From 65ead81bf77bc80a403bcfb367ab92d3af90f130 Mon Sep 17 00:00:00 2001 From: Thomas Munro Date: Wed, 11 Apr 2018 11:30:36 +1200 Subject: [PATCH] Add WL_EXIT_ON_PM_DEATH pseudo-event. Users of the WaitEventSet and WaitLatch() APIs can now choose between asking for WL_POSTMASTER_DEATH and then handling it explicitly, or asking for WL_EXIT_ON_PM_DEATH to trigger immediate exit on postmaster death. This reduces code duplication, since almost all callers want the latter. Repair all code that was previously ignoring postmaster death completely, or requesting the event but ignoring it, or requesting the event but then doing an unconditional PostmasterIsAlive() call every time through its event loop (which is an expensive syscall on platforms for which we don't have USE_POSTMASTER_DEATH_SIGNAL support). Assert that callers of WaitLatchXXX() under the postmaster remember to ask for either WL_POSTMASTER_DEATH or WL_EXIT_ON_PM_DEATH, to prevent future bugs. The only process that doesn't handle postmaster death is syslogger. It waits until all backends holding the write end of the syslog pipe (including the postmaster) have closed it by exiting, to be sure to capture any parting messages. By using the WaitEventSet API directly it avoids the new assertion, and as a by-product it may be slightly more efficient on platforms that have epoll(). Author: Thomas Munro Reviewed-by: Heikki Linnakangas, Kyotaro Horiguchi Discussion: https://postgr.es/m/CAEepm%3D1TCviRykkUb69ppWLr_V697rzd1j3eZsRMmbXvETfqbQ%40mail.gmail.com, https://postgr.es/m/CAEepm=2LqHzizbe7muD7-2yHUbTOoF7Q+qkSD5Q41kuhttRTwA@mail.gmail.com --- contrib/pg_prewarm/autoprewarm.c | 6 +-- contrib/postgres_fdw/connection.c | 6 ++- src/backend/access/transam/parallel.c | 8 +--- src/backend/access/transam/xlog.c | 6 +-- src/backend/executor/nodeGather.c | 3 +- src/backend/libpq/pqmq.c | 2 +- src/backend/postmaster/autovacuum.c | 16 ++------ src/backend/postmaster/bgwriter.c | 11 +----- src/backend/postmaster/checkpointer.c | 16 ++------ src/backend/postmaster/pgarch.c | 4 +- src/backend/postmaster/syslogger.c | 33 +++++++++++----- src/backend/postmaster/walwriter.c | 16 ++------ src/backend/replication/basebackup.c | 2 +- .../libpqwalreceiver/libpqwalreceiver.c | 13 +------ src/backend/replication/logical/launcher.c | 24 ++---------- src/backend/replication/logical/tablesync.c | 30 ++++---------- src/backend/replication/logical/worker.c | 6 +-- src/backend/replication/syncrep.c | 18 +++++---- src/backend/replication/walreceiver.c | 20 +--------- src/backend/replication/walsender.c | 31 +++------------ src/backend/storage/ipc/latch.c | 39 ++++++++++++++++++- src/backend/storage/ipc/shm_mq.c | 9 +++-- src/backend/storage/lmgr/condition_variable.c | 11 +----- src/backend/storage/lmgr/proc.c | 4 +- src/backend/utils/adt/misc.c | 2 +- src/include/storage/latch.h | 3 +- src/test/modules/test_shm_mq/setup.c | 3 +- src/test/modules/test_shm_mq/test.c | 3 +- src/test/modules/worker_spi/worker_spi.c | 6 +-- 29 files changed, 142 insertions(+), 209 deletions(-) diff --git a/contrib/pg_prewarm/autoprewarm.c b/contrib/pg_prewarm/autoprewarm.c index cc5e2dd89cd..06eaa3b1010 100644 --- a/contrib/pg_prewarm/autoprewarm.c +++ b/contrib/pg_prewarm/autoprewarm.c @@ -220,7 +220,7 @@ autoprewarm_main(Datum main_arg) { /* We're only dumping at shutdown, so just wait forever. */ rc = WaitLatch(&MyProc->procLatch, - WL_LATCH_SET | WL_POSTMASTER_DEATH, + WL_LATCH_SET | WL_EXIT_ON_PM_DEATH, -1L, PG_WAIT_EXTENSION); } @@ -249,15 +249,13 @@ autoprewarm_main(Datum main_arg) /* Sleep until the next dump time. */ rc = WaitLatch(&MyProc->procLatch, - WL_LATCH_SET | WL_TIMEOUT | WL_POSTMASTER_DEATH, + WL_LATCH_SET | WL_TIMEOUT | WL_EXIT_ON_PM_DEATH, delay_in_ms, PG_WAIT_EXTENSION); } /* Reset the latch, bail out if postmaster died, otherwise loop. */ ResetLatch(&MyProc->procLatch); - if (rc & WL_POSTMASTER_DEATH) - proc_exit(1); } /* diff --git a/contrib/postgres_fdw/connection.c b/contrib/postgres_fdw/connection.c index fe4893a8e05..a6509932dcf 100644 --- a/contrib/postgres_fdw/connection.c +++ b/contrib/postgres_fdw/connection.c @@ -546,7 +546,8 @@ pgfdw_get_result(PGconn *conn, const char *query) /* Sleep until there's something to do */ wc = WaitLatchOrSocket(MyLatch, - WL_LATCH_SET | WL_SOCKET_READABLE, + WL_LATCH_SET | WL_SOCKET_READABLE | + WL_EXIT_ON_PM_DEATH, PQsocket(conn), -1L, PG_WAIT_EXTENSION); ResetLatch(MyLatch); @@ -1152,7 +1153,8 @@ pgfdw_get_cleanup_result(PGconn *conn, TimestampTz endtime, PGresult **result) /* Sleep until there's something to do */ wc = WaitLatchOrSocket(MyLatch, - WL_LATCH_SET | WL_SOCKET_READABLE | WL_TIMEOUT, + WL_LATCH_SET | WL_SOCKET_READABLE | + WL_TIMEOUT | WL_EXIT_ON_PM_DEATH, PQsocket(conn), cur_timeout, PG_WAIT_EXTENSION); ResetLatch(MyLatch); diff --git a/src/backend/access/transam/parallel.c b/src/backend/access/transam/parallel.c index c1681184670..09c5e3b5393 100644 --- a/src/backend/access/transam/parallel.c +++ b/src/backend/access/transam/parallel.c @@ -676,13 +676,9 @@ WaitForParallelWorkersToAttach(ParallelContext *pcxt) * just end up waiting for the same worker again. */ rc = WaitLatch(MyLatch, - WL_LATCH_SET | WL_POSTMASTER_DEATH, + WL_LATCH_SET | WL_EXIT_ON_PM_DEATH, -1, WAIT_EVENT_BGWORKER_STARTUP); - /* emergency bailout if postmaster has died */ - if (rc & WL_POSTMASTER_DEATH) - proc_exit(1); - if (rc & WL_LATCH_SET) ResetLatch(MyLatch); } @@ -799,7 +795,7 @@ WaitForParallelWorkersToFinish(ParallelContext *pcxt) } } - WaitLatch(MyLatch, WL_LATCH_SET, -1, + WaitLatch(MyLatch, WL_LATCH_SET | WL_EXIT_ON_PM_DEATH, -1, WAIT_EVENT_PARALLEL_FINISH); ResetLatch(MyLatch); } diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 65db2e48d88..31b45215d27 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -6208,7 +6208,7 @@ recoveryApplyDelay(XLogReaderState *record) secs, microsecs / 1000); WaitLatch(&XLogCtl->recoveryWakeupLatch, - WL_LATCH_SET | WL_TIMEOUT | WL_POSTMASTER_DEATH, + WL_LATCH_SET | WL_TIMEOUT | WL_EXIT_ON_PM_DEATH, secs * 1000L + microsecs / 1000, WAIT_EVENT_RECOVERY_APPLY_DELAY); } @@ -12113,7 +12113,7 @@ WaitForWALToBecomeAvailable(XLogRecPtr RecPtr, bool randAccess, (secs * 1000 + usecs / 1000); WaitLatch(&XLogCtl->recoveryWakeupLatch, - WL_LATCH_SET | WL_TIMEOUT | WL_POSTMASTER_DEATH, + WL_LATCH_SET | WL_TIMEOUT | WL_EXIT_ON_PM_DEATH, wait_time, WAIT_EVENT_RECOVERY_WAL_STREAM); ResetLatch(&XLogCtl->recoveryWakeupLatch); now = GetCurrentTimestamp(); @@ -12289,7 +12289,7 @@ WaitForWALToBecomeAvailable(XLogRecPtr RecPtr, bool randAccess, * to react to a trigger file promptly. */ WaitLatch(&XLogCtl->recoveryWakeupLatch, - WL_LATCH_SET | WL_TIMEOUT | WL_POSTMASTER_DEATH, + WL_LATCH_SET | WL_TIMEOUT | WL_EXIT_ON_PM_DEATH, 5000L, WAIT_EVENT_RECOVERY_WAL_ALL); ResetLatch(&XLogCtl->recoveryWakeupLatch); break; diff --git a/src/backend/executor/nodeGather.c b/src/backend/executor/nodeGather.c index 4a700b7b30e..b1459b4b872 100644 --- a/src/backend/executor/nodeGather.c +++ b/src/backend/executor/nodeGather.c @@ -364,7 +364,8 @@ gather_readnext(GatherState *gatherstate) return NULL; /* Nothing to do except wait for developments. */ - WaitLatch(MyLatch, WL_LATCH_SET, 0, WAIT_EVENT_EXECUTE_GATHER); + WaitLatch(MyLatch, WL_LATCH_SET | WL_EXIT_ON_PM_DEATH, 0, + WAIT_EVENT_EXECUTE_GATHER); ResetLatch(MyLatch); nvisited = 0; } diff --git a/src/backend/libpq/pqmq.c b/src/backend/libpq/pqmq.c index 4fbc6b5115d..bb788fbd565 100644 --- a/src/backend/libpq/pqmq.c +++ b/src/backend/libpq/pqmq.c @@ -168,7 +168,7 @@ mq_putmessage(char msgtype, const char *s, size_t len) if (result != SHM_MQ_WOULD_BLOCK) break; - WaitLatch(MyLatch, WL_LATCH_SET, 0, + WaitLatch(MyLatch, WL_LATCH_SET | WL_EXIT_ON_PM_DEATH, 0, WAIT_EVENT_MQ_PUT_MESSAGE); ResetLatch(MyLatch); CHECK_FOR_INTERRUPTS(); diff --git a/src/backend/postmaster/autovacuum.c b/src/backend/postmaster/autovacuum.c index 978089575b8..2d7c65b7cbb 100644 --- a/src/backend/postmaster/autovacuum.c +++ b/src/backend/postmaster/autovacuum.c @@ -628,7 +628,6 @@ AutoVacLauncherMain(int argc, char *argv[]) struct timeval nap; TimestampTz current_time = 0; bool can_launch; - int rc; /* * This loop is a bit different from the normal use of WaitLatch, @@ -644,23 +643,16 @@ AutoVacLauncherMain(int argc, char *argv[]) * Wait until naptime expires or we get some type of signal (all the * signal handlers will wake us by calling SetLatch). */ - rc = WaitLatch(MyLatch, - WL_LATCH_SET | WL_TIMEOUT | WL_POSTMASTER_DEATH, - (nap.tv_sec * 1000L) + (nap.tv_usec / 1000L), - WAIT_EVENT_AUTOVACUUM_MAIN); + WaitLatch(MyLatch, + WL_LATCH_SET | WL_TIMEOUT | WL_EXIT_ON_PM_DEATH, + (nap.tv_sec * 1000L) + (nap.tv_usec / 1000L), + WAIT_EVENT_AUTOVACUUM_MAIN); ResetLatch(MyLatch); /* Process sinval catchup interrupts that happened while sleeping */ ProcessCatchupInterrupt(); - /* - * Emergency bailout if postmaster has died. This is to avoid the - * necessity for manual cleanup of all postmaster children. - */ - if (rc & WL_POSTMASTER_DEATH) - proc_exit(1); - /* the normal shutdown case */ if (got_SIGTERM) break; diff --git a/src/backend/postmaster/bgwriter.c b/src/backend/postmaster/bgwriter.c index b1e9bb2c537..3da7bdac1de 100644 --- a/src/backend/postmaster/bgwriter.c +++ b/src/backend/postmaster/bgwriter.c @@ -339,7 +339,7 @@ BackgroundWriterMain(void) * normal operation. */ rc = WaitLatch(MyLatch, - WL_LATCH_SET | WL_TIMEOUT | WL_POSTMASTER_DEATH, + WL_LATCH_SET | WL_TIMEOUT | WL_EXIT_ON_PM_DEATH, BgWriterDelay /* ms */ , WAIT_EVENT_BGWRITER_MAIN); /* @@ -366,20 +366,13 @@ BackgroundWriterMain(void) StrategyNotifyBgWriter(MyProc->pgprocno); /* Sleep ... */ rc = WaitLatch(MyLatch, - WL_LATCH_SET | WL_TIMEOUT | WL_POSTMASTER_DEATH, + WL_LATCH_SET | WL_TIMEOUT | WL_EXIT_ON_PM_DEATH, BgWriterDelay * HIBERNATE_FACTOR, WAIT_EVENT_BGWRITER_HIBERNATE); /* Reset the notification request in case we timed out */ StrategyNotifyBgWriter(-1); } - /* - * Emergency bailout if postmaster has died. This is to avoid the - * necessity for manual cleanup of all postmaster children. - */ - if (rc & WL_POSTMASTER_DEATH) - exit(1); - prev_hibernate = can_hibernate; } } diff --git a/src/backend/postmaster/checkpointer.c b/src/backend/postmaster/checkpointer.c index 1a033093c53..8c2cdda3f5a 100644 --- a/src/backend/postmaster/checkpointer.c +++ b/src/backend/postmaster/checkpointer.c @@ -344,7 +344,6 @@ CheckpointerMain(void) pg_time_t now; int elapsed_secs; int cur_timeout; - int rc; /* Clear any already-pending wakeups */ ResetLatch(MyLatch); @@ -545,17 +544,10 @@ CheckpointerMain(void) cur_timeout = Min(cur_timeout, XLogArchiveTimeout - elapsed_secs); } - rc = WaitLatch(MyLatch, - WL_LATCH_SET | WL_TIMEOUT | WL_POSTMASTER_DEATH, - cur_timeout * 1000L /* convert to ms */ , - WAIT_EVENT_CHECKPOINTER_MAIN); - - /* - * Emergency bailout if postmaster has died. This is to avoid the - * necessity for manual cleanup of all postmaster children. - */ - if (rc & WL_POSTMASTER_DEATH) - exit(1); + WaitLatch(MyLatch, + WL_LATCH_SET | WL_TIMEOUT | WL_EXIT_ON_PM_DEATH, + cur_timeout * 1000L /* convert to ms */ , + WAIT_EVENT_CHECKPOINTER_MAIN); } } diff --git a/src/backend/postmaster/pgarch.c b/src/backend/postmaster/pgarch.c index 885e85ad8af..e43ac4d4f47 100644 --- a/src/backend/postmaster/pgarch.c +++ b/src/backend/postmaster/pgarch.c @@ -393,6 +393,8 @@ pgarch_MainLoop(void) WAIT_EVENT_ARCHIVER_MAIN); if (rc & WL_TIMEOUT) wakened = true; + if (rc & WL_POSTMASTER_DEATH) + time_to_stop = true; } else wakened = true; @@ -403,7 +405,7 @@ pgarch_MainLoop(void) * or after completing one more archiving cycle after receiving * SIGUSR2. */ - } while (PostmasterIsAlive() && !time_to_stop); + } while (!time_to_stop); } /* diff --git a/src/backend/postmaster/syslogger.c b/src/backend/postmaster/syslogger.c index 29bdcec8958..b0aa38f0ae8 100644 --- a/src/backend/postmaster/syslogger.c +++ b/src/backend/postmaster/syslogger.c @@ -168,6 +168,7 @@ SysLoggerMain(int argc, char *argv[]) char *currentLogFilename; int currentLogRotationAge; pg_time_t now; + WaitEventSet *wes; now = MyStartTime; @@ -298,6 +299,22 @@ SysLoggerMain(int argc, char *argv[]) */ whereToSendOutput = DestNone; + /* + * Set up a reusable WaitEventSet object we'll use to wait for our latch, + * and (except on Windows) our socket. + * + * Unlike all other postmaster child processes, we'll ignore postmaster + * death because we want to collect final log output from all backends and + * then exit last. We'll do that by running until we see EOF on the + * syslog pipe, which implies that all other backends have exited + * (including the postmaster). + */ + wes = CreateWaitEventSet(CurrentMemoryContext, 2); + AddWaitEventToSet(wes, WL_LATCH_SET, PGINVALID_SOCKET, MyLatch, NULL); +#ifndef WIN32 + AddWaitEventToSet(wes, WL_SOCKET_READABLE, syslogPipe[0], NULL, NULL); +#endif + /* main worker loop */ for (;;) { @@ -305,6 +322,7 @@ SysLoggerMain(int argc, char *argv[]) int size_rotation_for = 0; long cur_timeout; int cur_flags; + WaitEvent event; #ifndef WIN32 int rc; @@ -452,13 +470,10 @@ SysLoggerMain(int argc, char *argv[]) * Sleep until there's something to do */ #ifndef WIN32 - rc = WaitLatchOrSocket(MyLatch, - WL_LATCH_SET | WL_SOCKET_READABLE | cur_flags, - syslogPipe[0], - cur_timeout, - WAIT_EVENT_SYSLOGGER_MAIN); + rc = WaitEventSetWait(wes, cur_timeout, &event, 1, + WAIT_EVENT_SYSLOGGER_MAIN); - if (rc & WL_SOCKET_READABLE) + if (rc == 1 && event.events == WL_SOCKET_READABLE) { int bytesRead; @@ -505,10 +520,8 @@ SysLoggerMain(int argc, char *argv[]) */ LeaveCriticalSection(&sysloggerSection); - (void) WaitLatch(MyLatch, - WL_LATCH_SET | cur_flags, - cur_timeout, - WAIT_EVENT_SYSLOGGER_MAIN); + (void) WaitEventSetWait(wes, cur_timeout, &event, 1, + WAIT_EVENT_SYSLOGGER_MAIN); EnterCriticalSection(&sysloggerSection); #endif /* WIN32 */ diff --git a/src/backend/postmaster/walwriter.c b/src/backend/postmaster/walwriter.c index fb66bceeedf..f0e59dca7d1 100644 --- a/src/backend/postmaster/walwriter.c +++ b/src/backend/postmaster/walwriter.c @@ -227,7 +227,6 @@ WalWriterMain(void) for (;;) { long cur_timeout; - int rc; /* * Advertise whether we might hibernate in this cycle. We do this @@ -280,17 +279,10 @@ WalWriterMain(void) else cur_timeout = WalWriterDelay * HIBERNATE_FACTOR; - rc = WaitLatch(MyLatch, - WL_LATCH_SET | WL_TIMEOUT | WL_POSTMASTER_DEATH, - cur_timeout, - WAIT_EVENT_WAL_WRITER_MAIN); - - /* - * Emergency bailout if postmaster has died. This is to avoid the - * necessity for manual cleanup of all postmaster children. - */ - if (rc & WL_POSTMASTER_DEATH) - exit(1); + WaitLatch(MyLatch, + WL_LATCH_SET | WL_TIMEOUT | WL_EXIT_ON_PM_DEATH, + cur_timeout, + WAIT_EVENT_WAL_WRITER_MAIN); } } diff --git a/src/backend/replication/basebackup.c b/src/backend/replication/basebackup.c index 3f1eae38a92..4b671327fd8 100644 --- a/src/backend/replication/basebackup.c +++ b/src/backend/replication/basebackup.c @@ -1686,7 +1686,7 @@ throttle(size_t increment) * the maximum time to sleep. Thus the cast to long is safe. */ wait_result = WaitLatch(MyLatch, - WL_LATCH_SET | WL_TIMEOUT | WL_POSTMASTER_DEATH, + WL_LATCH_SET | WL_TIMEOUT | WL_EXIT_ON_PM_DEATH, (long) (sleep / 1000), WAIT_EVENT_BASE_BACKUP_THROTTLE); diff --git a/src/backend/replication/libpqwalreceiver/libpqwalreceiver.c b/src/backend/replication/libpqwalreceiver/libpqwalreceiver.c index 1e1695ef4f4..310bf04d371 100644 --- a/src/backend/replication/libpqwalreceiver/libpqwalreceiver.c +++ b/src/backend/replication/libpqwalreceiver/libpqwalreceiver.c @@ -186,16 +186,11 @@ libpqrcv_connect(const char *conninfo, bool logical, const char *appname, io_flag = WL_SOCKET_WRITEABLE; rc = WaitLatchOrSocket(MyLatch, - WL_POSTMASTER_DEATH | - WL_LATCH_SET | io_flag, + WL_EXIT_ON_PM_DEATH | WL_LATCH_SET | io_flag, PQsocket(conn->streamConn), 0, WAIT_EVENT_LIBPQWALRECEIVER_CONNECT); - /* Emergency bailout? */ - if (rc & WL_POSTMASTER_DEATH) - exit(1); - /* Interrupted? */ if (rc & WL_LATCH_SET) { @@ -610,16 +605,12 @@ libpqrcv_PQexec(PGconn *streamConn, const char *query) * replication connection. */ rc = WaitLatchOrSocket(MyLatch, - WL_POSTMASTER_DEATH | WL_SOCKET_READABLE | + WL_EXIT_ON_PM_DEATH | WL_SOCKET_READABLE | WL_LATCH_SET, PQsocket(streamConn), 0, WAIT_EVENT_LIBPQWALRECEIVER_RECEIVE); - /* Emergency bailout? */ - if (rc & WL_POSTMASTER_DEATH) - exit(1); - /* Interrupted? */ if (rc & WL_LATCH_SET) { diff --git a/src/backend/replication/logical/launcher.c b/src/backend/replication/logical/launcher.c index ada16adb67b..4833ff9d48c 100644 --- a/src/backend/replication/logical/launcher.c +++ b/src/backend/replication/logical/launcher.c @@ -221,13 +221,9 @@ WaitForReplicationWorkerAttach(LogicalRepWorker *worker, * about the worker attach. But we don't expect to have to wait long. */ rc = WaitLatch(MyLatch, - WL_LATCH_SET | WL_TIMEOUT | WL_POSTMASTER_DEATH, + WL_LATCH_SET | WL_TIMEOUT | WL_EXIT_ON_PM_DEATH, 10L, WAIT_EVENT_BGWORKER_STARTUP); - /* emergency bailout if postmaster has died */ - if (rc & WL_POSTMASTER_DEATH) - proc_exit(1); - if (rc & WL_LATCH_SET) { ResetLatch(MyLatch); @@ -498,13 +494,9 @@ logicalrep_worker_stop(Oid subid, Oid relid) /* Wait a bit --- we don't expect to have to wait long. */ rc = WaitLatch(MyLatch, - WL_LATCH_SET | WL_TIMEOUT | WL_POSTMASTER_DEATH, + WL_LATCH_SET | WL_TIMEOUT | WL_EXIT_ON_PM_DEATH, 10L, WAIT_EVENT_BGWORKER_STARTUP); - /* emergency bailout if postmaster has died */ - if (rc & WL_POSTMASTER_DEATH) - proc_exit(1); - if (rc & WL_LATCH_SET) { ResetLatch(MyLatch); @@ -546,13 +538,9 @@ logicalrep_worker_stop(Oid subid, Oid relid) /* Wait a bit --- we don't expect to have to wait long. */ rc = WaitLatch(MyLatch, - WL_LATCH_SET | WL_TIMEOUT | WL_POSTMASTER_DEATH, + WL_LATCH_SET | WL_TIMEOUT | WL_EXIT_ON_PM_DEATH, 10L, WAIT_EVENT_BGWORKER_SHUTDOWN); - /* emergency bailout if postmaster has died */ - if (rc & WL_POSTMASTER_DEATH) - proc_exit(1); - if (rc & WL_LATCH_SET) { ResetLatch(MyLatch); @@ -1072,14 +1060,10 @@ ApplyLauncherMain(Datum main_arg) /* Wait for more work. */ rc = WaitLatch(MyLatch, - WL_LATCH_SET | WL_TIMEOUT | WL_POSTMASTER_DEATH, + WL_LATCH_SET | WL_TIMEOUT | WL_EXIT_ON_PM_DEATH, wait_time, WAIT_EVENT_LOGICAL_LAUNCHER_MAIN); - /* emergency bailout if postmaster has died */ - if (rc & WL_POSTMASTER_DEATH) - proc_exit(1); - if (rc & WL_LATCH_SET) { ResetLatch(MyLatch); diff --git a/src/backend/replication/logical/tablesync.c b/src/backend/replication/logical/tablesync.c index acc6498567d..159234ef3c7 100644 --- a/src/backend/replication/logical/tablesync.c +++ b/src/backend/replication/logical/tablesync.c @@ -159,7 +159,6 @@ finish_sync_worker(void) static bool wait_for_relation_state_change(Oid relid, char expected_state) { - int rc; char state; for (;;) @@ -192,13 +191,9 @@ wait_for_relation_state_change(Oid relid, char expected_state) if (!worker) return false; - rc = WaitLatch(MyLatch, - WL_LATCH_SET | WL_TIMEOUT | WL_POSTMASTER_DEATH, - 1000L, WAIT_EVENT_LOGICAL_SYNC_STATE_CHANGE); - - /* emergency bailout if postmaster has died */ - if (rc & WL_POSTMASTER_DEATH) - proc_exit(1); + WaitLatch(MyLatch, + WL_LATCH_SET | WL_TIMEOUT | WL_EXIT_ON_PM_DEATH, + 1000L, WAIT_EVENT_LOGICAL_SYNC_STATE_CHANGE); ResetLatch(MyLatch); } @@ -250,13 +245,9 @@ wait_for_worker_state_change(char expected_state) * but use a timeout in case it dies without sending one. */ rc = WaitLatch(MyLatch, - WL_LATCH_SET | WL_TIMEOUT | WL_POSTMASTER_DEATH, + WL_LATCH_SET | WL_TIMEOUT | WL_EXIT_ON_PM_DEATH, 1000L, WAIT_EVENT_LOGICAL_SYNC_STATE_CHANGE); - /* emergency bailout if postmaster has died */ - if (rc & WL_POSTMASTER_DEATH) - proc_exit(1); - if (rc & WL_LATCH_SET) ResetLatch(MyLatch); } @@ -593,7 +584,6 @@ copy_read_data(void *outbuf, int minread, int maxread) while (maxread > 0 && bytesread < minread) { pgsocket fd = PGINVALID_SOCKET; - int rc; int len; char *buf = NULL; @@ -632,14 +622,10 @@ copy_read_data(void *outbuf, int minread, int maxread) /* * Wait for more data or latch. */ - rc = WaitLatchOrSocket(MyLatch, - WL_SOCKET_READABLE | WL_LATCH_SET | - WL_TIMEOUT | WL_POSTMASTER_DEATH, - fd, 1000L, WAIT_EVENT_LOGICAL_SYNC_DATA); - - /* Emergency bailout if postmaster has died */ - if (rc & WL_POSTMASTER_DEATH) - proc_exit(1); + WaitLatchOrSocket(MyLatch, + WL_SOCKET_READABLE | WL_LATCH_SET | + WL_TIMEOUT | WL_EXIT_ON_PM_DEATH, + fd, 1000L, WAIT_EVENT_LOGICAL_SYNC_DATA); ResetLatch(MyLatch); } diff --git a/src/backend/replication/logical/worker.c b/src/backend/replication/logical/worker.c index 2054abe6532..41097647443 100644 --- a/src/backend/replication/logical/worker.c +++ b/src/backend/replication/logical/worker.c @@ -1257,14 +1257,10 @@ LogicalRepApplyLoop(XLogRecPtr last_received) rc = WaitLatchOrSocket(MyLatch, WL_SOCKET_READABLE | WL_LATCH_SET | - WL_TIMEOUT | WL_POSTMASTER_DEATH, + WL_TIMEOUT | WL_EXIT_ON_PM_DEATH, fd, wait_time, WAIT_EVENT_LOGICAL_APPLY_MAIN); - /* Emergency bailout if postmaster has died */ - if (rc & WL_POSTMASTER_DEATH) - proc_exit(1); - if (rc & WL_LATCH_SET) { ResetLatch(MyLatch); diff --git a/src/backend/replication/syncrep.c b/src/backend/replication/syncrep.c index 75d26817192..7fc784f5916 100644 --- a/src/backend/replication/syncrep.c +++ b/src/backend/replication/syncrep.c @@ -214,6 +214,8 @@ SyncRepWaitForLSN(XLogRecPtr lsn, bool commit) */ for (;;) { + int rc; + /* Must reset the latch before testing state. */ ResetLatch(MyLatch); @@ -266,25 +268,25 @@ SyncRepWaitForLSN(XLogRecPtr lsn, bool commit) break; } + /* + * Wait on latch. Any condition that should wake us up will set the + * latch, so no need for timeout. + */ + rc = WaitLatch(MyLatch, WL_LATCH_SET | WL_POSTMASTER_DEATH, -1, + WAIT_EVENT_SYNC_REP); + /* * If the postmaster dies, we'll probably never get an * acknowledgement, because all the wal sender processes will exit. So * just bail out. */ - if (!PostmasterIsAlive()) + if (rc & WL_POSTMASTER_DEATH) { ProcDiePending = true; whereToSendOutput = DestNone; SyncRepCancelWait(); break; } - - /* - * Wait on latch. Any condition that should wake us up will set the - * latch, so no need for timeout. - */ - WaitLatch(MyLatch, WL_LATCH_SET | WL_POSTMASTER_DEATH, -1, - WAIT_EVENT_SYNC_REP); } /* diff --git a/src/backend/replication/walreceiver.c b/src/backend/replication/walreceiver.c index 6f4b3538ac4..31c5de6f6ac 100644 --- a/src/backend/replication/walreceiver.c +++ b/src/backend/replication/walreceiver.c @@ -507,7 +507,7 @@ WalReceiverMain(void) */ Assert(wait_fd != PGINVALID_SOCKET); rc = WaitLatchOrSocket(walrcv->latch, - WL_POSTMASTER_DEATH | WL_SOCKET_READABLE | + WL_EXIT_ON_PM_DEATH | WL_SOCKET_READABLE | WL_TIMEOUT | WL_LATCH_SET, wait_fd, NAPTIME_PER_CYCLE, @@ -528,15 +528,6 @@ WalReceiverMain(void) XLogWalRcvSendReply(true, false); } } - if (rc & WL_POSTMASTER_DEATH) - { - /* - * Emergency bailout if postmaster has died. This is to - * avoid the necessity for manual cleanup of all - * postmaster children. - */ - exit(1); - } if (rc & WL_TIMEOUT) { /* @@ -677,13 +668,6 @@ WalRcvWaitForStartPosition(XLogRecPtr *startpoint, TimeLineID *startpointTLI) { ResetLatch(walrcv->latch); - /* - * Emergency bailout if postmaster has died. This is to avoid the - * necessity for manual cleanup of all postmaster children. - */ - if (!PostmasterIsAlive()) - exit(1); - ProcessWalRcvInterrupts(); SpinLockAcquire(&walrcv->mutex); @@ -710,7 +694,7 @@ WalRcvWaitForStartPosition(XLogRecPtr *startpoint, TimeLineID *startpointTLI) } SpinLockRelease(&walrcv->mutex); - WaitLatch(walrcv->latch, WL_LATCH_SET | WL_POSTMASTER_DEATH, 0, + WaitLatch(walrcv->latch, WL_LATCH_SET | WL_EXIT_ON_PM_DEATH, 0, WAIT_EVENT_WAL_RECEIVER_WAIT_START); } diff --git a/src/backend/replication/walsender.c b/src/backend/replication/walsender.c index 3e51cf3d312..95f30476071 100644 --- a/src/backend/replication/walsender.c +++ b/src/backend/replication/walsender.c @@ -1216,7 +1216,7 @@ WalSndWriteData(LogicalDecodingContext *ctx, XLogRecPtr lsn, TransactionId xid, sleeptime = WalSndComputeSleeptime(GetCurrentTimestamp()); - wakeEvents = WL_LATCH_SET | WL_POSTMASTER_DEATH | + wakeEvents = WL_LATCH_SET | WL_EXIT_ON_PM_DEATH | WL_SOCKET_WRITEABLE | WL_SOCKET_READABLE | WL_TIMEOUT; /* Sleep until something happens or we time out */ @@ -1224,13 +1224,6 @@ WalSndWriteData(LogicalDecodingContext *ctx, XLogRecPtr lsn, TransactionId xid, MyProcPort->sock, sleeptime, WAIT_EVENT_WAL_SENDER_WRITE_DATA); - /* - * Emergency bailout if postmaster has died. This is to avoid the - * necessity for manual cleanup of all postmaster children. - */ - if (!PostmasterIsAlive()) - exit(1); - /* Clear any already-pending wakeups */ ResetLatch(MyLatch); @@ -1310,13 +1303,6 @@ WalSndWaitForWal(XLogRecPtr loc) { long sleeptime; - /* - * Emergency bailout if postmaster has died. This is to avoid the - * necessity for manual cleanup of all postmaster children. - */ - if (!PostmasterIsAlive()) - exit(1); - /* Clear any already-pending wakeups */ ResetLatch(MyLatch); @@ -1408,8 +1394,8 @@ WalSndWaitForWal(XLogRecPtr loc) */ sleeptime = WalSndComputeSleeptime(GetCurrentTimestamp()); - wakeEvents = WL_LATCH_SET | WL_POSTMASTER_DEATH | - WL_SOCKET_READABLE | WL_TIMEOUT; + wakeEvents = WL_LATCH_SET | WL_EXIT_ON_PM_DEATH | + WL_SOCKET_READABLE | WL_TIMEOUT; if (pq_is_send_pending()) wakeEvents |= WL_SOCKET_WRITEABLE; @@ -2124,13 +2110,6 @@ WalSndLoop(WalSndSendDataCallback send_data) */ for (;;) { - /* - * Emergency bailout if postmaster has died. This is to avoid the - * necessity for manual cleanup of all postmaster children. - */ - if (!PostmasterIsAlive()) - exit(1); - /* Clear any already-pending wakeups */ ResetLatch(MyLatch); @@ -2220,8 +2199,8 @@ WalSndLoop(WalSndSendDataCallback send_data) long sleeptime; int wakeEvents; - wakeEvents = WL_LATCH_SET | WL_POSTMASTER_DEATH | WL_TIMEOUT | - WL_SOCKET_READABLE; + wakeEvents = WL_LATCH_SET | WL_EXIT_ON_PM_DEATH | WL_TIMEOUT | + WL_SOCKET_READABLE; /* * Use fresh timestamp, not last_processed, to reduce the chance diff --git a/src/backend/storage/ipc/latch.c b/src/backend/storage/ipc/latch.c index f6dda9cc9ac..7a76493269a 100644 --- a/src/backend/storage/ipc/latch.c +++ b/src/backend/storage/ipc/latch.c @@ -92,6 +92,13 @@ struct WaitEventSet Latch *latch; int latch_pos; + /* + * WL_EXIT_ON_PM_DEATH is converted to WL_POSTMASTER_DEATH, but this flag + * is set so that we'll exit immediately if postmaster death is detected, + * instead of returning. + */ + bool exit_on_postmaster_death; + #if defined(WAIT_USE_EPOLL) int epoll_fd; /* epoll_wait returns events in a user provided arrays, allocate once */ @@ -370,10 +377,19 @@ WaitLatchOrSocket(volatile Latch *latch, int wakeEvents, pgsocket sock, AddWaitEventToSet(set, WL_LATCH_SET, PGINVALID_SOCKET, (Latch *) latch, NULL); + /* Postmaster-managed callers must handle postmaster death somehow. */ + Assert(!IsUnderPostmaster || + (wakeEvents & (WL_EXIT_ON_PM_DEATH)) != 0 || + (wakeEvents & (WL_POSTMASTER_DEATH)) != 0); + if (wakeEvents & WL_POSTMASTER_DEATH && IsUnderPostmaster) AddWaitEventToSet(set, WL_POSTMASTER_DEATH, PGINVALID_SOCKET, NULL, NULL); + if (wakeEvents & WL_EXIT_ON_PM_DEATH && IsUnderPostmaster) + AddWaitEventToSet(set, WL_EXIT_ON_PM_DEATH, PGINVALID_SOCKET, + NULL, NULL); + if (wakeEvents & WL_SOCKET_MASK) { int ev; @@ -562,6 +578,7 @@ CreateWaitEventSet(MemoryContext context, int nevents) set->latch = NULL; set->nevents_space = nevents; + set->exit_on_postmaster_death = false; #if defined(WAIT_USE_EPOLL) #ifdef EPOLL_CLOEXEC @@ -646,6 +663,7 @@ FreeWaitEventSet(WaitEventSet *set) * - WL_SOCKET_CONNECTED: Wait for socket connection to be established, * can be combined with other WL_SOCKET_* events (on non-Windows * platforms, this is the same as WL_SOCKET_WRITEABLE) + * - WL_EXIT_ON_PM_DEATH: Exit immediately if the postmaster dies * * Returns the offset in WaitEventSet->events (starting from 0), which can be * used to modify previously added wait events using ModifyWaitEvent(). @@ -671,6 +689,12 @@ AddWaitEventToSet(WaitEventSet *set, uint32 events, pgsocket fd, Latch *latch, /* not enough space */ Assert(set->nevents < set->nevents_space); + if (events == WL_EXIT_ON_PM_DEATH) + { + events = WL_POSTMASTER_DEATH; + set->exit_on_postmaster_death = true; + } + if (latch) { if (latch->owner_pid != MyProcPid) @@ -926,6 +950,7 @@ WaitEventSetWait(WaitEventSet *set, long timeout, instr_time start_time; instr_time cur_time; long cur_timeout = -1; + int i; Assert(nevents > 0); @@ -986,7 +1011,6 @@ WaitEventSetWait(WaitEventSet *set, long timeout, occurred_events->user_data = set->events[set->latch_pos].user_data; occurred_events->events = WL_LATCH_SET; - occurred_events++; returned_events++; break; @@ -1021,6 +1045,19 @@ WaitEventSetWait(WaitEventSet *set, long timeout, pgstat_report_wait_end(); + /* + * Exit immediately if the postmaster died and the caller asked for + * automatic exit in that case. + */ + if (set->exit_on_postmaster_death) + { + for (i = 0; i < returned_events; ++i) + { + if (occurred_events[i].events == WL_POSTMASTER_DEATH) + exit(1); + } + } + return returned_events; } diff --git a/src/backend/storage/ipc/shm_mq.c b/src/backend/storage/ipc/shm_mq.c index fde71afd479..0b9ecf81451 100644 --- a/src/backend/storage/ipc/shm_mq.c +++ b/src/backend/storage/ipc/shm_mq.c @@ -949,7 +949,8 @@ shm_mq_send_bytes(shm_mq_handle *mqh, Size nbytes, const void *data, * at top of loop, because setting an already-set latch is much * cheaper than setting one that has been reset. */ - WaitLatch(MyLatch, WL_LATCH_SET, 0, WAIT_EVENT_MQ_SEND); + WaitLatch(MyLatch, WL_LATCH_SET | WL_EXIT_ON_PM_DEATH, 0, + WAIT_EVENT_MQ_SEND); /* Reset the latch so we don't spin. */ ResetLatch(MyLatch); @@ -1093,7 +1094,8 @@ shm_mq_receive_bytes(shm_mq_handle *mqh, Size bytes_needed, bool nowait, * loop, because setting an already-set latch is much cheaper than * setting one that has been reset. */ - WaitLatch(MyLatch, WL_LATCH_SET, 0, WAIT_EVENT_MQ_RECEIVE); + WaitLatch(MyLatch, WL_LATCH_SET | WL_EXIT_ON_PM_DEATH, 0, + WAIT_EVENT_MQ_RECEIVE); /* Reset the latch so we don't spin. */ ResetLatch(MyLatch); @@ -1181,7 +1183,8 @@ shm_mq_wait_internal(shm_mq *mq, PGPROC **ptr, BackgroundWorkerHandle *handle) } /* Wait to be signalled. */ - WaitLatch(MyLatch, WL_LATCH_SET, 0, WAIT_EVENT_MQ_INTERNAL); + WaitLatch(MyLatch, WL_LATCH_SET | WL_EXIT_ON_PM_DEATH, 0, + WAIT_EVENT_MQ_INTERNAL); /* Reset the latch so we don't spin. */ ResetLatch(MyLatch); diff --git a/src/backend/storage/lmgr/condition_variable.c b/src/backend/storage/lmgr/condition_variable.c index ef1d5baf016..851c09ce7ad 100644 --- a/src/backend/storage/lmgr/condition_variable.c +++ b/src/backend/storage/lmgr/condition_variable.c @@ -72,7 +72,7 @@ ConditionVariablePrepareToSleep(ConditionVariable *cv) new_event_set = CreateWaitEventSet(TopMemoryContext, 2); AddWaitEventToSet(new_event_set, WL_LATCH_SET, PGINVALID_SOCKET, MyLatch, NULL); - AddWaitEventToSet(new_event_set, WL_POSTMASTER_DEATH, PGINVALID_SOCKET, + AddWaitEventToSet(new_event_set, WL_EXIT_ON_PM_DEATH, PGINVALID_SOCKET, NULL, NULL); /* Don't set cv_wait_event_set until we have a correct WES. */ cv_wait_event_set = new_event_set; @@ -156,15 +156,6 @@ ConditionVariableSleep(ConditionVariable *cv, uint32 wait_event_info) */ WaitEventSetWait(cv_wait_event_set, -1, &event, 1, wait_event_info); - if (event.events & WL_POSTMASTER_DEATH) - { - /* - * Emergency bailout if postmaster has died. This is to avoid the - * necessity for manual cleanup of all postmaster children. - */ - exit(1); - } - /* Reset latch before examining the state of the wait list. */ ResetLatch(MyLatch); diff --git a/src/backend/storage/lmgr/proc.c b/src/backend/storage/lmgr/proc.c index 6f9aaa52faf..a9c6fd9383c 100644 --- a/src/backend/storage/lmgr/proc.c +++ b/src/backend/storage/lmgr/proc.c @@ -1263,7 +1263,7 @@ ProcSleep(LOCALLOCK *locallock, LockMethod lockMethodTable) } else { - WaitLatch(MyLatch, WL_LATCH_SET, 0, + WaitLatch(MyLatch, WL_LATCH_SET | WL_EXIT_ON_PM_DEATH, 0, PG_WAIT_LOCK | locallock->tag.lock.locktag_type); ResetLatch(MyLatch); /* check for deadlocks first, as that's probably log-worthy */ @@ -1776,7 +1776,7 @@ CheckDeadLockAlert(void) void ProcWaitForSignal(uint32 wait_event_info) { - WaitLatch(MyLatch, WL_LATCH_SET, 0, wait_event_info); + WaitLatch(MyLatch, WL_LATCH_SET | WL_EXIT_ON_PM_DEATH, 0, wait_event_info); ResetLatch(MyLatch); CHECK_FOR_INTERRUPTS(); } diff --git a/src/backend/utils/adt/misc.c b/src/backend/utils/adt/misc.c index b24dece23f8..81c72d8bf80 100644 --- a/src/backend/utils/adt/misc.c +++ b/src/backend/utils/adt/misc.c @@ -573,7 +573,7 @@ pg_sleep(PG_FUNCTION_ARGS) break; (void) WaitLatch(MyLatch, - WL_LATCH_SET | WL_TIMEOUT, + WL_LATCH_SET | WL_TIMEOUT | WL_EXIT_ON_PM_DEATH, delay_ms, WAIT_EVENT_PG_SLEEP); ResetLatch(MyLatch); diff --git a/src/include/storage/latch.h b/src/include/storage/latch.h index fd8735b7f5f..039a82e1a3a 100644 --- a/src/include/storage/latch.h +++ b/src/include/storage/latch.h @@ -126,8 +126,9 @@ typedef struct Latch #define WL_SOCKET_WRITEABLE (1 << 2) #define WL_TIMEOUT (1 << 3) /* not for WaitEventSetWait() */ #define WL_POSTMASTER_DEATH (1 << 4) +#define WL_EXIT_ON_PM_DEATH (1 << 5) #ifdef WIN32 -#define WL_SOCKET_CONNECTED (1 << 5) +#define WL_SOCKET_CONNECTED (1 << 6) #else /* avoid having to deal with case on platforms not requiring it */ #define WL_SOCKET_CONNECTED WL_SOCKET_WRITEABLE diff --git a/src/test/modules/test_shm_mq/setup.c b/src/test/modules/test_shm_mq/setup.c index 97e8617b3e9..af88110885d 100644 --- a/src/test/modules/test_shm_mq/setup.c +++ b/src/test/modules/test_shm_mq/setup.c @@ -280,7 +280,8 @@ wait_for_workers_to_become_ready(worker_state *wstate, } /* Wait to be signalled. */ - WaitLatch(MyLatch, WL_LATCH_SET, 0, PG_WAIT_EXTENSION); + WaitLatch(MyLatch, WL_LATCH_SET | WL_EXIT_ON_PM_DEATH, 0, + PG_WAIT_EXTENSION); /* Reset the latch so we don't spin. */ ResetLatch(MyLatch); diff --git a/src/test/modules/test_shm_mq/test.c b/src/test/modules/test_shm_mq/test.c index ebab9866017..b9621c294d1 100644 --- a/src/test/modules/test_shm_mq/test.c +++ b/src/test/modules/test_shm_mq/test.c @@ -231,7 +231,8 @@ test_shm_mq_pipelined(PG_FUNCTION_ARGS) * have read or written data and therefore there may now be work * for us to do. */ - WaitLatch(MyLatch, WL_LATCH_SET, 0, PG_WAIT_EXTENSION); + WaitLatch(MyLatch, WL_LATCH_SET | WL_EXIT_ON_PM_DEATH, 0, + PG_WAIT_EXTENSION); ResetLatch(MyLatch); CHECK_FOR_INTERRUPTS(); } diff --git a/src/test/modules/worker_spi/worker_spi.c b/src/test/modules/worker_spi/worker_spi.c index 0d705a3f2ed..52a3208527e 100644 --- a/src/test/modules/worker_spi/worker_spi.c +++ b/src/test/modules/worker_spi/worker_spi.c @@ -226,15 +226,11 @@ worker_spi_main(Datum main_arg) * background process goes away immediately in an emergency. */ rc = WaitLatch(MyLatch, - WL_LATCH_SET | WL_TIMEOUT | WL_POSTMASTER_DEATH, + WL_LATCH_SET | WL_TIMEOUT | WL_EXIT_ON_PM_DEATH, worker_spi_naptime * 1000L, PG_WAIT_EXTENSION); ResetLatch(MyLatch); - /* emergency bailout if postmaster has died */ - if (rc & WL_POSTMASTER_DEATH) - proc_exit(1); - CHECK_FOR_INTERRUPTS(); /* -- 2.17.0