Create an internal semaphore API that is not tied to SysV semaphores.

As proof of concept, provide an alternate implementation based on POSIX
semaphores.  Also push the SysV shared-memory implementation into a
separate file so that it can be replaced conveniently.

8 files changed, 152 insertions, 127 deletions

diff --git a/src/include/miscadmin.h b/src/include/miscadmin.h
index 99f7fae88bf..09e1c0fe635 100644
--- a/src/include/miscadmin.h
+++ b/src/include/miscadmin.h
@@ -12,7 +12,7 @@
  * Portions Copyright (c) 1996-2001, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $Id: miscadmin.h,v 1.102 2002/04/04 04:25:51 momjian Exp $
+ * $Id: miscadmin.h,v 1.103 2002/05/05 00:03:29 tgl Exp $
  *
  * NOTES
  *	  some of the information in this file should be moved to
@@ -23,7 +23,8 @@
 #ifndef MISCADMIN_H
 #define MISCADMIN_H
 
-#include "storage/ipc.h"
+#include <sys/types.h>
+
 
 /*****************************************************************************
  *	  System interrupt and critical section handling
@@ -291,8 +292,8 @@ extern void BaseInit(void);
 extern bool CreateDataDirLockFile(const char *datadir, bool amPostmaster);
 extern bool CreateSocketLockFile(const char *socketfile, bool amPostmaster);
 extern void TouchSocketLockFile(void);
-extern void RecordSharedMemoryInLockFile(IpcMemoryKey shmKey,
-							 IpcMemoryId shmId);
+extern void RecordSharedMemoryInLockFile(unsigned long id1,
+										 unsigned long id2);
 
 extern void ValidatePgVersion(const char *path);
 
diff --git a/src/include/pg_config.h.in b/src/include/pg_config.h.in
index 80b904181d3..b3a5b8d6e1b 100644
--- a/src/include/pg_config.h.in
+++ b/src/include/pg_config.h.in
@@ -8,7 +8,7 @@
  * or in pg_config.h afterwards.  Of course, if you edit pg_config.h, then your
  * changes will be overwritten the next time you run configure.
  *
- * $Id: pg_config.h.in,v 1.23 2002/04/21 19:48:19 thomas Exp $
+ * $Id: pg_config.h.in,v 1.24 2002/05/05 00:03:29 tgl Exp $
  */
 
 #ifndef PG_CONFIG_H
@@ -711,6 +711,15 @@ extern int fdatasync(int fildes);
 /* Define if you have syslog() */
 #undef HAVE_SYSLOG
 
+/* Define exactly one of these symbols to select semaphore implementation */
+#undef USE_NAMED_POSIX_SEMAPHORES
+#undef USE_UNNAMED_POSIX_SEMAPHORES
+#undef USE_SYSV_SEMAPHORES
+
+/* Define exactly one of these symbols to select shared-mem implementation */
+#undef USE_SYSV_SHARED_MEMORY
+
+
 /*
  *------------------------------------------------------------------------
  * Part 4: pull in system-specific declarations.
diff --git a/src/include/storage/ipc.h b/src/include/storage/ipc.h
index 6a96ea2d125..5ebee36b9f7 100644
--- a/src/include/storage/ipc.h
+++ b/src/include/storage/ipc.h
@@ -3,63 +3,21 @@
  * ipc.h
  *	  POSTGRES inter-process communication definitions.
  *
+ * This file is misnamed, as it no longer has much of anything directly
+ * to do with IPC.  The functionality here is concerned with managing
+ * exit-time cleanup for either a postmaster or a backend.
+ *
  *
  * Portions Copyright (c) 1996-2001, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $Id: ipc.h,v 1.54 2001/11/05 17:46:35 momjian Exp $
+ * $Id: ipc.h,v 1.55 2002/05/05 00:03:29 tgl Exp $
  *
- * Some files that would normally need to include only sys/ipc.h must
- * instead include this file because on Ultrix, sys/ipc.h is not designed
- * to be included multiple times.  This file (by virtue of the ifndef IPC_H)
- * is.
  *-------------------------------------------------------------------------
  */
 #ifndef IPC_H
 #define IPC_H
 
-#include <sys/types.h>
-#ifdef HAVE_SYS_IPC_H
-#include <sys/ipc.h>
-#endif   /* HAVE_SYS_IPC_H */
-
-#ifndef HAVE_UNION_SEMUN
-union semun
-{
-	int			val;
-	struct semid_ds *buf;
-	unsigned short *array;
-};
-#endif
-
-
-/* generic IPC definitions */
-
-#define IPCProtection	(0600)	/* access/modify by user only */
-
-/* semaphore definitions */
-
-typedef uint32 IpcSemaphoreKey; /* semaphore key passed to semget(2) */
-typedef int IpcSemaphoreId;		/* semaphore ID returned by semget(2) */
-
-#define IPC_NMAXSEM		32		/* maximum number of semaphores per semID */
-
-#define PGSemaMagic  537		/* must be less than SEMVMX */
-
-/* shared memory definitions */
-
-typedef uint32 IpcMemoryKey;	/* shared memory key passed to shmget(2) */
-typedef int IpcMemoryId;		/* shared memory ID returned by shmget(2) */
-
-typedef struct PGShmemHeader	/* standard header for all Postgres shmem */
-{
-	int32		magic;			/* magic # to identify Postgres segments */
-#define PGShmemMagic  679834892
-	pid_t		creatorPID;		/* PID of creating process */
-	uint32		totalsize;		/* total size of segment */
-	uint32		freeoffset;		/* offset to first free space */
-} PGShmemHeader;
-
 
 /* ipc.c */
 extern bool proc_exit_inprogress;
@@ -70,24 +28,9 @@ extern void on_proc_exit(void (*function) (), Datum arg);
 extern void on_shmem_exit(void (*function) (), Datum arg);
 extern void on_exit_reset(void);
 
-extern void IpcInitKeyAssignment(int port);
-
-extern IpcSemaphoreId IpcSemaphoreCreate(int numSems, int permission,
-				   int semStartValue,
-				   bool removeOnExit);
-extern void IpcSemaphoreKill(IpcSemaphoreId semId);
-extern void IpcSemaphoreLock(IpcSemaphoreId semId, int sem, bool interruptOK);
-extern void IpcSemaphoreUnlock(IpcSemaphoreId semId, int sem);
-extern bool IpcSemaphoreTryLock(IpcSemaphoreId semId, int sem);
-extern int	IpcSemaphoreGetValue(IpcSemaphoreId semId, int sem);
-
-extern PGShmemHeader *IpcMemoryCreate(uint32 size, bool makePrivate,
-				int permission);
-
-extern bool SharedMemoryIsInUse(IpcMemoryKey shmKey, IpcMemoryId shmId);
-
 /* ipci.c */
 extern void CreateSharedMemoryAndSemaphores(bool makePrivate,
-								int maxBackends);
+											int maxBackends,
+											int port);
 
 #endif   /* IPC_H */
diff --git a/src/include/storage/pg_sema.h b/src/include/storage/pg_sema.h
new file mode 100644
index 00000000000..65c1936c29f
--- /dev/null
+++ b/src/include/storage/pg_sema.h
@@ -0,0 +1,76 @@
+/*-------------------------------------------------------------------------
+ *
+ * pg_sema.h
+ *	  Platform-independent API for semaphores.
+ *
+ * PostgreSQL requires counting semaphores (the kind that keep track of
+ * multiple unlock operations, and will allow an equal number of subsequent
+ * lock operations before blocking).  The underlying implementation is
+ * not the same on every platform.  This file defines the API that must
+ * be provided by each port.
+ *
+ *
+ * Portions Copyright (c) 1996-2001, PostgreSQL Global Development Group
+ * Portions Copyright (c) 1994, Regents of the University of California
+ *
+ * $Id: pg_sema.h,v 1.1 2002/05/05 00:03:29 tgl Exp $
+ *
+ *-------------------------------------------------------------------------
+ */
+#ifndef PG_SEMA_H
+#define PG_SEMA_H
+
+/*
+ * PGSemaphoreData and pointer type PGSemaphore are the data structure
+ * representing an individual semaphore.  The contents of PGSemaphoreData
+ * vary across implementations and must never be touched by platform-
+ * independent code.  PGSemaphoreData structures are always allocated
+ * in shared memory (to support implementations where the data changes during
+ * lock/unlock).
+ *
+ * pg_config.h must define exactly one of the USE_xxx_SEMAPHORES symbols.
+ */
+
+#ifdef USE_NAMED_POSIX_SEMAPHORES
+
+#include <semaphore.h>
+
+typedef sem_t *PGSemaphoreData;
+
+#endif
+
+#ifdef USE_UNNAMED_POSIX_SEMAPHORES
+
+#include <semaphore.h>
+
+typedef sem_t PGSemaphoreData;
+
+#endif
+
+#ifdef USE_SYSV_SEMAPHORES
+
+typedef struct PGSemaphoreData
+{
+	int			semId;			/* semaphore set identifier */
+	int			semNum;			/* semaphore number within set */
+} PGSemaphoreData;
+
+#endif
+
+typedef PGSemaphoreData *PGSemaphore;
+
+
+/* Module initialization (called during postmaster start or shmem reinit) */
+extern void PGReserveSemaphores(int maxSemas, int port);
+/* Initialize a PGSemaphore structure to represent a sema with count 1 */
+extern void PGSemaphoreCreate(PGSemaphore sema);
+/* Reset a previously-initialized PGSemaphore to have count 0 */
+extern void PGSemaphoreReset(PGSemaphore sema);
+/* Lock a semaphore (decrement count), blocking if count would be < 0 */
+extern void PGSemaphoreLock(PGSemaphore sema, bool interruptOK);
+/* Unlock a semaphore (increment count) */
+extern void PGSemaphoreUnlock(PGSemaphore sema);
+/* Lock a semaphore only if able to do so without blocking */
+extern bool PGSemaphoreTryLock(PGSemaphore sema);
+
+#endif   /* PG_SEMA_H */
diff --git a/src/include/storage/pg_shmem.h b/src/include/storage/pg_shmem.h
new file mode 100644
index 00000000000..fa35e63f725
--- /dev/null
+++ b/src/include/storage/pg_shmem.h
@@ -0,0 +1,44 @@
+/*-------------------------------------------------------------------------
+ *
+ * pg_shmem.h
+ *	  Platform-independent API for shared memory support.
+ *
+ * Every port is expected to support shared memory with approximately
+ * SysV-ish semantics; in particular, a memory block is not anonymous
+ * but has an ID, and we must be able to tell whether there are any
+ * remaining processes attached to a block of a specified ID.
+ *
+ * To simplify life for the SysV implementation, the ID is assumed to
+ * consist of two unsigned long values (these are key and ID in SysV
+ * terms).  Other platforms may ignore the second value if they need
+ * only one ID number.
+ *
+ *
+ * Portions Copyright (c) 1996-2001, PostgreSQL Global Development Group
+ * Portions Copyright (c) 1994, Regents of the University of California
+ *
+ * $Id: pg_shmem.h,v 1.1 2002/05/05 00:03:29 tgl Exp $
+ *
+ *-------------------------------------------------------------------------
+ */
+#ifndef PG_SHMEM_H
+#define PG_SHMEM_H
+
+#include <sys/types.h>
+
+
+typedef struct PGShmemHeader	/* standard header for all Postgres shmem */
+{
+	int32		magic;			/* magic # to identify Postgres segments */
+#define PGShmemMagic  679834892
+	pid_t		creatorPID;		/* PID of creating process */
+	uint32		totalsize;		/* total size of segment */
+	uint32		freeoffset;		/* offset to first free space */
+} PGShmemHeader;
+
+
+extern PGShmemHeader *PGSharedMemoryCreate(uint32 size, bool makePrivate,
+										   int port);
+extern bool PGSharedMemoryIsInUse(unsigned long id1, unsigned long id2);
+
+#endif   /* PG_SHMEM_H */
diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h
index e1700b23727..e276fddf9cd 100644
--- a/src/include/storage/proc.h
+++ b/src/include/storage/proc.h
@@ -7,7 +7,7 @@
  * Portions Copyright (c) 1996-2001, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $Id: proc.h,v 1.54 2001/11/06 00:38:26 tgl Exp $
+ * $Id: proc.h,v 1.55 2002/05/05 00:03:29 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -16,16 +16,10 @@
 
 #include "access/xlog.h"
 #include "storage/backendid.h"
-#include "storage/ipc.h"
 #include "storage/lock.h"
+#include "storage/pg_sema.h"
 
 
-typedef struct
-{
-	IpcSemaphoreId semId;		/* SysV semaphore set ID */
-	int			semNum;			/* semaphore number within set */
-} SEMA;
-
 /*
  * Each backend has a PROC struct in shared memory.  There is also a list of
  * currently-unused PROC structs that will be reallocated to new backends.
@@ -39,7 +33,7 @@ struct PROC
 	/* proc->links MUST BE FIRST IN STRUCT (see ProcSleep,ProcWakeup,etc) */
 	SHM_QUEUE	links;			/* list link if process is in a list */
 
-	SEMA		sem;			/* ONE semaphore to sleep on */
+	PGSemaphoreData sem;		/* ONE semaphore to sleep on */
 	int			errType;		/* STATUS_OK or STATUS_ERROR after wakeup */
 
 	TransactionId xid;			/* transaction currently being executed by
@@ -84,47 +78,11 @@ extern PROC *MyProc;
 
 /*
  * There is one ProcGlobal struct for the whole installation.
- *
- * PROC_NSEMS_PER_SET is the number of semaphores in each sys-V semaphore set
- * we allocate.  It must be no more than 32 (or however many bits in an int
- * on your machine), or our free-semaphores bitmap won't work.  It also must
- * be *less than* your kernel's SEMMSL (max semaphores per set) parameter,
- * which is often around 25.  (Less than, because we allocate one extra sema
- * in each set for identification purposes.)
- *
- * PROC_SEM_MAP_ENTRIES is the number of semaphore sets we need to allocate
- * to keep track of up to maxBackends backends.  Note that we need one extra
- * semaphore (see storage/lmgr/proc.c), so the computation may look wrong,
- * but it's right.
  */
-#define  PROC_NSEMS_PER_SET		16
-#define  PROC_SEM_MAP_ENTRIES(maxBackends)	((maxBackends)/PROC_NSEMS_PER_SET+1)
-
-typedef struct
-{
-	/* info about a single set of per-process semaphores */
-	IpcSemaphoreId procSemId;
-	int32		freeSemMap;
-
-	/*
-	 * In freeSemMap, bit i is set if the i'th semaphore of this sema set
-	 * is allocated to a process.  (i counts from 0 at the LSB)
-	 */
-} SEM_MAP_ENTRY;
-
 typedef struct PROC_HDR
 {
 	/* Head of list of free PROC structures */
 	SHMEM_OFFSET freeProcs;
-
-	/* Info about semaphore sets used for per-process semaphores */
-	int			semMapEntries;
-
-	/*
-	 * VARIABLE LENGTH ARRAY: actual length is semMapEntries. THIS MUST BE
-	 * LAST IN THE STRUCT DECLARATION.
-	 */
-	SEM_MAP_ENTRY procSemMap[1];
 } PROC_HDR;
 
 
@@ -135,6 +93,7 @@ extern int	DeadlockTimeout;
 /*
  * Function Prototypes
  */
+extern int	ProcGlobalSemas(int maxBackends);
 extern void InitProcGlobal(int maxBackends);
 extern void InitProcess(void);
 extern void InitDummyProcess(void);
diff --git a/src/include/storage/s_lock.h b/src/include/storage/s_lock.h
index be79b51a17b..41a79102d78 100644
--- a/src/include/storage/s_lock.h
+++ b/src/include/storage/s_lock.h
@@ -63,14 +63,14 @@
  * Portions Copyright (c) 1996-2001, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- *	  $Id: s_lock.h,v 1.97 2002/01/29 15:44:42 tgl Exp $
+ *	  $Id: s_lock.h,v 1.98 2002/05/05 00:03:29 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
 #ifndef S_LOCK_H
 #define S_LOCK_H
 
-#include "storage/ipc.h"
+#include "storage/pg_sema.h"
 
 
 #if defined(HAS_TEST_AND_SET)
@@ -438,7 +438,6 @@ extern slock_t wc_tas(volatile slock_t *lock);
  * AIX (POWER)
  *
  * Note that slock_t on POWER/POWER2/PowerPC is int instead of char
- * (see storage/ipc.h).
  */
 #define TAS(lock)	cs((int *) (lock), 0, 1)
 #endif	 /* _AIX */
@@ -462,17 +461,11 @@ extern slock_t wc_tas(volatile slock_t *lock);
 #else							/* !HAS_TEST_AND_SET */
 
 /*
- * Fake spinlock implementation using SysV semaphores --- slow and prone
+ * Fake spinlock implementation using semaphores --- slow and prone
  * to fall foul of kernel limits on number of semaphores, so don't use this
  * unless you must!  The subroutines appear in spin.c.
  */
-
-typedef struct
-{
-	/* reference to semaphore used to implement this spinlock */
-	IpcSemaphoreId semId;
-	int			sem;
-} slock_t;
+typedef PGSemaphoreData slock_t;
 
 extern bool s_lock_free_sema(volatile slock_t *lock);
 extern void s_unlock_sema(volatile slock_t *lock);
diff --git a/src/include/storage/spin.h b/src/include/storage/spin.h
index 6ec0552ca64..8279af0efb0 100644
--- a/src/include/storage/spin.h
+++ b/src/include/storage/spin.h
@@ -40,7 +40,7 @@
  * Portions Copyright (c) 1996-2001, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $Id: spin.h,v 1.19 2001/11/05 17:46:35 momjian Exp $
+ * $Id: spin.h,v 1.20 2002/05/05 00:03:29 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -72,6 +72,6 @@
 #define SpinLockFree(lock)	S_LOCK_FREE(lock)
 
 
-extern void CreateSpinlocks(void);
+extern int	SpinlockSemas(void);
 
 #endif   /* SPIN_H */