* the backend's "backend/libpq" is quite separate from "interfaces/libpq".
* All that remains is similarities of names to trap the unwary...
*
- * Portions Copyright (c) 1996-2008, PostgreSQL Global Development Group
+ * Portions Copyright (c) 1996-2014, PostgreSQL Global Development Group
* Portions Copyright (c) 1994, Regents of the University of California
*
- * $PostgreSQL: pgsql/src/backend/libpq/pqcomm.c,v 1.198 2008/01/01 19:45:49 momjian Exp $
+ * src/backend/libpq/pqcomm.c
*
*-------------------------------------------------------------------------
*/
* StreamServerPort - Open postmaster's server port
* StreamConnection - Create new connection with client
* StreamClose - Close a client/backend connection
- * TouchSocketFile - Protect socket file against /tmp cleaners
+ * TouchSocketFiles - Protect socket files against /tmp cleaners
* pq_init - initialize libpq at backend startup
* pq_comm_reset - reset libpq during error recovery
* pq_close - shutdown libpq at backend exit
* pq_peekbyte - peek at next byte from connection
* pq_putbytes - send bytes to connection (not flushed until pq_flush)
* pq_flush - flush pending output
+ * pq_flush_if_writable - flush pending output if writable without blocking
+ * pq_getbyte_if_available - get a byte if available without blocking
*
* message-level I/O (and old-style-COPY-OUT cruft):
* pq_putmessage - send a normal message (suppressed in COPY OUT mode)
+ * pq_putmessage_noblock - buffer a normal message (suppressed in COPY OUT)
* pq_startcopyout - inform libpq that a COPY OUT transfer is beginning
* pq_endcopyout - end a COPY OUT transfer
*
#ifdef HAVE_UTIME_H
#include <utime.h>
#endif
+#ifdef WIN32_ONLY_COMPILER /* mstcpip.h is missing on mingw */
+#include <mstcpip.h>
+#endif
#include "libpq/ip.h"
#include "libpq/libpq.h"
#include "miscadmin.h"
#include "storage/ipc.h"
#include "utils/guc.h"
+#include "utils/memutils.h"
/*
* Configuration options
int Unix_socket_permissions;
char *Unix_socket_group;
+/* Where the Unix socket files are (list of palloc'd strings) */
+static List *sock_paths = NIL;
-/* Where the Unix socket file is */
-static char sock_path[MAXPGPATH];
+PQcommMethods *PqCommMethods;
/*
- * Buffers for low-level I/O
+ * Buffers for low-level I/O.
+ *
+ * The receive buffer is fixed size. Send buffer is usually 8k, but can be
+ * enlarged by pq_putmessage_noblock() if the message doesn't fit otherwise.
*/
-#define PQ_BUFFER_SIZE 8192
+#define PQ_SEND_BUFFER_SIZE 8192
+#define PQ_RECV_BUFFER_SIZE 8192
-static char PqSendBuffer[PQ_BUFFER_SIZE];
+static char *PqSendBuffer;
+static int PqSendBufferSize; /* Size send buffer */
static int PqSendPointer; /* Next index to store a byte in PqSendBuffer */
+static int PqSendStart; /* Next index to send a byte in PqSendBuffer */
-static char PqRecvBuffer[PQ_BUFFER_SIZE];
+static char PqRecvBuffer[PQ_RECV_BUFFER_SIZE];
static int PqRecvPointer; /* Next index to read a byte from PqRecvBuffer */
static int PqRecvLength; /* End of data available in PqRecvBuffer */
/* Internal functions */
-static void pq_close(int code, Datum arg);
+static void socket_comm_reset(void);
+static void socket_close(int code, Datum arg);
+static void socket_set_nonblocking(bool nonblocking);
+static int socket_flush(void);
+static int socket_flush_if_writable(void);
+static bool socket_is_send_pending(void);
+static int socket_putmessage(char msgtype, const char *s, size_t len);
+static void socket_putmessage_noblock(char msgtype, const char *s, size_t len);
+static void socket_startcopyout(void);
+static void socket_endcopyout(bool errorAbort);
static int internal_putbytes(const char *s, size_t len);
static int internal_flush(void);
+static void socket_set_nonblocking(bool nonblocking);
#ifdef HAVE_UNIX_SOCKETS
-static int Lock_AF_UNIX(unsigned short portNumber, char *unixSocketName);
-static int Setup_AF_UNIX(void);
+static int Lock_AF_UNIX(char *unixSocketDir, char *unixSocketPath);
+static int Setup_AF_UNIX(char *sock_path);
#endif /* HAVE_UNIX_SOCKETS */
+PQcommMethods PQcommSocketMethods;
+
+static PQcommMethods PqCommSocketMethods = {
+ socket_comm_reset,
+ socket_flush,
+ socket_flush_if_writable,
+ socket_is_send_pending,
+ socket_putmessage,
+ socket_putmessage_noblock,
+ socket_startcopyout,
+ socket_endcopyout
+};
+
/* --------------------------------
* pq_init - initialize libpq at backend startup
void
pq_init(void)
{
- PqSendPointer = PqRecvPointer = PqRecvLength = 0;
+ PqCommMethods = &PqCommSocketMethods;
+ PqSendBufferSize = PQ_SEND_BUFFER_SIZE;
+ PqSendBuffer = MemoryContextAlloc(TopMemoryContext, PqSendBufferSize);
+ PqSendPointer = PqSendStart = PqRecvPointer = PqRecvLength = 0;
PqCommBusy = false;
DoingCopyOut = false;
- on_proc_exit(pq_close, 0);
+ on_proc_exit(socket_close, 0);
}
/* --------------------------------
- * pq_comm_reset - reset libpq during error recovery
+ * socket_comm_reset - reset libpq during error recovery
*
* This is called from error recovery at the outer idle loop. It's
* just to get us out of trouble if we somehow manage to elog() from
* inside a pqcomm.c routine (which ideally will never happen, but...)
* --------------------------------
*/
-void
-pq_comm_reset(void)
+static void
+socket_comm_reset(void)
{
/* Do not throw away pending data, but do reset the busy flag */
PqCommBusy = false;
}
/* --------------------------------
- * pq_close - shutdown libpq at backend exit
+ * socket_close - shutdown libpq at backend exit
*
* Note: in a standalone backend MyProcPort will be null,
* don't crash during exit...
* --------------------------------
*/
static void
-pq_close(int code, Datum arg)
+socket_close(int code, Datum arg)
{
if (MyProcPort != NULL)
{
* transport layer reports connection closure, and you can be sure the
* backend has exited.
*
- * We do set sock to -1 to prevent any further I/O, though.
+ * We do set sock to PGINVALID_SOCKET to prevent any further I/O,
+ * though.
*/
- MyProcPort->sock = -1;
+ MyProcPort->sock = PGINVALID_SOCKET;
}
}
/* StreamDoUnlink()
* Shutdown routine for backend connection
- * If a Unix socket is used for communication, explicitly close it.
+ * If any Unix sockets are used for communication, explicitly close them.
*/
#ifdef HAVE_UNIX_SOCKETS
static void
StreamDoUnlink(int code, Datum arg)
{
- Assert(sock_path[0]);
- unlink(sock_path);
+ ListCell *l;
+
+ /* Loop through all created sockets... */
+ foreach(l, sock_paths)
+ {
+ char *sock_path = (char *) lfirst(l);
+
+ unlink(sock_path);
+ }
+ /* Since we're about to exit, no need to reclaim storage */
+ sock_paths = NIL;
}
#endif /* HAVE_UNIX_SOCKETS */
/*
* StreamServerPort -- open a "listening" port to accept connections.
*
- * Successfully opened sockets are added to the ListenSocket[] array,
- * at the first position that isn't -1.
+ * family should be AF_UNIX or AF_UNSPEC; portNumber is the port number.
+ * For AF_UNIX ports, hostName should be NULL and unixSocketDir must be
+ * specified. For TCP ports, hostName is either NULL for all interfaces or
+ * the interface to listen on, and unixSocketDir is ignored (can be NULL).
+ *
+ * Successfully opened sockets are added to the ListenSocket[] array (of
+ * length MaxListen), at the first position that isn't PGINVALID_SOCKET.
*
* RETURNS: STATUS_OK or STATUS_ERROR
*/
int
StreamServerPort(int family, char *hostName, unsigned short portNumber,
- char *unixSocketName,
- int ListenSocket[], int MaxListen)
+ char *unixSocketDir,
+ pgsocket ListenSocket[], int MaxListen)
{
- int fd,
- err;
+ pgsocket fd;
+ int err;
int maxconn;
int ret;
char portNumberStr[32];
int listen_index = 0;
int added = 0;
+#ifdef HAVE_UNIX_SOCKETS
+ char unixSocketPath[MAXPGPATH];
+#endif
#if !defined(WIN32) || defined(IPV6_V6ONLY)
int one = 1;
#endif
#ifdef HAVE_UNIX_SOCKETS
if (family == AF_UNIX)
{
- /* Lock_AF_UNIX will also fill in sock_path. */
- if (Lock_AF_UNIX(portNumber, unixSocketName) != STATUS_OK)
+ /*
+ * Create unixSocketPath from portNumber and unixSocketDir and lock
+ * that file path
+ */
+ UNIXSOCK_PATH(unixSocketPath, portNumber, unixSocketDir);
+ if (strlen(unixSocketPath) >= UNIXSOCK_PATH_BUFLEN)
+ {
+ ereport(LOG,
+ (errmsg("Unix-domain socket path \"%s\" is too long (maximum %d bytes)",
+ unixSocketPath,
+ (int) (UNIXSOCK_PATH_BUFLEN - 1))));
return STATUS_ERROR;
- service = sock_path;
+ }
+ if (Lock_AF_UNIX(unixSocketDir, unixSocketPath) != STATUS_OK)
+ return STATUS_ERROR;
+ service = unixSocketPath;
}
else
#endif /* HAVE_UNIX_SOCKETS */
/* See if there is still room to add 1 more socket. */
for (; listen_index < MaxListen; listen_index++)
{
- if (ListenSocket[listen_index] == -1)
+ if (ListenSocket[listen_index] == PGINVALID_SOCKET)
break;
}
if (listen_index >= MaxListen)
break;
}
- if ((fd = socket(addr->ai_family, SOCK_STREAM, 0)) < 0)
+ if ((fd = socket(addr->ai_family, SOCK_STREAM, 0)) == PGINVALID_SOCKET)
{
ereport(LOG,
(errcode_for_socket_access(),
/*
* Note: This might fail on some OS's, like Linux older than
* 2.4.21-pre3, that don't have the IPV6_V6ONLY socket option, and map
- * ipv4 addresses to ipv6. It will show ::ffff:ipv4 for all ipv4
+ * ipv4 addresses to ipv6. It will show ::ffff:ipv4 for all ipv4
* connections.
*/
err = bind(fd, addr->ai_addr, addr->ai_addrlen);
(IS_AF_UNIX(addr->ai_family)) ?
errhint("Is another postmaster already running on port %d?"
" If not, remove socket file \"%s\" and retry.",
- (int) portNumber, sock_path) :
+ (int) portNumber, service) :
errhint("Is another postmaster already running on port %d?"
" If not, wait a few seconds and retry.",
(int) portNumber)));
#ifdef HAVE_UNIX_SOCKETS
if (addr->ai_family == AF_UNIX)
{
- if (Setup_AF_UNIX() != STATUS_OK)
+ if (Setup_AF_UNIX(service) != STATUS_OK)
{
closesocket(fd);
break;
* Lock_AF_UNIX -- configure unix socket file path
*/
static int
-Lock_AF_UNIX(unsigned short portNumber, char *unixSocketName)
+Lock_AF_UNIX(char *unixSocketDir, char *unixSocketPath)
{
- UNIXSOCK_PATH(sock_path, portNumber, unixSocketName);
-
/*
* Grab an interlock file associated with the socket file.
+ *
+ * Note: there are two reasons for using a socket lock file, rather than
+ * trying to interlock directly on the socket itself. First, it's a lot
+ * more portable, and second, it lets us remove any pre-existing socket
+ * file without race conditions.
*/
- CreateSocketLockFile(sock_path, true);
+ CreateSocketLockFile(unixSocketPath, true, unixSocketDir);
/*
* Once we have the interlock, we can safely delete any pre-existing
* socket file to avoid failure at bind() time.
*/
- unlink(sock_path);
+ unlink(unixSocketPath);
+
+ /*
+ * Arrange to unlink the socket file(s) at proc_exit. If this is the
+ * first one, set up the on_proc_exit function to do it; then add this
+ * socket file to the list of files to unlink.
+ */
+ if (sock_paths == NIL)
+ on_proc_exit(StreamDoUnlink, 0);
+
+ sock_paths = lappend(sock_paths, pstrdup(unixSocketPath));
return STATUS_OK;
}
* Setup_AF_UNIX -- configure unix socket permissions
*/
static int
-Setup_AF_UNIX(void)
+Setup_AF_UNIX(char *sock_path)
{
- /* Arrange to unlink the socket file at exit */
- on_proc_exit(StreamDoUnlink, 0);
-
/*
* Fix socket ownership/permission if requested. Note we must do this
* before we listen() to avoid a window where unwanted connections could
* RETURNS: STATUS_OK or STATUS_ERROR
*/
int
-StreamConnection(int server_fd, Port *port)
+StreamConnection(pgsocket server_fd, Port *port)
{
/* accept connection and fill in the client (remote) address */
port->raddr.salen = sizeof(port->raddr.addr);
if ((port->sock = accept(server_fd,
(struct sockaddr *) & port->raddr.addr,
- &port->raddr.salen)) < 0)
+ &port->raddr.salen)) == PGINVALID_SOCKET)
{
ereport(LOG,
(errcode_for_socket_access(),
* This is a Win32 socket optimization. The ideal size is 32k.
* http://support.microsoft.com/kb/823764/EN-US/
*/
- on = PQ_BUFFER_SIZE * 4;
+ on = PQ_SEND_BUFFER_SIZE * 4;
if (setsockopt(port->sock, SOL_SOCKET, SO_SNDBUF, (char *) &on,
sizeof(on)) < 0)
{
* we do NOT want to send anything to the far end.
*/
void
-StreamClose(int sock)
+StreamClose(pgsocket sock)
{
closesocket(sock);
}
/*
- * TouchSocketFile -- mark socket file as recently accessed
+ * TouchSocketFiles -- mark socket files as recently accessed
*
* This routine should be called every so often to ensure that the socket
- * file has a recent mod date (ordinary operations on sockets usually won't
- * change the mod date). That saves it from being removed by
+ * files have a recent mod date (ordinary operations on sockets usually won't
+ * change the mod date). That saves them from being removed by
* overenthusiastic /tmp-directory-cleaner daemons. (Another reason we should
* never have put the socket file in /tmp...)
*/
void
-TouchSocketFile(void)
+TouchSocketFiles(void)
{
- /* Do nothing if we did not create a socket... */
- if (sock_path[0] != '\0')
+ ListCell *l;
+
+ /* Loop through all created sockets... */
+ foreach(l, sock_paths)
{
+ char *sock_path = (char *) lfirst(l);
+
/*
* utime() is POSIX standard, utimes() is a common alternative. If we
* have neither, there's no way to affect the mod or access time of
* --------------------------------
*/
+/* --------------------------------
+ * socket_set_nonblocking - set socket blocking/non-blocking
+ *
+ * Sets the socket non-blocking if nonblocking is TRUE, or sets it
+ * blocking otherwise.
+ * --------------------------------
+ */
+static void
+socket_set_nonblocking(bool nonblocking)
+{
+ if (MyProcPort == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_CONNECTION_DOES_NOT_EXIST),
+ errmsg("there is no client connection")));
+
+ if (MyProcPort->noblock == nonblocking)
+ return;
+
+#ifdef WIN32
+ pgwin32_noblock = nonblocking ? 1 : 0;
+#else
+
+ /*
+ * Use COMMERROR on failure, because ERROR would try to send the error to
+ * the client, which might require changing the mode again, leading to
+ * infinite recursion.
+ */
+ if (nonblocking)
+ {
+ if (!pg_set_noblock(MyProcPort->sock))
+ ereport(COMMERROR,
+ (errmsg("could not set socket to nonblocking mode: %m")));
+ }
+ else
+ {
+ if (!pg_set_block(MyProcPort->sock))
+ ereport(COMMERROR,
+ (errmsg("could not set socket to blocking mode: %m")));
+ }
+#endif
+ MyProcPort->noblock = nonblocking;
+}
/* --------------------------------
* pq_recvbuf - load some bytes into the input buffer
PqRecvLength = PqRecvPointer = 0;
}
+ /* Ensure that we're in blocking mode */
+ socket_set_nonblocking(false);
+
/* Can fill buffer from PqRecvLength and upwards */
for (;;)
{
int r;
r = secure_read(MyProcPort, PqRecvBuffer + PqRecvLength,
- PQ_BUFFER_SIZE - PqRecvLength);
+ PQ_RECV_BUFFER_SIZE - PqRecvLength);
if (r < 0)
{
return (unsigned char) PqRecvBuffer[PqRecvPointer];
}
+/* --------------------------------
+ * pq_getbyte_if_available - get a single byte from connection,
+ * if available
+ *
+ * The received byte is stored in *c. Returns 1 if a byte was read,
+ * 0 if no data was available, or EOF if trouble.
+ * --------------------------------
+ */
+int
+pq_getbyte_if_available(unsigned char *c)
+{
+ int r;
+
+ if (PqRecvPointer < PqRecvLength)
+ {
+ *c = PqRecvBuffer[PqRecvPointer++];
+ return 1;
+ }
+
+ /* Put the socket into non-blocking mode */
+ socket_set_nonblocking(true);
+
+ r = secure_read(MyProcPort, c, 1);
+ if (r < 0)
+ {
+ /*
+ * Ok if no data available without blocking or interrupted (though
+ * EINTR really shouldn't happen with a non-blocking socket). Report
+ * other errors.
+ */
+ if (errno == EAGAIN || errno == EWOULDBLOCK || errno == EINTR)
+ r = 0;
+ else
+ {
+ /*
+ * Careful: an ereport() that tries to write to the client would
+ * cause recursion to here, leading to stack overflow and core
+ * dump! This message must go *only* to the postmaster log.
+ */
+ ereport(COMMERROR,
+ (errcode_for_socket_access(),
+ errmsg("could not receive data from client: %m")));
+ r = EOF;
+ }
+ }
+ else if (r == 0)
+ {
+ /* EOF detected */
+ r = EOF;
+ }
+
+ return r;
+}
+
/* --------------------------------
* pq_getbytes - get a known number of bytes from connection
*
if (len > 0)
{
/*
- * Allocate space for message. If we run out of room (ridiculously
+ * Allocate space for message. If we run out of room (ridiculously
* large message), we will elog(ERROR), but we want to discard the
* message body so as not to lose communication sync.
*/
while (len > 0)
{
/* If buffer is full, then flush it out */
- if (PqSendPointer >= PQ_BUFFER_SIZE)
+ if (PqSendPointer >= PqSendBufferSize)
+ {
+ socket_set_nonblocking(false);
if (internal_flush())
return EOF;
- amount = PQ_BUFFER_SIZE - PqSendPointer;
+ }
+ amount = PqSendBufferSize - PqSendPointer;
if (amount > len)
amount = len;
memcpy(PqSendBuffer + PqSendPointer, s, amount);
}
/* --------------------------------
- * pq_flush - flush pending output
+ * socket_flush - flush pending output
*
* returns 0 if OK, EOF if trouble
* --------------------------------
*/
-int
-pq_flush(void)
+static int
+socket_flush(void)
{
int res;
if (PqCommBusy)
return 0;
PqCommBusy = true;
+ socket_set_nonblocking(false);
res = internal_flush();
PqCommBusy = false;
return res;
}
+/* --------------------------------
+ * internal_flush - flush pending output
+ *
+ * Returns 0 if OK (meaning everything was sent, or operation would block
+ * and the socket is in non-blocking mode), or EOF if trouble.
+ * --------------------------------
+ */
static int
internal_flush(void)
{
static int last_reported_send_errno = 0;
- char *bufptr = PqSendBuffer;
+ char *bufptr = PqSendBuffer + PqSendStart;
char *bufend = PqSendBuffer + PqSendPointer;
while (bufptr < bufend)
if (errno == EINTR)
continue; /* Ok if we were interrupted */
+ /*
+ * Ok if no data writable without blocking, and the socket is in
+ * non-blocking mode.
+ */
+ if (errno == EAGAIN ||
+ errno == EWOULDBLOCK)
+ {
+ return 0;
+ }
+
/*
* Careful: an ereport() that tries to write to the client would
* cause recursion to here, leading to stack overflow and core
/*
* We drop the buffered data anyway so that processing can
- * continue, even though we'll probably quit soon.
+ * continue, even though we'll probably quit soon. We also set a
+ * flag that'll cause the next CHECK_FOR_INTERRUPTS to terminate
+ * the connection.
*/
- PqSendPointer = 0;
+ PqSendStart = PqSendPointer = 0;
+ ClientConnectionLost = 1;
+ InterruptPending = 1;
return EOF;
}
last_reported_send_errno = 0; /* reset after any successful send */
bufptr += r;
+ PqSendStart += r;
}
- PqSendPointer = 0;
+ PqSendStart = PqSendPointer = 0;
return 0;
}
+/* --------------------------------
+ * pq_flush_if_writable - flush pending output if writable without blocking
+ *
+ * Returns 0 if OK, or EOF if trouble.
+ * --------------------------------
+ */
+static int
+socket_flush_if_writable(void)
+{
+ int res;
+
+ /* Quick exit if nothing to do */
+ if (PqSendPointer == PqSendStart)
+ return 0;
+
+ /* No-op if reentrant call */
+ if (PqCommBusy)
+ return 0;
+
+ /* Temporarily put the socket into non-blocking mode */
+ socket_set_nonblocking(true);
+
+ PqCommBusy = true;
+ res = internal_flush();
+ PqCommBusy = false;
+ return res;
+}
+
+/* --------------------------------
+ * socket_is_send_pending - is there any pending data in the output buffer?
+ * --------------------------------
+ */
+static bool
+socket_is_send_pending(void)
+{
+ return (PqSendStart < PqSendPointer);
+}
/* --------------------------------
* Message-level I/O routines begin here.
/* --------------------------------
- * pq_putmessage - send a normal message (suppressed in COPY OUT mode)
+ * socket_putmessage - send a normal message (suppressed in COPY OUT mode)
*
* If msgtype is not '\0', it is a message type code to place before
* the message body. If msgtype is '\0', then the message has no type
* returns 0 if OK, EOF if trouble
* --------------------------------
*/
-int
-pq_putmessage(char msgtype, const char *s, size_t len)
+static int
+socket_putmessage(char msgtype, const char *s, size_t len)
{
if (DoingCopyOut || PqCommBusy)
return 0;
}
/* --------------------------------
- * pq_startcopyout - inform libpq that an old-style COPY OUT transfer
+ * pq_putmessage_noblock - like pq_putmessage, but never blocks
+ *
+ * If the output buffer is too small to hold the message, the buffer
+ * is enlarged.
+ */
+static void
+socket_putmessage_noblock(char msgtype, const char *s, size_t len)
+{
+ int res PG_USED_FOR_ASSERTS_ONLY;
+ int required;
+
+ /*
+ * Ensure we have enough space in the output buffer for the message header
+ * as well as the message itself.
+ */
+ required = PqSendPointer + 1 + 4 + len;
+ if (required > PqSendBufferSize)
+ {
+ PqSendBuffer = repalloc(PqSendBuffer, required);
+ PqSendBufferSize = required;
+ }
+ res = pq_putmessage(msgtype, s, len);
+ Assert(res == 0); /* should not fail when the message fits in
+ * buffer */
+}
+
+
+/* --------------------------------
+ * socket_startcopyout - inform libpq that an old-style COPY OUT transfer
* is beginning
* --------------------------------
*/
-void
-pq_startcopyout(void)
+static void
+socket_startcopyout(void)
{
DoingCopyOut = true;
}
/* --------------------------------
- * pq_endcopyout - end an old-style COPY OUT transfer
+ * socket_endcopyout - end an old-style COPY OUT transfer
*
* If errorAbort is indicated, we are aborting a COPY OUT due to an error,
* and must send a terminator line. Since a partial data line might have
* not allow binary transfers, so a textual terminator is always correct.
* --------------------------------
*/
-void
-pq_endcopyout(bool errorAbort)
+static void
+socket_endcopyout(bool errorAbort)
{
if (!DoingCopyOut)
return;
DoingCopyOut = false;
}
-
/*
* Support for TCP Keepalive parameters
*/
+/*
+ * On Windows, we need to set both idle and interval at the same time.
+ * We also cannot reset them to the default (setting to zero will
+ * actually set them to zero, not default), therefor we fallback to
+ * the out-of-the-box default instead.
+ */
+#if defined(WIN32) && defined(SIO_KEEPALIVE_VALS)
+static int
+pq_setkeepaliveswin32(Port *port, int idle, int interval)
+{
+ struct tcp_keepalive ka;
+ DWORD retsize;
+
+ if (idle <= 0)
+ idle = 2 * 60 * 60; /* default = 2 hours */
+ if (interval <= 0)
+ interval = 1; /* default = 1 second */
+
+ ka.onoff = 1;
+ ka.keepalivetime = idle * 1000;
+ ka.keepaliveinterval = interval * 1000;
+
+ if (WSAIoctl(port->sock,
+ SIO_KEEPALIVE_VALS,
+ (LPVOID) &ka,
+ sizeof(ka),
+ NULL,
+ 0,
+ &retsize,
+ NULL,
+ NULL)
+ != 0)
+ {
+ elog(LOG, "WSAIoctl(SIO_KEEPALIVE_VALS) failed: %ui",
+ WSAGetLastError());
+ return STATUS_ERROR;
+ }
+ if (port->keepalives_idle != idle)
+ port->keepalives_idle = idle;
+ if (port->keepalives_interval != interval)
+ port->keepalives_interval = interval;
+ return STATUS_OK;
+}
+#endif
+
int
pq_getkeepalivesidle(Port *port)
{
-#ifdef TCP_KEEPIDLE
+#if defined(TCP_KEEPIDLE) || defined(TCP_KEEPALIVE) || defined(WIN32)
if (port == NULL || IS_AF_UNIX(port->laddr.addr.ss_family))
return 0;
if (port->default_keepalives_idle == 0)
{
+#ifndef WIN32
ACCEPT_TYPE_ARG3 size = sizeof(port->default_keepalives_idle);
+#ifdef TCP_KEEPIDLE
if (getsockopt(port->sock, IPPROTO_TCP, TCP_KEEPIDLE,
(char *) &port->default_keepalives_idle,
&size) < 0)
elog(LOG, "getsockopt(TCP_KEEPIDLE) failed: %m");
port->default_keepalives_idle = -1; /* don't know */
}
+#else
+ if (getsockopt(port->sock, IPPROTO_TCP, TCP_KEEPALIVE,
+ (char *) &port->default_keepalives_idle,
+ &size) < 0)
+ {
+ elog(LOG, "getsockopt(TCP_KEEPALIVE) failed: %m");
+ port->default_keepalives_idle = -1; /* don't know */
+ }
+#endif /* TCP_KEEPIDLE */
+#else /* WIN32 */
+ /* We can't get the defaults on Windows, so return "don't know" */
+ port->default_keepalives_idle = -1;
+#endif /* WIN32 */
}
return port->default_keepalives_idle;
if (port == NULL || IS_AF_UNIX(port->laddr.addr.ss_family))
return STATUS_OK;
-#ifdef TCP_KEEPIDLE
+#if defined(TCP_KEEPIDLE) || defined(TCP_KEEPALIVE) || defined(SIO_KEEPALIVE_VALS)
if (idle == port->keepalives_idle)
return STATUS_OK;
+#ifndef WIN32
if (port->default_keepalives_idle <= 0)
{
if (pq_getkeepalivesidle(port) < 0)
if (idle == 0)
idle = port->default_keepalives_idle;
+#ifdef TCP_KEEPIDLE
if (setsockopt(port->sock, IPPROTO_TCP, TCP_KEEPIDLE,
(char *) &idle, sizeof(idle)) < 0)
{
elog(LOG, "setsockopt(TCP_KEEPIDLE) failed: %m");
return STATUS_ERROR;
}
+#else
+ if (setsockopt(port->sock, IPPROTO_TCP, TCP_KEEPALIVE,
+ (char *) &idle, sizeof(idle)) < 0)
+ {
+ elog(LOG, "setsockopt(TCP_KEEPALIVE) failed: %m");
+ return STATUS_ERROR;
+ }
+#endif
port->keepalives_idle = idle;
-#else
+#else /* WIN32 */
+ return pq_setkeepaliveswin32(port, idle, port->keepalives_interval);
+#endif
+#else /* TCP_KEEPIDLE || SIO_KEEPALIVE_VALS */
if (idle != 0)
{
- elog(LOG, "setsockopt(TCP_KEEPIDLE) not supported");
+ elog(LOG, "setting the keepalive idle time is not supported");
return STATUS_ERROR;
}
#endif
-
return STATUS_OK;
}
int
pq_getkeepalivesinterval(Port *port)
{
-#ifdef TCP_KEEPINTVL
+#if defined(TCP_KEEPINTVL) || defined(SIO_KEEPALIVE_VALS)
if (port == NULL || IS_AF_UNIX(port->laddr.addr.ss_family))
return 0;
if (port->default_keepalives_interval == 0)
{
+#ifndef WIN32
ACCEPT_TYPE_ARG3 size = sizeof(port->default_keepalives_interval);
if (getsockopt(port->sock, IPPROTO_TCP, TCP_KEEPINTVL,
elog(LOG, "getsockopt(TCP_KEEPINTVL) failed: %m");
port->default_keepalives_interval = -1; /* don't know */
}
+#else
+ /* We can't get the defaults on Windows, so return "don't know" */
+ port->default_keepalives_interval = -1;
+#endif /* WIN32 */
}
return port->default_keepalives_interval;
if (port == NULL || IS_AF_UNIX(port->laddr.addr.ss_family))
return STATUS_OK;
-#ifdef TCP_KEEPINTVL
+#if defined(TCP_KEEPINTVL) || defined (SIO_KEEPALIVE_VALS)
if (interval == port->keepalives_interval)
return STATUS_OK;
+#ifndef WIN32
if (port->default_keepalives_interval <= 0)
{
if (pq_getkeepalivesinterval(port) < 0)
}
port->keepalives_interval = interval;
+#else /* WIN32 */
+ return pq_setkeepaliveswin32(port, port->keepalives_idle, interval);
+#endif
#else
if (interval != 0)
{