing a cancel/die interrupt to be accepted. This is OK because we
* have recorded the fact that we are waiting for a lock, and so
* LockErrorCleanup will clean up if cancel/die happens.
*/
LWLockRelease(partitionLock);
/*
* Also, now that we will successfully clean up after an ereport, it's
* safe to check to see if there's a buffer pin deadlock against the
* Startup process. Of course, that's only necessary if we're doing Hot
* Standby and are not the Startup process ourselves.
*/
if (RecoveryInProgress() && !InRecovery)
CheckRecoveryConflictDeadlock();
/* Reset deadlock_state before enabling the timeout handler */
deadlock_state = DS_NOT_YET_CHECKED;
/*
* Set timer so we can wake up after awhile and check for a deadlock. If a
* deadlock is detected, the handler releases the process's semaphore and
* sets MyProc->waitStatus = STATUS_ERROR, allowing us to know that we
* must report failure rather than success.
*
* By delaying the check until we've waited for a bit, we can avoid
* running the rather expensive deadlock-check code in most cases.
*/
enable_timeout_after(DEADLOCK_TIMEOUT, DeadlockTimeout);
/*
* If someone wakes us between LWLockRelease and PGSemaphoreLock,
* PGSemaphoreLock will not block. The wakeup is "saved" by the semaphore
* implementation. While this is normally good, there are cases where a
* saved wakeup might be leftover from a previous operation (for example,
* we aborted ProcWaitForSignal just before someone did ProcSendSignal).
* So, loop to wait again if the waitStatus shows we haven't been granted
* nor denied the lock yet.
*
* We pass interruptOK = true, which eliminates a window in which
* cancel/die interrupts would be held off undesirably. This is a promise
* that we don't mind losing control to a cancel/die interrupt here. We
* don't, because we have no shared-state-change work to do after being
* granted the lock (the grantor did it all). We do have to worry about
* canceling the deadlock timeout and updating the locallock table, but if
* we lose control to an error, LockErrorCleanup will fix that up.
*/
do
{
PGSemaphoreLock(&MyProc->sem, true);
/*
* waitStatus could change from STATUS_WAITING to something else
* asynchronously. Read it just once per loop to prevent surprising
* behavior (such as missing log messages).
*/
myWaitStatus = MyProc->waitStatus;
/*
* If we are not deadlocked, but are waiting on an autovacuum-induced
* task, send a signal to interrupt it.
*/
if (deadlock_state == DS_BLOCKED_BY_AUTOVACUUM && allow_autovacuum_cancel)
{
PGPROC *autovac = GetBlockingAutoVacuumPgproc();
PGXACT *autovac_pgxact = &ProcGlobal->allPgXact[autovac->pgprocno];
LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE);
/*
* Only do it if the worker is not working to protect against Xid
* wraparound.
*/
if ((autovac != NULL) &&
(autovac_pgxact->vacuumFlags & PROC_IS_AUTOVACUUM) &&
!(autovac_pgxact->vacuumFlags & PROC_VACUUM_FOR_WRAPAROUND))
{
int pid = autovac->pid;
StringInfoData locktagbuf;
StringInfoData logbuf; /* errdetail for server log */
initStringInfo(&locktagbuf);
initStringInfo(&logbuf);
DescribeLockTag(&locktagbuf, &lock->tag);
appendStringInfo(&logbuf,
_("Process %d waits for %s on %s"),
MyProcPid,
GetLockmodeName(lock->tag.locktag_lockmethodid,
lockmode),
locktagbuf.data);
/* release lock as quickly as possible */
LWLockRelease(ProcArrayLock);
ereport(LOG,
(errmsg("sending cancel to blocking autovacuum PID %d",
pid),
errdetail_log("%s", logbuf.data)));
pfree(logbuf.data);
pfree(locktagbuf.data);
/* send the autovacuum worker Back to Old Kent Road */
if (kill(pid, SIGINT) < 0)
{
/* Just a warning to allow multiple callers */
ereport(WARNING,
(errmsg("could not send signal to process %d: %m",
pid)));
}
}
else
LWLockRelease(ProcArrayLock);
/* prevent signal from being resent more than once */
allow_autovacuum_cancel = false;
}
/*
* If awoken after the deadlock check interrupt has run, and
* log_lock_waits is on, then report about the wait.
*/
if (log_lock_waits && deadlock_state != DS_NOT_YET_CHECKED)
{
StringInfoData buf;
const char *modename;
long secs;
int usecs;
long msecs;
initStringInfo(&buf);
DescribeLockTag(&buf, &locallock->tag.lock);
modename = GetLockmodeName(locallock->tag.lock.locktag_lockmethodid,
lockmode);
TimestampDifference(get_timeout_start_time(DEADLOCK_TIMEOUT),
GetCurrentTimestamp(),
&secs, &usecs);
msecs = secs * 1000 + usecs / 1000;
usecs = usecs % 1000;
if (deadlock_state == DS_SOFT_DEADLOCK)
ereport(LOG,
(errmsg("process %d avoided deadlock for %s on %s by rearranging queue order after %ld.%03d ms",
MyProcPid, modename, buf.data, msecs, usecs)));
else if (deadlock_state == DS_HARD_DEADLOCK)
{
/*
* This message is a bit redundant with the error that will be
* reported subsequently, but in some cases the error report
* might not make it to the log (eg, if it's caught by an
* exception handler), and we want to ensure all long-wait
* events get logged.
*/
ereport(LOG,
(errmsg("process %d detected deadlock while waiting for %s on %s after %ld.%03d ms",
MyProcPid, modename, buf.data, msecs, usecs)));
}
if (myWaitStatus == STATUS_WAITING)
ereport(LOG,
(errmsg("process %d still waiting for %s on %s after %ld.%03d ms",
MyProcPid, modename, buf.data, msecs, usecs)));
else if (myWaitStatus == STATUS_OK)
ereport(LOG,
(errmsg("process %d acquired %s on %s after %ld.%03d ms",
MyProcPid, modename, buf.data, msecs, usecs)));
else
{
Assert(myWaitStatus == STATUS_ERROR);
/*
* Currently, the deadlock checker always kicks its own
* process, which means that we'll only see STATUS_ERROR when
* deadlock_state == DS_HARD_DEADLOCK, and there's no need to
* print redundant messages. But for completeness and
* future-proofing, print a message if it looks like someone
* else kicked us off the lock.
*/
if (deadlock_state != DS_HARD_DEADLOCK)
ereport(LOG,
(errmsg("process %d failed to acquire %s on %s after %ld.%03d ms",
MyProcPid, modename, buf.data, msecs, usecs)));
}
/*
* At this point we might still need to wait for the lock. Reset
* state so we don't print the above messages again.
*/
deadlock_state = DS_NO_DEADLOCK;
pfree(buf.data);
}
} while (myWaitStatus == STATUS_WAITING);
/*
* Disable the timer, if it's still running
*/
disable_timeout(DEADLOCK_TIMEOUT, false);
/*
* Re-acquire the lock table's partition lock. We have to do this to hold
* off cancel/die interrupts before we can mess with lockAwaited (else we
* might have a missed or duplicated locallock update).
*/
LWLockAcquire(partitionLock, LW_EXCLUSIVE);
/*
* We no longer want LockErrorCleanup to do anything.
*/
lockAwaited = NULL;
/*
* If we got the lock, be sure to remember it in the locallock table.
*/
if (MyProc->waitStatus == STATUS_OK)
GrantAwaitedLock();
/*
* We don't have to do anything else, because the awaker did all the
* necessary update of the lock table and MyProc.
*/
return MyProc->waitStatus;
}
/*
* ProcWaitForSignal - wait for a signal from another backend.
*
* This can share the semaphore normally used for waiting for locks,
* since a backend could never be waiting for a lock and a signal at
* the same time. As with locks, it's OK if the signal arrives just
* before we actually reach the waiting state. Also as with locks,
* it's necessary that the caller be robust against bogus wakeups:
* always check that the desired state has occurred, and wait again
* if not. This copes with possible "leftover" wakeups.
*/
void
ProcWaitForSignal(void)
{
PGSemaphoreLock(&MyProc->sem, true);
}
/*
* LWLockAcquire - acquire a lightweight lock in the specified mode
*
* If the lock is not available, sleep until it is.
*
* Side effect: cancel/die interrupts are held off until lock release.
*/
void
LWLockAcquire(LWLockId lockid, LWLockMode mode)
{
volatile LWLock *lock = LWLockArray + lockid;
PGPROC *proc = MyProc;
bool retry = false;
int extraWaits = 0;
PRINT_LWDEBUG("LWLockAcquire", lockid, lock);
/*
* We can't wait if we haven't got a PGPROC. This should only occur
* during bootstrap or shared memory initialization. Put an Assert
* here to catch unsafe coding practices.
*/
Assert(!(proc == NULL && IsUnderPostmaster));
/*
* Lock out cancel/die interrupts until we exit the code section
* protected by the LWLock. This ensures that interrupts will not
* interfere with manipulations of data structures in shared memory.
*/
HOLD_INTERRUPTS();
/*
* Loop here to try to acquire lock after each time we are signaled by
* LWLockRelease.
*
* NOTE: it might seem better to have LWLockRelease actually grant us the
* lock, rather than retrying and possibly having to go back to sleep.
* But in practice that is no good because it means a process swap for
* every lock acquisition when two or more processes are contending
* for the same lock. Since LWLocks are normally used to protect
* not-very-long sections of computation, a process needs to be able
* to acquire and release the same lock many times during a single CPU
* time slice, even in the presence of contention. The efficiency of
* being able to do that outweighs the inefficiency of sometimes
* wasting a process dispatch cycle because the lock is not free when
* a released waiter finally gets to run. See pgsql-hackers archives
* for 29-Dec-01.
*/
for (;;)
{
bool mustwait;
/* Acquire mutex. Time spent holding mutex should be short! */
SpinLockAcquire_NoHoldoff(&lock->mutex);
/* If retrying, allow LWLockRelease to release waiters again */
if (retry)
lock->releaseOK = true;
/* If I can get the lock, do so quickly. */
if (mode == LW_EXCLUSIVE)
{
if (lock->exclusive == 0 && lock->shared == 0)
{
lock->exclusive++;
mustwait = false;
}
else
mustwait = true;
}
else
{
if (lock->exclusive == 0)
{
lock->shared++;
mustwait = false;
}
else
mustwait = true;
}
if (!mustwait)
break; /* got the lock */
/*
* Add myself to wait queue.
*
* If we don't have a PGPROC structure, there's no way to wait. This
* should never occur, since MyProc should only be null during
* shared memory initialization.
*/
if (proc == NULL)
elog(FATAL, "cannot wait without a PGPROC structure");
proc->lwWaiting = true;
proc->lwExclusive = (mode == LW_EXCLUSIVE);
proc->lwWaitLink = NULL;
if (lock->head == NULL)
lock->head = proc;
else
lock->tail->lwWaitLink = proc;
lock->tail = proc;
/* Can release the mutex now */
SpinLockRelease_NoHoldoff(&lock->mutex);
/*
* Wait until awakened.
*
* Since we share the process wait semaphore with the regular lock
* manager and ProcWaitForSignal, and we may need to acquire an
* LWLock while one of those is pending, it is possible that we
* get awakened for a reason other than being signaled by
* LWLockRelease. If so, loop back and wait again. Once we've
* gotten the LWLock, re-increment the sema by the number of
* additional signals received, so that the lock manager or signal
* manager will see the received signal when it next waits.
*/
LOG_LWDEBUG("LWLockAcquire", lockid, "waiting");
for (;;)
{
/* "false" means cannot accept cancel/die interrupt here. */
PGSemaphoreLock(&proc->sem, false);
if (!proc->lwWaiting)
break;
extraWaits++;
}
LOG_LWDEBUG("LWLockAcquire", lockid, "awakened");
/* Now loop back and try to acquire lock again. */
retry = true;
}
/* We are done updating shared state of the lock itself. */
SpinLockRelease_NoHoldoff(&lock->mutex);
/* Add lock to list of locks held by this backend */
Assert(num_held_lwlocks < MAX_SIMUL_LWLOCKS);
held_lwlocks[num_held_lwlocks++] = lockid;
/*
* Fix the process wait semaphore's count for any absorbed wakeups.
*/
while (extraWaits-- > 0)
PGSemaphoreUnlock(&proc->sem);
}
/*
* ProcSleep -- put a process to sleep on the specified lock
*
* Caller must have set MyProc->heldLocks to reflect locks already held
* on the lockable object by this process (under all XIDs).
*
* The lock table's partition lock must be held at entry, and will be held
* at exit.
*
* Result: STATUS_OK if we acquired the lock, STATUS_ERROR if not (deadlock).
*
* ASSUME: that no one will fiddle with the queue until after
* we release the partition lock.
*
* NOTES: The process queue is now a priority queue for locking.
*
* P() on the semaphore should put us to sleep. The process
* semaphore is normally zero, so when we try to acquire it, we sleep.
*/
int
ProcSleep(LOCALLOCK *locallock, LockMethod lockMethodTable)
{
LOCKMODE lockmode = locallock->tag.mode;
LOCK *lock = locallock->lock;
PROCLOCK *proclock = locallock->proclock;
uint32 hashcode = locallock->hashcode;
LWLockId partitionLock = LockHashPartitionLock(hashcode);
PROC_QUEUE *waitQueue = &(lock->waitProcs);
LOCKMASK myHeldLocks = MyProc->heldLocks;
bool early_deadlock = false;
PGPROC *proc;
int i;
/*
* Determine where to add myself in the wait queue.
*
* Normally I should go at the end of the queue. However, if I already
* hold locks that conflict with the request of any previous waiter, put
* myself in the queue just in front of the first such waiter. This is not
* a necessary step, since deadlock detection would move me to before that
* waiter anyway; but it's relatively cheap to detect such a conflict
* immediately, and avoid delaying till deadlock timeout.
*
* Special case: if I find I should go in front of some waiter, check to
* see if I conflict with already-held locks or the requests before that
* waiter. If not, then just grant myself the requested lock immediately.
* This is the same as the test for immediate grant in LockAcquire, except
* we are only considering the part of the wait queue before my insertion
* point.
*/
if (myHeldLocks != 0)
{
LOCKMASK aheadRequests = 0;
proc = (PGPROC *) MAKE_PTR(waitQueue->links.next);
for (i = 0; i < waitQueue->size; i++)
{
/* Must he wait for me? */
if (lockMethodTable->conflictTab[proc->waitLockMode] & myHeldLocks)
{
/* Must I wait for him ? */
if (lockMethodTable->conflictTab[lockmode] & proc->heldLocks)
{
/*
* Yes, so we have a deadlock. Easiest way to clean up
* correctly is to call RemoveFromWaitQueue(), but we
* can't do that until we are *on* the wait queue. So, set
* a flag to check below, and break out of loop. Also,
* record deadlock info for later message.
*/
RememberSimpleDeadLock(MyProc, lockmode, lock, proc);
early_deadlock = true;
break;
}
/* I must go before this waiter. Check special case. */
if ((lockMethodTable->conflictTab[lockmode] & aheadRequests) == 0 &&
LockCheckConflicts(lockMethodTable,
lockmode,
lock,
proclock,
MyProc) == STATUS_OK)
{
/* Skip the wait and just grant myself the lock. */
GrantLock(lock, proclock, lockmode);
GrantAwaitedLock();
return STATUS_OK;
}
/* Break out of loop to put myself before him */
break;
}
/* Nope, so advance to next waiter */
aheadRequests |= LOCKBIT_ON(proc->waitLockMode);
proc = (PGPROC *) MAKE_PTR(proc->links.next);
}
/*
* If we fall out of loop normally, proc points to waitQueue head, so
* we will insert at tail of queue as desired.
*/
}
else
{
/* I hold no locks, so I can't push in front of anyone. */
proc = (PGPROC *) &(waitQueue->links);
}
/*
* Insert self into queue, ahead of the given proc (or at tail of queue).
*/
SHMQueueInsertBefore(&(proc->links), &(MyProc->links));
waitQueue->size++;
lock->waitMask |= LOCKBIT_ON(lockmode);
/* Set up wait information in PGPROC object, too */
MyProc->waitLock = lock;
MyProc->waitProcLock = proclock;
MyProc->waitLockMode = lockmode;
MyProc->waitStatus = STATUS_WAITING;
/*
* If we detected deadlock, give up without waiting. This must agree with
* CheckDeadLock's recovery code, except that we shouldn't release the
* semaphore since we haven't tried to lock it yet.
*/
if (early_deadlock)
{
RemoveFromWaitQueue(MyProc, hashcode);
return STATUS_ERROR;
}
/* mark that we are waiting for a lock */
lockAwaited = locallock;
/*
* Release the lock table's partition lock.
*
* NOTE: this may also cause us to exit critical-section state, possibly
* allowing a cancel/die interrupt to be accepted. This is OK because we
* have recorded the fact that we are waiting for a lock, and so
* LockWaitCancel will clean up if cancel/die happens.
*/
LWLockRelease(partitionLock);
/*
* Set timer so we can wake up after awhile and check for a deadlock. If a
* deadlock is detected, the handler releases the process's semaphore and
* sets MyProc->waitStatus = STATUS_ERROR, allowing us to know that we
* must report failure rather than success.
*
* By delaying the check until we've waited for a bit, we can avoid
* running the rather expensive deadlock-check code in most cases.
*/
if (!enable_sig_alarm(DeadlockTimeout, false))
elog(FATAL, "could not set timer for process wakeup");
/*
* If someone wakes us between LWLockRelease and PGSemaphoreLock,
* PGSemaphoreLock will not block. The wakeup is "saved" by the semaphore
* implementation. While this is normally good, there are cases where a
* saved wakeup might be leftover from a previous operation (for example,
* we aborted ProcWaitForSignal just before someone did ProcSendSignal).
* So, loop to wait again if the waitStatus shows we haven't been granted
* nor denied the lock yet.
*
* We pass interruptOK = true, which eliminates a window in which
* cancel/die interrupts would be held off undesirably. This is a promise
* that we don't mind losing control to a cancel/die interrupt here. We
* don't, because we have no shared-state-change work to do after being
* granted the lock (the grantor did it all). We do have to worry about
* updating the locallock table, but if we lose control to an error,
* LockWaitCancel will fix that up.
*/
do
{
PGSemaphoreLock(&MyProc->sem, true);
} while (MyProc->waitStatus == STATUS_WAITING);
/*
* Disable the timer, if it's still running
*/
if (!disable_sig_alarm(false))
elog(FATAL, "could not disable timer for process wakeup");
/*
* Re-acquire the lock table's partition lock. We have to do this to hold
* off cancel/die interrupts before we can mess with lockAwaited (else we
* might have a missed or duplicated locallock update).
*/
LWLockAcquire(partitionLock, LW_EXCLUSIVE);
/*
* We no longer want LockWaitCancel to do anything.
*/
lockAwaited = NULL;
/*
* If we got the lock, be sure to remember it in the locallock table.
*/
if (MyProc->waitStatus == STATUS_OK)
GrantAwaitedLock();
/*
* We don't have to do anything else, because the awaker did all the
* necessary update of the lock table and MyProc.
*/
return MyProc->waitStatus;
}