parallel.c File Reference

#include "postgres_fe.h"
#include <sys/select.h>
#include <sys/wait.h>
#include <signal.h>
#include <unistd.h>
#include <fcntl.h>
#include "fe_utils/string_utils.h"
#include "parallel.h"
#include "pg_backup_utils.h"
#include "port/pg_bswap.h"

Include dependency graph for parallel.c:

Go to the source code of this file.

Data Structures
struct	ParallelSlot
struct	ShutdownInformation
struct	DumpSignalInformation

Macros
#define	PIPE_READ   0
#define	PIPE_WRITE   1
#define	NO_SLOT   (-1) /* Failure result for GetIdleWorker() */
#define	WORKER_IS_RUNNING(workerStatus)    ((workerStatus) == WRKR_IDLE || (workerStatus) == WRKR_WORKING)
#define	pgpipe(a)   pipe(a)
#define	piperead(a, b, c)   read(a,b,c)
#define	pipewrite(a, b, c)   write(a,b,c)
#define	write_stderr(str)
#define	messageStartsWith(msg, prefix)    (strncmp(msg, prefix, strlen(prefix)) == 0)

Typedefs
typedef struct ShutdownInformation	ShutdownInformation
typedef struct DumpSignalInformation	DumpSignalInformation

Enumerations
enum	T_WorkerStatus { WRKR_NOT_STARTED = 0 , WRKR_IDLE , WRKR_WORKING , WRKR_TERMINATED }

Functions
static ParallelSlot *	GetMyPSlot (ParallelState *pstate)
static void	archive_close_connection (int code, void *arg)
static void	ShutdownWorkersHard (ParallelState *pstate)
static void	WaitForTerminatingWorkers (ParallelState *pstate)
static void	setup_cancel_handler (void)
static void	set_cancel_pstate (ParallelState *pstate)
static void	set_cancel_slot_archive (ParallelSlot *slot, ArchiveHandle *AH)
static void	RunWorker (ArchiveHandle *AH, ParallelSlot *slot)
static int	GetIdleWorker (ParallelState *pstate)
static bool	HasEveryWorkerTerminated (ParallelState *pstate)
static void	lockTableForWorker (ArchiveHandle *AH, TocEntry *te)
static void	WaitForCommands (ArchiveHandle *AH, int pipefd[2])
static bool	ListenToWorkers (ArchiveHandle *AH, ParallelState *pstate, bool do_wait)
static char *	getMessageFromLeader (int pipefd[2])
static void	sendMessageToLeader (int pipefd[2], const char *str)
static int	select_loop (int maxFd, fd_set *workerset)
static char *	getMessageFromWorker (ParallelState *pstate, bool do_wait, int *worker)
static void	sendMessageToWorker (ParallelState *pstate, int worker, const char *str)
static char *	readMessageFromPipe (int fd)
void	init_parallel_dump_utils (void)
void	on_exit_close_archive (Archive *AHX)
static void	sigTermHandler (SIGNAL_ARGS)
void	set_archive_cancel_info (ArchiveHandle *AH, PGconn *conn)
ParallelState *	ParallelBackupStart (ArchiveHandle *AH)
void	ParallelBackupEnd (ArchiveHandle *AH, ParallelState *pstate)
static void	buildWorkerCommand (ArchiveHandle *AH, TocEntry *te, T_Action act, char *buf, int buflen)
static void	parseWorkerCommand (ArchiveHandle *AH, TocEntry **te, T_Action *act, const char *msg)
static void	buildWorkerResponse (ArchiveHandle *AH, TocEntry *te, T_Action act, int status, char *buf, int buflen)
static int	parseWorkerResponse (ArchiveHandle *AH, TocEntry *te, const char *msg)
void	DispatchJobForTocEntry (ArchiveHandle *AH, ParallelState *pstate, TocEntry *te, T_Action act, ParallelCompletionPtr callback, void *callback_data)
bool	IsEveryWorkerIdle (ParallelState *pstate)
void	WaitForWorkers (ArchiveHandle *AH, ParallelState *pstate, WFW_WaitOption mode)

Variables
static ShutdownInformation	shutdown_info
static volatile DumpSignalInformation	signal_info

Macro Definition Documentation

messageStartsWith

#define messageStartsWith	(		msg,
			prefix
	)		(strncmp(msg, prefix, strlen(prefix)) == 0)

Definition at line 226 of file parallel.c.

NO_SLOT

#define NO_SLOT   (-1) /* Failure result for GetIdleWorker() */

Definition at line 72 of file parallel.c.

pgpipe

#define pgpipe

(

a

)

pipe(a)

Definition at line 137 of file parallel.c.

PIPE_READ

#define PIPE_READ   0

Definition at line 69 of file parallel.c.

PIPE_WRITE

#define PIPE_WRITE   1

Definition at line 70 of file parallel.c.

piperead

#define piperead	(		a,
			b,
			c
	)		read(a,b,c)

Definition at line 138 of file parallel.c.

pipewrite

#define pipewrite	(		a,
			b,
			c
	)		write(a,b,c)

Definition at line 139 of file parallel.c.

WORKER_IS_RUNNING

#define WORKER_IS_RUNNING

(

workerStatus

)

((workerStatus) == WRKR_IDLE || (workerStatus) == WRKR_WORKING)

Definition at line 83 of file parallel.c.

write_stderr

#define write_stderr

(

str

)

Value:

    do { \
        const char *str_ = (str); \
        int     rc_; \
        rc_ = write(fileno(stderr), str_, strlen(str_)); \
        (void) rc_; \
    } while (0)

Definition at line 184 of file parallel.c.

Typedef Documentation

DumpSignalInformation

typedef struct DumpSignalInformation DumpSignalInformation

ShutdownInformation

typedef struct ShutdownInformation ShutdownInformation

Enumeration Type Documentation

T_WorkerStatus

enum T_WorkerStatus

Enumerator
WRKR_NOT_STARTED
WRKR_IDLE
WRKR_WORKING
WRKR_TERMINATED

Definition at line 75 of file parallel.c.

{
    WRKR_NOT_STARTED = 0,
    WRKR_IDLE,
    WRKR_WORKING,
    WRKR_TERMINATED
} T_WorkerStatus;

Function Documentation

archive_close_connection()

static void archive_close_connection ( int  code, void *  arg  )

static

Definition at line 339 of file parallel.c.

{
    ShutdownInformation *si = (ShutdownInformation *) arg;
 
    if (si->pstate)
    {
        /* In parallel mode, must figure out who we are */
        ParallelSlot *slot = GetMyPSlot(si->pstate);
 
        if (!slot)
        {
            /*
             * We're the leader.  Forcibly shut down workers, then close our
             * own database connection, if any.
             */
            ShutdownWorkersHard(si->pstate);
 
            if (si->AHX)
                DisconnectDatabase(si->AHX);
        }
        else
        {
            /*
             * We're a worker.  Shut down our own DB connection if any.  On
             * Windows, we also have to close our communication sockets, to
             * emulate what will happen on Unix when the worker process exits.
             * (Without this, if this is a premature exit, the leader would
             * fail to detect it because there would be no EOF condition on
             * the other end of the pipe.)
             */
            if (slot->AH)
                DisconnectDatabase(&(slot->AH->public));
 
#ifdef WIN32
            closesocket(slot->pipeRevRead);
            closesocket(slot->pipeRevWrite);
#endif
        }
    }
    else
    {
        /* Non-parallel operation: just kill the leader DB connection */
        if (si->AHX)
            DisconnectDatabase(si->AHX);
    }
}

References ParallelSlot::AH, ShutdownInformation::AHX, arg, closesocket, DisconnectDatabase(), GetMyPSlot(), ParallelSlot::pipeRevRead, ParallelSlot::pipeRevWrite, ShutdownInformation::pstate, _archiveHandle::public, and ShutdownWorkersHard().

Referenced by on_exit_close_archive().

buildWorkerCommand()

static void buildWorkerCommand ( ArchiveHandle *  AH, TocEntry *  te, T_Action  act, char *  buf, int  buflen  )

static

Definition at line 1108 of file parallel.c.

{
    if (act == ACT_DUMP)
        snprintf(buf, buflen, "DUMP %d", te->dumpId);
    else if (act == ACT_RESTORE)
        snprintf(buf, buflen, "RESTORE %d", te->dumpId);
    else
        Assert(false);
}

References ACT_DUMP, ACT_RESTORE, Assert(), buf, _tocEntry::dumpId, and snprintf.

Referenced by DispatchJobForTocEntry().

buildWorkerResponse()

static void buildWorkerResponse ( ArchiveHandle *  AH, TocEntry *  te, T_Action  act, int  status, char *  buf, int  buflen  )

static

Definition at line 1156 of file parallel.c.

{
    snprintf(buf, buflen, "OK %d %d %d",
             te->dumpId,
             status,
             status == WORKER_IGNORED_ERRORS ? AH->public.n_errors : 0);
}

References buf, _tocEntry::dumpId, Archive::n_errors, _archiveHandle::public, snprintf, status(), and WORKER_IGNORED_ERRORS.

Referenced by WaitForCommands().

DispatchJobForTocEntry()

void DispatchJobForTocEntry	(	ArchiveHandle *	AH,
		ParallelState *	pstate,
		TocEntry *	te,
		T_Action	act,
		ParallelCompletionPtr	callback,
		void *	callback_data
	)

Definition at line 1205 of file parallel.c.

{
    int         worker;
    char        buf[256];
 
    /* Get a worker, waiting if none are idle */
    while ((worker = GetIdleWorker(pstate)) == NO_SLOT)
        WaitForWorkers(AH, pstate, WFW_ONE_IDLE);
 
    /* Construct and send command string */
    buildWorkerCommand(AH, te, act, buf, sizeof(buf));
 
    sendMessageToWorker(pstate, worker, buf);
 
    /* Remember worker is busy, and which TocEntry it's working on */
    pstate->parallelSlot[worker].workerStatus = WRKR_WORKING;
    pstate->parallelSlot[worker].callback = callback;
    pstate->parallelSlot[worker].callback_data = callback_data;
    pstate->te[worker] = te;
}

References buf, buildWorkerCommand(), ParallelSlot::callback, callback(), ParallelSlot::callback_data, GetIdleWorker(), NO_SLOT, ParallelState::parallelSlot, sendMessageToWorker(), ParallelState::te, WaitForWorkers(), WFW_ONE_IDLE, ParallelSlot::workerStatus, and WRKR_WORKING.

Referenced by restore_toc_entries_parallel(), and WriteDataChunks().

GetIdleWorker()

static int GetIdleWorker ( ParallelState *  pstate )

static

Definition at line 1236 of file parallel.c.

{
    int         i;
 
    for (i = 0; i < pstate->numWorkers; i++)
    {
        if (pstate->parallelSlot[i].workerStatus == WRKR_IDLE)
            return i;
    }
    return NO_SLOT;
}

References i, NO_SLOT, ParallelState::numWorkers, ParallelState::parallelSlot, ParallelSlot::workerStatus, and WRKR_IDLE.

Referenced by DispatchJobForTocEntry(), and WaitForWorkers().

getMessageFromLeader()

static char * getMessageFromLeader ( int  pipefd[2] )

static

Definition at line 1516 of file parallel.c.

{
    return readMessageFromPipe(pipefd[PIPE_READ]);
}

References PIPE_READ, and readMessageFromPipe().

Referenced by WaitForCommands().

getMessageFromWorker()

static char * getMessageFromWorker ( ParallelState *  pstate, bool  do_wait, int *  worker  )

static

Definition at line 1579 of file parallel.c.

{
    int         i;
    fd_set      workerset;
    int         maxFd = -1;
    struct timeval nowait = {0, 0};
 
    /* construct bitmap of socket descriptors for select() */
    FD_ZERO(&workerset);
    for (i = 0; i < pstate->numWorkers; i++)
    {
        if (!WORKER_IS_RUNNING(pstate->parallelSlot[i].workerStatus))
            continue;
        FD_SET(pstate->parallelSlot[i].pipeRead, &workerset);
        if (pstate->parallelSlot[i].pipeRead > maxFd)
            maxFd = pstate->parallelSlot[i].pipeRead;
    }
 
    if (do_wait)
    {
        i = select_loop(maxFd, &workerset);
        Assert(i != 0);
    }
    else
    {
        if ((i = select(maxFd + 1, &workerset, NULL, NULL, &nowait)) == 0)
            return NULL;
    }
 
    if (i < 0)
        pg_fatal("%s() failed: %m", "select");
 
    for (i = 0; i < pstate->numWorkers; i++)
    {
        char       *msg;
 
        if (!WORKER_IS_RUNNING(pstate->parallelSlot[i].workerStatus))
            continue;
        if (!FD_ISSET(pstate->parallelSlot[i].pipeRead, &workerset))
            continue;
 
        /*
         * Read the message if any.  If the socket is ready because of EOF,
         * we'll return NULL instead (and the socket will stay ready, so the
         * condition will persist).
         *
         * Note: because this is a blocking read, we'll wait if only part of
         * the message is available.  Waiting a long time would be bad, but
         * since worker status messages are short and are always sent in one
         * operation, it shouldn't be a problem in practice.
         */
        msg = readMessageFromPipe(pstate->parallelSlot[i].pipeRead);
        *worker = i;
        return msg;
    }
    Assert(false);
    return NULL;
}

References Assert(), do_wait, i, ParallelState::numWorkers, ParallelState::parallelSlot, pg_fatal, ParallelSlot::pipeRead, readMessageFromPipe(), select, select_loop(), WORKER_IS_RUNNING, and ParallelSlot::workerStatus.

Referenced by ListenToWorkers().

GetMyPSlot()

static ParallelSlot * GetMyPSlot ( ParallelState *  pstate )

static

Definition at line 264 of file parallel.c.

{
    int         i;
 
    for (i = 0; i < pstate->numWorkers; i++)
    {
#ifdef WIN32
        if (pstate->parallelSlot[i].threadId == GetCurrentThreadId())
#else
        if (pstate->parallelSlot[i].pid == getpid())
#endif
            return &(pstate->parallelSlot[i]);
    }
 
    return NULL;
}

References i, ParallelState::numWorkers, ParallelState::parallelSlot, and ParallelSlot::pid.

Referenced by archive_close_connection().

HasEveryWorkerTerminated()

static bool HasEveryWorkerTerminated ( ParallelState *  pstate )

static

Definition at line 1252 of file parallel.c.

{
    int         i;
 
    for (i = 0; i < pstate->numWorkers; i++)
    {
        if (WORKER_IS_RUNNING(pstate->parallelSlot[i].workerStatus))
            return false;
    }
    return true;
}

References i, ParallelState::numWorkers, ParallelState::parallelSlot, WORKER_IS_RUNNING, and ParallelSlot::workerStatus.

Referenced by WaitForTerminatingWorkers().

init_parallel_dump_utils()

void init_parallel_dump_utils

(

void

)

Definition at line 236 of file parallel.c.

{
#ifdef WIN32
    if (!parallel_init_done)
    {
        WSADATA     wsaData;
        int         err;
 
        /* Prepare for threaded operation */
        tls_index = TlsAlloc();
        mainThreadId = GetCurrentThreadId();
 
        /* Initialize socket access */
        err = WSAStartup(MAKEWORD(2, 2), &wsaData);
        if (err != 0)
            pg_fatal("%s() failed: error code %d", "WSAStartup", err);
 
        parallel_init_done = true;
    }
#endif
}

References err(), and pg_fatal.

Referenced by main().

IsEveryWorkerIdle()

bool IsEveryWorkerIdle

(

ParallelState *

pstate

)

Definition at line 1268 of file parallel.c.

{
    int         i;
 
    for (i = 0; i < pstate->numWorkers; i++)
    {
        if (pstate->parallelSlot[i].workerStatus != WRKR_IDLE)
            return false;
    }
    return true;
}

References i, ParallelState::numWorkers, ParallelState::parallelSlot, ParallelSlot::workerStatus, and WRKR_IDLE.

Referenced by ParallelBackupEnd(), restore_toc_entries_parallel(), and WaitForWorkers().

ListenToWorkers()

static bool ListenToWorkers ( ArchiveHandle *  AH, ParallelState *  pstate, bool  do_wait  )

static

Definition at line 1398 of file parallel.c.

{
    int         worker;
    char       *msg;
 
    /* Try to collect a status message */
    msg = getMessageFromWorker(pstate, do_wait, &worker);
 
    if (!msg)
    {
        /* If do_wait is true, we must have detected EOF on some socket */
        if (do_wait)
            pg_fatal("a worker process died unexpectedly");
        return false;
    }
 
    /* Process it and update our idea of the worker's status */
    if (messageStartsWith(msg, "OK "))
    {
        ParallelSlot *slot = &pstate->parallelSlot[worker];
        TocEntry   *te = pstate->te[worker];
        int         status;
 
        status = parseWorkerResponse(AH, te, msg);
        slot->callback(AH, te, status, slot->callback_data);
        slot->workerStatus = WRKR_IDLE;
        pstate->te[worker] = NULL;
    }
    else
        pg_fatal("invalid message received from worker: \"%s\"",
                 msg);
 
    /* Free the string returned from getMessageFromWorker */
    free(msg);
 
    return true;
}

References ParallelSlot::callback, ParallelSlot::callback_data, do_wait, free, getMessageFromWorker(), messageStartsWith, ParallelState::parallelSlot, parseWorkerResponse(), pg_fatal, status(), ParallelState::te, ParallelSlot::workerStatus, and WRKR_IDLE.

Referenced by WaitForWorkers().

lockTableForWorker()

static void lockTableForWorker ( ArchiveHandle *  AH, TocEntry *  te  )

static

Definition at line 1301 of file parallel.c.

{
    const char *qualId;
    PQExpBuffer query;
    PGresult   *res;
 
    /* Nothing to do for BLOBS */
    if (strcmp(te->desc, "BLOBS") == 0)
        return;
 
    query = createPQExpBuffer();
 
    qualId = fmtQualifiedId(te->namespace, te->tag);
 
    appendPQExpBuffer(query, "LOCK TABLE %s IN ACCESS SHARE MODE NOWAIT",
                      qualId);
 
    res = PQexec(AH->connection, query->data);
 
    if (!res || PQresultStatus(res) != PGRES_COMMAND_OK)
        pg_fatal("could not obtain lock on relation \"%s\"\n"
                 "This usually means that someone requested an ACCESS EXCLUSIVE lock "
                 "on the table after the pg_dump parent process had gotten the "
                 "initial ACCESS SHARE lock on the table.", qualId);
 
    PQclear(res);
    destroyPQExpBuffer(query);
}

References appendPQExpBuffer(), _archiveHandle::connection, createPQExpBuffer(), PQExpBufferData::data, _tocEntry::desc, destroyPQExpBuffer(), fmtQualifiedId(), pg_fatal, PGRES_COMMAND_OK, PQclear(), PQexec(), PQresultStatus(), res, and _tocEntry::tag.

Referenced by WaitForCommands().

on_exit_close_archive()

void on_exit_close_archive

(

Archive *

AHX

)

Definition at line 328 of file parallel.c.

{
    shutdown_info.AHX = AHX;
    on_exit_nicely(archive_close_connection, &shutdown_info);
}

References ShutdownInformation::AHX, archive_close_connection(), on_exit_nicely(), and shutdown_info.

Referenced by main().

ParallelBackupEnd()

void ParallelBackupEnd	(	ArchiveHandle *	AH,
		ParallelState *	pstate
	)

Definition at line 1059 of file parallel.c.

{
    int         i;
 
    /* No work if non-parallel */
    if (pstate->numWorkers == 1)
        return;
 
    /* There should not be any unfinished jobs */
    Assert(IsEveryWorkerIdle(pstate));
 
    /* Close the sockets so that the workers know they can exit */
    for (i = 0; i < pstate->numWorkers; i++)
    {
        closesocket(pstate->parallelSlot[i].pipeRead);
        closesocket(pstate->parallelSlot[i].pipeWrite);
    }
 
    /* Wait for them to exit */
    WaitForTerminatingWorkers(pstate);
 
    /*
     * Unlink pstate from shutdown_info, so the exit handler will not try to
     * use it; and likewise unlink from signal_info.
     */
    shutdown_info.pstate = NULL;
    set_cancel_pstate(NULL);
 
    /* Release state (mere neatnik-ism, since we're about to terminate) */
    free(pstate->te);
    free(pstate->parallelSlot);
    free(pstate);
}

References Assert(), closesocket, free, i, IsEveryWorkerIdle(), ParallelState::numWorkers, ParallelState::parallelSlot, ParallelSlot::pipeRead, ParallelSlot::pipeWrite, ShutdownInformation::pstate, set_cancel_pstate(), shutdown_info, ParallelState::te, and WaitForTerminatingWorkers().

Referenced by _CloseArchive(), and RestoreArchive().

ParallelBackupStart()

ParallelState* ParallelBackupStart

(

ArchiveHandle *

AH

)

Definition at line 897 of file parallel.c.

{
    ParallelState *pstate;
    int         i;
 
    Assert(AH->public.numWorkers > 0);
 
    pstate = (ParallelState *) pg_malloc(sizeof(ParallelState));
 
    pstate->numWorkers = AH->public.numWorkers;
    pstate->te = NULL;
    pstate->parallelSlot = NULL;
 
    if (AH->public.numWorkers == 1)
        return pstate;
 
    /* Create status arrays, being sure to initialize all fields to 0 */
    pstate->te = (TocEntry **)
        pg_malloc0(pstate->numWorkers * sizeof(TocEntry *));
    pstate->parallelSlot = (ParallelSlot *)
        pg_malloc0(pstate->numWorkers * sizeof(ParallelSlot));
 
#ifdef WIN32
    /* Make fmtId() and fmtQualifiedId() use thread-local storage */
    getLocalPQExpBuffer = getThreadLocalPQExpBuffer;
#endif
 
    /*
     * Set the pstate in shutdown_info, to tell the exit handler that it must
     * clean up workers as well as the main database connection.  But we don't
     * set this in signal_info yet, because we don't want child processes to
     * inherit non-NULL signal_info.pstate.
     */
    shutdown_info.pstate = pstate;
 
    /*
     * Temporarily disable query cancellation on the leader connection.  This
     * ensures that child processes won't inherit valid AH->connCancel
     * settings and thus won't try to issue cancels against the leader's
     * connection.  No harm is done if we fail while it's disabled, because
     * the leader connection is idle at this point anyway.
     */
    set_archive_cancel_info(AH, NULL);
 
    /* Ensure stdio state is quiesced before forking */
    fflush(NULL);
 
    /* Create desired number of workers */
    for (i = 0; i < pstate->numWorkers; i++)
    {
#ifdef WIN32
        WorkerInfo *wi;
        uintptr_t   handle;
#else
        pid_t       pid;
#endif
        ParallelSlot *slot = &(pstate->parallelSlot[i]);
        int         pipeMW[2],
                    pipeWM[2];
 
        /* Create communication pipes for this worker */
        if (pgpipe(pipeMW) < 0 || pgpipe(pipeWM) < 0)
            pg_fatal("could not create communication channels: %m");
 
        /* leader's ends of the pipes */
        slot->pipeRead = pipeWM[PIPE_READ];
        slot->pipeWrite = pipeMW[PIPE_WRITE];
        /* child's ends of the pipes */
        slot->pipeRevRead = pipeMW[PIPE_READ];
        slot->pipeRevWrite = pipeWM[PIPE_WRITE];
 
#ifdef WIN32
        /* Create transient structure to pass args to worker function */
        wi = (WorkerInfo *) pg_malloc(sizeof(WorkerInfo));
 
        wi->AH = AH;
        wi->slot = slot;
 
        handle = _beginthreadex(NULL, 0, (void *) &init_spawned_worker_win32,
                                wi, 0, &(slot->threadId));
        slot->hThread = handle;
        slot->workerStatus = WRKR_IDLE;
#else                           /* !WIN32 */
        pid = fork();
        if (pid == 0)
        {
            /* we are the worker */
            int         j;
 
            /* this is needed for GetMyPSlot() */
            slot->pid = getpid();
 
            /* instruct signal handler that we're in a worker now */
            signal_info.am_worker = true;
 
            /* close read end of Worker -> Leader */
            closesocket(pipeWM[PIPE_READ]);
            /* close write end of Leader -> Worker */
            closesocket(pipeMW[PIPE_WRITE]);
 
            /*
             * Close all inherited fds for communication of the leader with
             * previously-forked workers.
             */
            for (j = 0; j < i; j++)
            {
                closesocket(pstate->parallelSlot[j].pipeRead);
                closesocket(pstate->parallelSlot[j].pipeWrite);
            }
 
            /* Run the worker ... */
            RunWorker(AH, slot);
 
            /* We can just exit(0) when done */
            exit(0);
        }
        else if (pid < 0)
        {
            /* fork failed */
            pg_fatal("could not create worker process: %m");
        }
 
        /* In Leader after successful fork */
        slot->pid = pid;
        slot->workerStatus = WRKR_IDLE;
 
        /* close read end of Leader -> Worker */
        closesocket(pipeMW[PIPE_READ]);
        /* close write end of Worker -> Leader */
        closesocket(pipeWM[PIPE_WRITE]);
#endif                          /* WIN32 */
    }
 
    /*
     * Having forked off the workers, disable SIGPIPE so that leader isn't
     * killed if it tries to send a command to a dead worker.  We don't want
     * the workers to inherit this setting, though.
     */
#ifndef WIN32
    pqsignal(SIGPIPE, SIG_IGN);
#endif
 
    /*
     * Re-establish query cancellation on the leader connection.
     */
    set_archive_cancel_info(AH, AH->connection);
 
    /*
     * Tell the cancel signal handler to forward signals to worker processes,
     * too.  (As with query cancel, we did not need this earlier because the
     * workers have not yet been given anything to do; if we die before this
     * point, any already-started workers will see EOF and quit promptly.)
     */
    set_cancel_pstate(pstate);
 
    return pstate;
}

References DumpSignalInformation::am_worker, Assert(), closesocket, _archiveHandle::connection, exit(), fflush(), getLocalPQExpBuffer, i, j, ParallelState::numWorkers, Archive::numWorkers, ParallelState::parallelSlot, pg_fatal, pg_malloc(), pg_malloc0(), pgpipe, ParallelSlot::pid, PIPE_READ, PIPE_WRITE, ParallelSlot::pipeRead, ParallelSlot::pipeRevRead, ParallelSlot::pipeRevWrite, ParallelSlot::pipeWrite, pqsignal(), ShutdownInformation::pstate, _archiveHandle::public, RunWorker(), set_archive_cancel_info(), set_cancel_pstate(), shutdown_info, SIG_IGN, signal_info, SIGPIPE, ParallelState::te, ParallelSlot::workerStatus, and WRKR_IDLE.

Referenced by _CloseArchive(), and RestoreArchive().

parseWorkerCommand()

static void parseWorkerCommand ( ArchiveHandle *  AH, TocEntry **  te, T_Action *  act, const char *  msg  )

static

Definition at line 1123 of file parallel.c.

{
    DumpId      dumpId;
    int         nBytes;
 
    if (messageStartsWith(msg, "DUMP "))
    {
        *act = ACT_DUMP;
        sscanf(msg, "DUMP %d%n", &dumpId, &nBytes);
        Assert(nBytes == strlen(msg));
        *te = getTocEntryByDumpId(AH, dumpId);
        Assert(*te != NULL);
    }
    else if (messageStartsWith(msg, "RESTORE "))
    {
        *act = ACT_RESTORE;
        sscanf(msg, "RESTORE %d%n", &dumpId, &nBytes);
        Assert(nBytes == strlen(msg));
        *te = getTocEntryByDumpId(AH, dumpId);
        Assert(*te != NULL);
    }
    else
        pg_fatal("unrecognized command received from leader: \"%s\"",
                 msg);
}

References ACT_DUMP, ACT_RESTORE, Assert(), getTocEntryByDumpId(), messageStartsWith, and pg_fatal.

Referenced by WaitForCommands().

parseWorkerResponse()

static int parseWorkerResponse ( ArchiveHandle *  AH, TocEntry *  te, const char *  msg  )

static

Definition at line 1171 of file parallel.c.

{
    DumpId      dumpId;
    int         nBytes,
                n_errors;
    int         status = 0;
 
    if (messageStartsWith(msg, "OK "))
    {
        sscanf(msg, "OK %d %d %d%n", &dumpId, &status, &n_errors, &nBytes);
 
        Assert(dumpId == te->dumpId);
        Assert(nBytes == strlen(msg));
 
        AH->public.n_errors += n_errors;
    }
    else
        pg_fatal("invalid message received from worker: \"%s\"",
                 msg);
 
    return status;
}

References Assert(), _tocEntry::dumpId, messageStartsWith, Archive::n_errors, pg_fatal, _archiveHandle::public, and status().

Referenced by ListenToWorkers().

readMessageFromPipe()

static char * readMessageFromPipe ( int  fd )

static

Definition at line 1662 of file parallel.c.

{
    char       *msg;
    int         msgsize,
                bufsize;
    int         ret;
 
    /*
     * In theory, if we let piperead() read multiple bytes, it might give us
     * back fragments of multiple messages.  (That can't actually occur, since
     * neither leader nor workers send more than one message without waiting
     * for a reply, but we don't wish to assume that here.)  For simplicity,
     * read a byte at a time until we get the terminating '\0'.  This method
     * is a bit inefficient, but since this is only used for relatively short
     * command and status strings, it shouldn't matter.
     */
    bufsize = 64;               /* could be any number */
    msg = (char *) pg_malloc(bufsize);
    msgsize = 0;
    for (;;)
    {
        Assert(msgsize < bufsize);
        ret = piperead(fd, msg + msgsize, 1);
        if (ret <= 0)
            break;              /* error or connection closure */
 
        Assert(ret == 1);
 
        if (msg[msgsize] == '\0')
            return msg;         /* collected whole message */
 
        msgsize++;
        if (msgsize == bufsize) /* enlarge buffer if needed */
        {
            bufsize += 16;      /* could be any number */
            msg = (char *) pg_realloc(msg, bufsize);
        }
    }
 
    /* Other end has closed the connection */
    pg_free(msg);
    return NULL;
}

References Assert(), bufsize, fd(), pg_free(), pg_malloc(), pg_realloc(), and piperead.

Referenced by getMessageFromLeader(), and getMessageFromWorker().

RunWorker()

static void RunWorker ( ArchiveHandle *  AH, ParallelSlot *  slot  )

static

Definition at line 829 of file parallel.c.

{
    int         pipefd[2];
 
    /* fetch child ends of pipes */
    pipefd[PIPE_READ] = slot->pipeRevRead;
    pipefd[PIPE_WRITE] = slot->pipeRevWrite;
 
    /*
     * Clone the archive so that we have our own state to work with, and in
     * particular our own database connection.
     *
     * We clone on Unix as well as Windows, even though technically we don't
     * need to because fork() gives us a copy in our own address space
     * already.  But CloneArchive resets the state information and also clones
     * the database connection which both seem kinda helpful.
     */
    AH = CloneArchive(AH);
 
    /* Remember cloned archive where signal handler can find it */
    set_cancel_slot_archive(slot, AH);
 
    /*
     * Call the setup worker function that's defined in the ArchiveHandle.
     */
    (AH->SetupWorkerPtr) ((Archive *) AH);
 
    /*
     * Execute commands until done.
     */
    WaitForCommands(AH, pipefd);
 
    /*
     * Disconnect from database and clean up.
     */
    set_cancel_slot_archive(slot, NULL);
    DisconnectDatabase(&(AH->public));
    DeCloneArchive(AH);
}

References CloneArchive(), DeCloneArchive(), DisconnectDatabase(), PIPE_READ, PIPE_WRITE, ParallelSlot::pipeRevRead, ParallelSlot::pipeRevWrite, _archiveHandle::public, set_cancel_slot_archive(), _archiveHandle::SetupWorkerPtr, and WaitForCommands().

Referenced by ParallelBackupStart().

select_loop()

static int select_loop ( int  maxFd, fd_set *  workerset  )

static

Definition at line 1540 of file parallel.c.

{
    int         i;
    fd_set      saveSet = *workerset;
 
    for (;;)
    {
        *workerset = saveSet;
        i = select(maxFd + 1, workerset, NULL, NULL, NULL);
 
#ifndef WIN32
        if (i < 0 && errno == EINTR)
            continue;
#else
        if (i == SOCKET_ERROR && WSAGetLastError() == WSAEINTR)
            continue;
#endif
        break;
    }
 
    return i;
}

References EINTR, i, and select.

Referenced by getMessageFromWorker().

sendMessageToLeader()

static void sendMessageToLeader ( int  pipefd[2], const char *  str  )

static

Definition at line 1527 of file parallel.c.

{
    int         len = strlen(str) + 1;
 
    if (pipewrite(pipefd[PIPE_WRITE], str, len) != len)
        pg_fatal("could not write to the communication channel: %m");
}

References len, pg_fatal, PIPE_WRITE, pipewrite, and generate_unaccent_rules::str.

Referenced by WaitForCommands().

sendMessageToWorker()

static void sendMessageToWorker ( ParallelState *  pstate, int  worker, const char *  str  )

static

Definition at line 1644 of file parallel.c.

{
    int         len = strlen(str) + 1;
 
    if (pipewrite(pstate->parallelSlot[worker].pipeWrite, str, len) != len)
    {
        pg_fatal("could not write to the communication channel: %m");
    }
}

References len, ParallelState::parallelSlot, pg_fatal, ParallelSlot::pipeWrite, pipewrite, and generate_unaccent_rules::str.

Referenced by DispatchJobForTocEntry().

set_archive_cancel_info()

void set_archive_cancel_info	(	ArchiveHandle *	AH,
		PGconn *	conn
	)

Definition at line 730 of file parallel.c.

{
    PGcancel   *oldConnCancel;
 
    /*
     * Activate the interrupt handler if we didn't yet in this process.  On
     * Windows, this also initializes signal_info_lock; therefore it's
     * important that this happen at least once before we fork off any
     * threads.
     */
    setup_cancel_handler();
 
    /*
     * On Unix, we assume that storing a pointer value is atomic with respect
     * to any possible signal interrupt.  On Windows, use a critical section.
     */
 
#ifdef WIN32
    EnterCriticalSection(&signal_info_lock);
#endif
 
    /* Free the old one if we have one */
    oldConnCancel = AH->connCancel;
    /* be sure interrupt handler doesn't use pointer while freeing */
    AH->connCancel = NULL;
 
    if (oldConnCancel != NULL)
        PQfreeCancel(oldConnCancel);
 
    /* Set the new one if specified */
    if (conn)
        AH->connCancel = PQgetCancel(conn);
 
    /*
     * On Unix, there's only ever one active ArchiveHandle per process, so we
     * can just set signal_info.myAH unconditionally.  On Windows, do that
     * only in the main thread; worker threads have to make sure their
     * ArchiveHandle appears in the pstate data, which is dealt with in
     * RunWorker().
     */
#ifndef WIN32
    signal_info.myAH = AH;
#else
    if (mainThreadId == GetCurrentThreadId())
        signal_info.myAH = AH;
#endif
 
#ifdef WIN32
    LeaveCriticalSection(&signal_info_lock);
#endif
}

References conn, _archiveHandle::connCancel, DumpSignalInformation::myAH, PQfreeCancel(), PQgetCancel(), setup_cancel_handler(), and signal_info.

Referenced by ConnectDatabase(), DisconnectDatabase(), and ParallelBackupStart().

set_cancel_pstate()

static void set_cancel_pstate ( ParallelState *  pstate )

static

Definition at line 789 of file parallel.c.

{
#ifdef WIN32
    EnterCriticalSection(&signal_info_lock);
#endif
 
    signal_info.pstate = pstate;
 
#ifdef WIN32
    LeaveCriticalSection(&signal_info_lock);
#endif
}

References DumpSignalInformation::pstate, and signal_info.

Referenced by ParallelBackupEnd(), and ParallelBackupStart().

set_cancel_slot_archive()

static void set_cancel_slot_archive ( ParallelSlot *  slot, ArchiveHandle *  AH  )

static

Definition at line 809 of file parallel.c.

{
#ifdef WIN32
    EnterCriticalSection(&signal_info_lock);
#endif
 
    slot->AH = AH;
 
#ifdef WIN32
    LeaveCriticalSection(&signal_info_lock);
#endif
}

References ParallelSlot::AH.

Referenced by RunWorker().

setup_cancel_handler()

static void setup_cancel_handler ( void  )

static

Definition at line 608 of file parallel.c.

{
    /*
     * When forking, signal_info.handler_set will propagate into the new
     * process, but that's fine because the signal handler state does too.
     */
    if (!signal_info.handler_set)
    {
        signal_info.handler_set = true;
 
        pqsignal(SIGINT, sigTermHandler);
        pqsignal(SIGTERM, sigTermHandler);
        pqsignal(SIGQUIT, sigTermHandler);
    }
}

References DumpSignalInformation::handler_set, pqsignal(), signal_info, SIGQUIT, and sigTermHandler().

Referenced by main(), psql_setup_cancel_handler(), runInitSteps(), and set_archive_cancel_info().

ShutdownWorkersHard()

static void ShutdownWorkersHard ( ParallelState *  pstate )

static

Definition at line 395 of file parallel.c.

{
    int         i;
 
    /*
     * Close our write end of the sockets so that any workers waiting for
     * commands know they can exit.  (Note: some of the pipeWrite fields might
     * still be zero, if we failed to initialize all the workers.  Hence, just
     * ignore errors here.)
     */
    for (i = 0; i < pstate->numWorkers; i++)
        closesocket(pstate->parallelSlot[i].pipeWrite);
 
    /*
     * Force early termination of any commands currently in progress.
     */
#ifndef WIN32
    /* On non-Windows, send SIGTERM to each worker process. */
    for (i = 0; i < pstate->numWorkers; i++)
    {
        pid_t       pid = pstate->parallelSlot[i].pid;
 
        if (pid != 0)
            kill(pid, SIGTERM);
    }
#else
 
    /*
     * On Windows, send query cancels directly to the workers' backends.  Use
     * a critical section to ensure worker threads don't change state.
     */
    EnterCriticalSection(&signal_info_lock);
    for (i = 0; i < pstate->numWorkers; i++)
    {
        ArchiveHandle *AH = pstate->parallelSlot[i].AH;
        char        errbuf[1];
 
        if (AH != NULL && AH->connCancel != NULL)
            (void) PQcancel(AH->connCancel, errbuf, sizeof(errbuf));
    }
    LeaveCriticalSection(&signal_info_lock);
#endif
 
    /* Now wait for them to terminate. */
    WaitForTerminatingWorkers(pstate);
}

References ParallelSlot::AH, closesocket, _archiveHandle::connCancel, i, kill, ParallelState::numWorkers, ParallelState::parallelSlot, ParallelSlot::pid, ParallelSlot::pipeWrite, PQcancel(), and WaitForTerminatingWorkers().

Referenced by archive_close_connection().

sigTermHandler()

static void sigTermHandler ( SIGNAL_ARGS  )

static

Definition at line 545 of file parallel.c.

{
    int         i;
    char        errbuf[1];
 
    /*
     * Some platforms allow delivery of new signals to interrupt an active
     * signal handler.  That could muck up our attempt to send PQcancel, so
     * disable the signals that setup_cancel_handler enabled.
     */
    pqsignal(SIGINT, SIG_IGN);
    pqsignal(SIGTERM, SIG_IGN);
    pqsignal(SIGQUIT, SIG_IGN);
 
    /*
     * If we're in the leader, forward signal to all workers.  (It seems best
     * to do this before PQcancel; killing the leader transaction will result
     * in invalid-snapshot errors from active workers, which maybe we can
     * quiet by killing workers first.)  Ignore any errors.
     */
    if (signal_info.pstate != NULL)
    {
        for (i = 0; i < signal_info.pstate->numWorkers; i++)
        {
            pid_t       pid = signal_info.pstate->parallelSlot[i].pid;
 
            if (pid != 0)
                kill(pid, SIGTERM);
        }
    }
 
    /*
     * Send QueryCancel if we have a connection to send to.  Ignore errors,
     * there's not much we can do about them anyway.
     */
    if (signal_info.myAH != NULL && signal_info.myAH->connCancel != NULL)
        (void) PQcancel(signal_info.myAH->connCancel, errbuf, sizeof(errbuf));
 
    /*
     * Report we're quitting, using nothing more complicated than write(2).
     * When in parallel operation, only the leader process should do this.
     */
    if (!signal_info.am_worker)
    {
        if (progname)
        {
            write_stderr(progname);
            write_stderr(": ");
        }
        write_stderr("terminated by user\n");
    }
 
    /*
     * And die, using _exit() not exit() because the latter will invoke atexit
     * handlers that can fail if we interrupted related code.
     */
    _exit(1);
}

References DumpSignalInformation::am_worker, _archiveHandle::connCancel, i, kill, DumpSignalInformation::myAH, ParallelState::numWorkers, ParallelState::parallelSlot, ParallelSlot::pid, PQcancel(), pqsignal(), progname, DumpSignalInformation::pstate, SIG_IGN, signal_info, SIGQUIT, and write_stderr.

Referenced by setup_cancel_handler().

WaitForCommands()

static void WaitForCommands ( ArchiveHandle *  AH, int  pipefd[2]  )

static

Definition at line 1336 of file parallel.c.

{
    char       *command;
    TocEntry   *te;
    T_Action    act;
    int         status = 0;
    char        buf[256];
 
    for (;;)
    {
        if (!(command = getMessageFromLeader(pipefd)))
        {
            /* EOF, so done */
            return;
        }
 
        /* Decode the command */
        parseWorkerCommand(AH, &te, &act, command);
 
        if (act == ACT_DUMP)
        {
            /* Acquire lock on this table within the worker's session */
            lockTableForWorker(AH, te);
 
            /* Perform the dump command */
            status = (AH->WorkerJobDumpPtr) (AH, te);
        }
        else if (act == ACT_RESTORE)
        {
            /* Perform the restore command */
            status = (AH->WorkerJobRestorePtr) (AH, te);
        }
        else
            Assert(false);
 
        /* Return status to leader */
        buildWorkerResponse(AH, te, act, status, buf, sizeof(buf));
 
        sendMessageToLeader(pipefd, buf);
 
        /* command was pg_malloc'd and we are responsible for free()ing it. */
        free(command);
    }
}

References ACT_DUMP, ACT_RESTORE, Assert(), buf, buildWorkerResponse(), free, getMessageFromLeader(), lockTableForWorker(), parseWorkerCommand(), sendMessageToLeader(), status(), _archiveHandle::WorkerJobDumpPtr, and _archiveHandle::WorkerJobRestorePtr.

Referenced by RunWorker().

WaitForTerminatingWorkers()

static void WaitForTerminatingWorkers ( ParallelState *  pstate )

static

Definition at line 446 of file parallel.c.

{
    while (!HasEveryWorkerTerminated(pstate))
    {
        ParallelSlot *slot = NULL;
        int         j;
 
#ifndef WIN32
        /* On non-Windows, use wait() to wait for next worker to end */
        int         status;
        pid_t       pid = wait(&status);
 
        /* Find dead worker's slot, and clear the PID field */
        for (j = 0; j < pstate->numWorkers; j++)
        {
            slot = &(pstate->parallelSlot[j]);
            if (slot->pid == pid)
            {
                slot->pid = 0;
                break;
            }
        }
#else                           /* WIN32 */
        /* On Windows, we must use WaitForMultipleObjects() */
        HANDLE     *lpHandles = pg_malloc(sizeof(HANDLE) * pstate->numWorkers);
        int         nrun = 0;
        DWORD       ret;
        uintptr_t   hThread;
 
        for (j = 0; j < pstate->numWorkers; j++)
        {
            if (WORKER_IS_RUNNING(pstate->parallelSlot[j].workerStatus))
            {
                lpHandles[nrun] = (HANDLE) pstate->parallelSlot[j].hThread;
                nrun++;
            }
        }
        ret = WaitForMultipleObjects(nrun, lpHandles, false, INFINITE);
        Assert(ret != WAIT_FAILED);
        hThread = (uintptr_t) lpHandles[ret - WAIT_OBJECT_0];
        free(lpHandles);
 
        /* Find dead worker's slot, and clear the hThread field */
        for (j = 0; j < pstate->numWorkers; j++)
        {
            slot = &(pstate->parallelSlot[j]);
            if (slot->hThread == hThread)
            {
                /* For cleanliness, close handles for dead threads */
                CloseHandle((HANDLE) slot->hThread);
                slot->hThread = (uintptr_t) INVALID_HANDLE_VALUE;
                break;
            }
        }
#endif                          /* WIN32 */
 
        /* On all platforms, update workerStatus and te[] as well */
        Assert(j < pstate->numWorkers);
        slot->workerStatus = WRKR_TERMINATED;
        pstate->te[j] = NULL;
    }
}

References Assert(), free, HasEveryWorkerTerminated(), j, ParallelState::numWorkers, ParallelState::parallelSlot, pg_malloc(), ParallelSlot::pid, status(), ParallelState::te, WORKER_IS_RUNNING, ParallelSlot::workerStatus, and WRKR_TERMINATED.

Referenced by ParallelBackupEnd(), and ShutdownWorkersHard().

WaitForWorkers()

void WaitForWorkers	(	ArchiveHandle *	AH,
		ParallelState *	pstate,
		WFW_WaitOption	mode
	)

Definition at line 1451 of file parallel.c.

{
    bool        do_wait = false;
 
    /*
     * In GOT_STATUS mode, always block waiting for a message, since we can't
     * return till we get something.  In other modes, we don't block the first
     * time through the loop.
     */
    if (mode == WFW_GOT_STATUS)
    {
        /* Assert that caller knows what it's doing */
        Assert(!IsEveryWorkerIdle(pstate));
        do_wait = true;
    }
 
    for (;;)
    {
        /*
         * Check for status messages, even if we don't need to block.  We do
         * not try very hard to reap all available messages, though, since
         * there's unlikely to be more than one.
         */
        if (ListenToWorkers(AH, pstate, do_wait))
        {
            /*
             * If we got a message, we are done by definition for GOT_STATUS
             * mode, and we can also be certain that there's at least one idle
             * worker.  So we're done in all but ALL_IDLE mode.
             */
            if (mode != WFW_ALL_IDLE)
                return;
        }
 
        /* Check whether we must wait for new status messages */
        switch (mode)
        {
            case WFW_NO_WAIT:
                return;         /* never wait */
            case WFW_GOT_STATUS:
                Assert(false);  /* can't get here, because we waited */
                break;
            case WFW_ONE_IDLE:
                if (GetIdleWorker(pstate) != NO_SLOT)
                    return;
                break;
            case WFW_ALL_IDLE:
                if (IsEveryWorkerIdle(pstate))
                    return;
                break;
        }
 
        /* Loop back, and this time wait for something to happen */
        do_wait = true;
    }
}

References Assert(), do_wait, GetIdleWorker(), IsEveryWorkerIdle(), ListenToWorkers(), mode, NO_SLOT, WFW_ALL_IDLE, WFW_GOT_STATUS, WFW_NO_WAIT, and WFW_ONE_IDLE.

Referenced by DispatchJobForTocEntry(), restore_toc_entries_parallel(), and WriteDataChunks().

Variable Documentation

shutdown_info

ShutdownInformation shutdown_info

static

Definition at line 152 of file parallel.c.

Referenced by on_exit_close_archive(), ParallelBackupEnd(), and ParallelBackupStart().

signal_info

volatile DumpSignalInformation signal_info

static

Definition at line 173 of file parallel.c.

Referenced by ParallelBackupStart(), set_archive_cancel_info(), set_cancel_pstate(), setup_cancel_handler(), and sigTermHandler().