7.1 beta 3 CHANGES FOR QNX

Started by Maurizioalmost 25 years ago10 messages

maurizio.c@libero.it

almost 25 years ago

3 attachment(s)

I have compiled 7.1 b3 in QNX but to let postgresql works I changed something.

In src/backend/port/qnx4/sem.c
#define SEMMAX (PROC_NSEMS_PER_SET) ----------------------OLD
#define SEMMAX (PROC_NSEMS_PER_SET + 1) ----------------------NEW

in src/include/storage/s_lock.h
#if defined(__i386__) ----------------------OLD
#if (defined(__i386__) && !defined(__QNX__)) ----------------------NEW

in src/interfaces/ecpg/preproc/pgc.c
#ifndef ECHO ---------------------OLD
#ifndef (ECHO || defined(__QNX__)) ----------------------NEW

Attached are the files

Maurizio Cauci
DREAMTECH di Cauci Maurizio
Via Ronchetti, 2 - 21013 Gallarate (VA)
www.dreamtech-it.com

Attachments:

S_lock.happlication/octet-stream; name=S_lock.hDownload

/*-------------------------------------------------------------------------
 *
 * s_lock.h
 *	   This file contains the in-line portion of the implementation
 *	   of spinlocks.
 *
 * Portions Copyright (c) 1996-2001, PostgreSQL Global Development Group
 * Portions Copyright (c) 1994, Regents of the University of California
 *
 *
 * IDENTIFICATION
 *	  $Header: /home/projects/pgsql/cvsroot/pgsql/src/include/storage/s_lock.h,v 1.86 2001/01/24 19:43:28 momjian Exp $
 *
 *-------------------------------------------------------------------------
 */

/*----------
 * DESCRIPTION
 *	The public macros that must be provided are:
 *
 *	void S_INIT_LOCK(slock_t *lock)
 *		Initialize a spinlock (to the unlocked state).
 *
 *	void S_LOCK(slock_t *lock)
 *		Acquire a spinlock, waiting if necessary.
 *		Time out and abort() if unable to acquire the lock in a
 *		"reasonable" amount of time --- typically ~ 1 minute.
 *
 *	void S_UNLOCK(slock_t *lock)
 *		Unlock a previously acquired lock.
 *
 *	bool S_LOCK_FREE(slock_t *lock)
 *		Tests if the lock is free. Returns TRUE if free, FALSE if locked.
 *		This does *not* change the state of the lock.
 *
 *	int TAS(slock_t *lock)
 *		Atomic test-and-set instruction.  Attempt to acquire the lock,
 *		but do *not* wait.	Returns 0 if successful, nonzero if unable
 *		to acquire the lock.
 *
 *	TAS() is a lower-level part of the API, but is used directly in a
 *	few places that want to do other things while waiting for a lock.
 *	The S_LOCK() macro is equivalent to
 *
 *	void
 *	S_LOCK(slock_t *lock)
 *	{
 *		unsigned	spins = 0;
 *
 *		while (TAS(lock))
 *			S_LOCK_SLEEP(lock, spins++);
 *	}
 *
 *	where S_LOCK_SLEEP() checks for timeout and sleeps for a short
 *	interval.  Callers that want to perform useful work while waiting
 *	can write out this entire loop and insert the "useful work" inside
 *	the loop.
 *
 *	CAUTION to TAS() callers: on some platforms TAS() may sometimes
 *	report failure to acquire a lock even when the lock is not locked.
 *	For example, on Alpha TAS() will "fail" if interrupted.  Therefore
 *	TAS() must *always* be invoked in a retry loop as depicted, even when
 *	you are certain the lock is free.
 *
 *	On most supported platforms, TAS() uses a tas() function written
 *	in assembly language to execute a hardware atomic-test-and-set
 *	instruction.  Equivalent OS-supplied mutex routines could be used too.
 *
 *	If no system-specific TAS() is available (ie, HAS_TEST_AND_SET is not
 *	defined), then we fall back on an emulation that uses SysV semaphores.
 *	This emulation will be MUCH MUCH MUCH slower than a proper TAS()
 *	implementation, because of the cost of a kernel call per lock or unlock.
 *	An old report is that Postgres spends around 40% of its time in semop(2)
 *	when using the SysV semaphore code.
 *
 *	Note to implementors: there are default implementations for all these
 *	macros at the bottom of the file.  Check if your platform can use
 *	these or needs to override them.
 *----------
 */
#ifndef S_LOCK_H
#define S_LOCK_H

#include "storage/ipc.h"

/* Platform-independent out-of-line support routines */
//extern void s_lock(volatile slock_t *lock,
extern void s_lock(sem_t *lock,
	   const char *file, const int line);
extern void s_lock_sleep(unsigned spins, int microsec,
			 volatile slock_t *lock,
			 const char *file, const int line);


#if defined(HAS_TEST_AND_SET)


#if defined(__GNUC__)
/*************************************************************************
 * All the gcc inlines
 */

/*
 * Standard gcc asm format:
 *
	__asm__ __volatile__(
		"	command	\n"
		"	command	\n"
		"	command	\n"
:		"=r"(_res)			return value, in register
:		"r"(lock)			argument, 'lock pointer', in register
:		"r0");				inline code uses this register
 */


#if (defined(__i386__) && !defined(__QNX__))
#define TAS(lock) tas(lock)

static __inline__ int
tas(volatile slock_t *lock)
{
//	int _res = 1;
	register slock_t _res = 1;

	__asm__ __volatile__(
		"	lock			\n"
		"	xchgb	%0,%1	\n"
:		"=q"(_res), "=m"(*lock)
:		"0"(_res));
	return (int) _res;
}

#endif	 /* __i386__ */


#ifdef __ia64__
#define TAS(lock) tas(lock)

static __inline__ int
tas(volatile slock_t *lock)
{
	long int	ret;

	__asm__ __volatile__(
		"	xchg4 	%0=%1,%2	\n"
:		"=r"(ret), "=m"(*lock)
:		"r"(1), "1"(*lock)
:		"memory");

	return (int) ret;
}

#endif	 /* __ia64__ */


#if defined(__arm__) || defined(__arm__)
#define TAS(lock) tas(lock)

static __inline__ int
tas(volatile slock_t *lock)
{
	register slock_t _res = 1;

	__asm__ __volatile__(
		"	swpb 	%0, %0, [%3]	\n"
:		"=r"(_res), "=m"(*lock)
:		"0"(_res), "r"(lock));
	return (int) _res;
}

#endif	 /* __arm__ */

#if defined(__s390__)
/*
 * S/390 Linux
 */
#define TAS(lock)	   tas(lock)

static inline int
tas(volatile slock_t *lock)
{
	int			_res;

	__asm__	__volatile__(
		"	la	1,1			\n"
		"	l 	2,%2		\n"
		"	slr 0,0			\n"
		"	cs 	0,1,0(2)	\n"
		"	lr 	%1,0		\n"
:		"=m"(lock), "=d"(_res)
:		"m"(lock)
:		"0", "1", "2");

	return (_res);
}

#endif	 /* __s390__ */


#if defined(__sparc__)
#define TAS(lock) tas(lock)

static __inline__ int
tas(volatile slock_t *lock)
{
	register slock_t _res = 1;

	__asm__ __volatile__(
		"	ldstub	[%2], %0	\n"
:		"=r"(_res), "=m"(*lock)
:		"r"(lock));
	return (int) _res;
}

#endif	 /* __sparc__ */


#if defined(__mc68000__) && defined(__linux__)
#define TAS(lock) tas(lock)

static __inline__ int
tas(volatile slock_t *lock)
{
	register int rv;

	__asm__	__volatile__(
		"	tas	%1		\n"
		"	sne	%0		\n"
:		"=d"(rv), "=m"(*lock)
:		"1"(*lock)
:		"cc");

	return rv;
}

#endif	 /* defined(__mc68000__) && defined(__linux__) */


#if defined(NEED_VAX_TAS_ASM)
/*
 * VAXen -- even multiprocessor ones
 * (thanks to Tom Ivar Helbekkmo)
 */
#define TAS(lock) tas(lock)

static __inline__ int
tas(volatile slock_t *lock)
{
	register	_res;

	__asm__ __volatile__(
		"	movl 	$1, r0			\n"
		"	bbssi	$0, (%1), 1f	\n"
		"	clrl	r0				\n"
		"1:	movl	r0, %0			\n"
:		"=r"(_res)
:		"r"(lock)
:		"r0");
	return (int) _res;
}

#endif	 /* NEED_VAX_TAS_ASM */


#if defined(NEED_NS32K_TAS_ASM)
#define TAS(lock) tas(lock)

static __inline__ int
tas(volatile slock_t *lock)
{
	register	_res;

	__asm__ __volatile__(
		"	sbitb	0, %0	\n"
		"	sfsd	%1		\n"
:		"=m"(*lock), "=r"(_res));
	return (int) _res;
}

#endif	 /* NEED_NS32K_TAS_ASM */



#else							/* !__GNUC__ */

/***************************************************************************
 * All non-gcc inlines
 */

#if defined(NEED_I386_TAS_ASM) && defined(USE_UNIVEL_CC)
#define TAS(lock)	tas(lock)

asm int
tas(volatile slock_t *s_lock)
{
/* UNIVEL wants %mem in column 1, so we don't pg_indent this file */
%mem s_lock
	pushl %ebx
	movl s_lock, %ebx
	movl $255, %eax
	lock
	xchgb %al, (%ebx)
	popl %ebx
}

#endif	 /* defined(NEED_I386_TAS_ASM) && defined(USE_UNIVEL_CC) */

#endif	 /* defined(__GNUC__) */



/*************************************************************************
 * These are the platforms that do not use inline assembler (and hence
 * have common code for gcc and non-gcc compilers, if both are available).
 */


#if defined(__alpha)

/*
 * Correct multi-processor locking methods are explained in section 5.5.3
 * of the Alpha AXP Architecture Handbook, which at this writing can be
 * found at ftp://ftp.netbsd.org/pub/NetBSD/misc/dec-docs/index.html.
 * For gcc we implement the handbook's code directly with inline assembler.
 */
#if defined(__GNUC__)

#define TAS(lock)  tas(lock)
#define S_UNLOCK(lock)	\
do \
{\
	__asm__ __volatile__ ("	mb \n"); \
	*(lock) = 0; \
} while (0)

static __inline__ int
tas(volatile slock_t *lock)
{
	register slock_t _res;

	__asm__	__volatile__(
		"	ldq		$0, %0	\n"
		"	bne		$0, 2f	\n"
		"	ldq_l	%1, %0	\n"
		"	bne		%1, 2f	\n"
		"	mov		1,  $0	\n"
		"	stq_c	$0, %0	\n"
		"	beq		$0, 2f	\n"
		"	mb				\n"
		"	br		3f		\n"
		"2:	mov		1, %1	\n"
		"3:					\n"
:		"=m"(*lock), "=r"(_res)
:
:		"0");

	return (int) _res;
}

#else							/* !defined(__GNUC__) */

/*
 * The Tru64 compiler doesn't support gcc-style inline asm, but it does
 * have some builtin functions that accomplish much the same results.
 * For simplicity, slock_t is defined as long (ie, quadword) on Alpha
 * regardless of the compiler in use.  LOCK_LONG and UNLOCK_LONG only
 * operate on an int (ie, longword), but that's OK as long as we define
 * S_INIT_LOCK to zero out the whole quadword.
 */

#include <alpha/builtins.h>

#define S_INIT_LOCK(lock)  (*(lock) = 0)
#define TAS(lock)		   (__LOCK_LONG_RETRY((lock), 1) == 0)
#define S_UNLOCK(lock)	   __UNLOCK_LONG(lock)

#endif	 /* defined(__GNUC__) */

#endif	 /* __alpha */


#if defined(__hpux)
/*
 * HP-UX (PA-RISC)
 *
 * Note that slock_t on PA-RISC is a structure instead of char
 * (see include/port/hpux.h).
 *
 * a "set" slock_t has a single word cleared.  a "clear" slock_t has
 * all words set to non-zero. tas() is in tas.s
 */

#define S_UNLOCK(lock) \
	do { \
		volatile slock_t *lock_ = (volatile slock_t *) (lock); \
		lock_->sema[0] = lock_->sema[1] = \
		lock_->sema[2] = lock_->sema[3] = -1; \
	} while (0)

#define S_LOCK_FREE(lock)	( *(int *) (((long) (lock) + 15) & ~15) != 0)

#endif	 /* __hpux */


#if defined(__QNX__)
/*
 * QNX 4
 *
 * Note that slock_t under QNX is sem_t instead of char
 */
#define TAS(lock)		(sem_trywait((lock)) < 0)
#define S_UNLOCK(lock)	sem_post((lock))
#define S_INIT_LOCK(lock)		sem_init((lock), 1, 1)
#define S_LOCK_FREE(lock)		((lock)->value)
#endif	 /* __QNX__ */


#if defined(__sgi)
/*
 * SGI IRIX 5
 * slock_t is defined as a unsigned long. We use the standard SGI
 * mutex API.
 *
 * The following comment is left for historical reasons, but is probably
 * not a good idea since the mutex ABI is supported.
 *
 * This stuff may be supplemented in the future with Masato Kataoka's MIPS-II
 * assembly from his NECEWS SVR4 port, but we probably ought to retain this
 * for the R3000 chips out there.
 */
#include "mutex.h"
#define TAS(lock)	(test_and_set(lock,1))
#define S_UNLOCK(lock)	(test_then_and(lock,0))
#define S_INIT_LOCK(lock)	(test_then_and(lock,0))
#define S_LOCK_FREE(lock)	(test_then_add(lock,0) == 0)
#endif	 /* __sgi */

#if defined(sinix)
/*
 * SINIX / Reliant UNIX
 * slock_t is defined as a struct abilock_t, which has a single unsigned long
 * member. (Basically same as SGI)
 *
 */
#define TAS(lock)	(!acquire_lock(lock))
#define S_UNLOCK(lock)	release_lock(lock)
#define S_INIT_LOCK(lock)	init_lock(lock)
#define S_LOCK_FREE(lock)	(stat_lock(lock) == UNLOCKED)
#endif	 /* sinix */


#if defined(_AIX)
/*
 * 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 */


#if defined (nextstep)
/*
 * NEXTSTEP (mach)
 * slock_t is defined as a struct mutex.
 */

#define S_LOCK(lock)	mutex_lock(lock)
#define S_UNLOCK(lock)	mutex_unlock(lock)
#define S_INIT_LOCK(lock)	mutex_init(lock)
/* For Mach, we have to delve inside the entrails of `struct mutex'.  Ick! */
#define S_LOCK_FREE(alock)	((alock)->lock == 0)
#endif	 /* nextstep */



#else							/* !HAS_TEST_AND_SET */

/*
 * Fake spinlock implementation using SysV semaphores --- slow and prone
 * to fall foul of kernel limits on number of semaphores, so don't use this
 * unless you must!
 */

typedef struct
{
	/* reference to semaphore used to implement this spinlock */
	IpcSemaphoreId semId;
	int			sem;
} slock_t;

extern bool s_lock_free_sema(volatile slock_t *lock);
extern void s_unlock_sema(volatile slock_t *lock);
extern void s_init_lock_sema(volatile slock_t *lock);
extern int	tas_sema(volatile slock_t *lock);

#define S_LOCK_FREE(lock)	s_lock_free_sema(lock)
#define S_UNLOCK(lock)	 s_unlock_sema(lock)
#define S_INIT_LOCK(lock)	s_init_lock_sema(lock)
#define TAS(lock)	tas_sema(lock)

#endif	 /* HAS_TEST_AND_SET */



/****************************************************************************
 * Default Definitions - override these above as needed.
 */

#if !defined(S_LOCK)
#define S_LOCK(lock) \
	do { \
		if (TAS(lock)) \
			s_lock((lock), __FILE__, __LINE__); \
	} while (0)
#endif	 /* S_LOCK */

#if !defined(S_LOCK_SLEEP)
#define S_LOCK_SLEEP(lock,spins) \
	s_lock_sleep((spins), 0, (lock), __FILE__, __LINE__)
#endif	 /* S_LOCK_SLEEP */

#if !defined(S_LOCK_SLEEP_INTERVAL)
#define S_LOCK_SLEEP_INTERVAL(lock,spins,microsec) \
	s_lock_sleep((spins), (microsec), (lock), __FILE__, __LINE__)
#endif	 /* S_LOCK_SLEEP_INTERVAL */

#if !defined(S_LOCK_FREE)
#define S_LOCK_FREE(lock)	(*(lock) == 0)
#endif	 /* S_LOCK_FREE */

#if !defined(S_UNLOCK)
#define S_UNLOCK(lock)		(*(lock) = 0)
#endif	 /* S_UNLOCK */

#if !defined(S_INIT_LOCK)
#define S_INIT_LOCK(lock)	S_UNLOCK(lock)
#endif	 /* S_INIT_LOCK */

#if !defined(TAS)
extern int	tas(volatile slock_t *lock);		/* in port/.../tas.s, or
												 * s_lock.c */

#define TAS(lock)		tas(lock)
#endif	 /* TAS */


#endif	 /* S_LOCK_H */

Pgc.capplication/octet-stream; name=Pgc.cDownload

Sem.capplication/octet-stream; name=Sem.cDownload

Bruce Momjian

pgman@candle.pha.pa.us

almost 25 years ago

In reply to: Maurizio (#1)

1 attachment(s)

Re: 7.1 beta 3 CHANGES FOR QNX

I have applied the following patch for QNX. I had to do the ECHO test
in pgc.l because pgc.c is generated from pgc.l. Can you test this to
see it fixes the problem?

in [ Charset ISO-8859-1 unsupported, converting... ]

I have compiled 7.1 b3 in QNX but to let postgresql works I changed something.

In src/backend/port/qnx4/sem.c
#define SEMMAX (PROC_NSEMS_PER_SET) ----------------------OLD
#define SEMMAX (PROC_NSEMS_PER_SET + 1) ----------------------NEW

in src/include/storage/s_lock.h
#if defined(__i386__) ----------------------OLD
#if (defined(__i386__) && !defined(__QNX__)) ----------------------NEW

in src/interfaces/ecpg/preproc/pgc.c
#ifndef ECHO ---------------------OLD
#ifndef (ECHO || defined(__QNX__)) ----------------------NEW

Attached are the files

Maurizio Cauci
DREAMTECH di Cauci Maurizio
Via Ronchetti, 2 - 21013 Gallarate (VA)
www.dreamtech-it.com

[ Attachment, skipping... ]

-- 
  Bruce Momjian                        |  http://candle.pha.pa.us
  pgman@candle.pha.pa.us               |  (610) 853-3000
  +  If your life is a hard drive,     |  830 Blythe Avenue
  +  Christ can be your backup.        |  Drexel Hill, Pennsylvania 19026

Attachments:

/bjm/difftext/plainDownload

? config.log
? config.cache
? config.status
? GNUmakefile
? src/Makefile.custom
? src/GNUmakefile
? src/Makefile.global
? src/log
? src/crtags
? src/backend/port/Makefile
? src/bin/pg_dump/pg_dump
? src/bin/pg_dump/pg_restore
? src/bin/pg_dump/pg_dumpall
? src/include/config.h
? src/include/stamp-h
? src/interfaces/libpq/libpq.so.2.1
Index: src/backend/port/qnx4/sem.c
===================================================================
RCS file: /home/projects/pgsql/cvsroot/pgsql/src/backend/port/qnx4/sem.c,v
retrieving revision 1.3
diff -c -r1.3 sem.c
*** src/backend/port/qnx4/sem.c	2000/04/12 17:15:30	1.3
--- src/backend/port/qnx4/sem.c	2001/02/02 18:13:12
***************
*** 26,32 ****
  
  
  #define SETMAX	((MAXBACKENDS + PROC_NSEMS_PER_SET - 1) / PROC_NSEMS_PER_SET)
! #define SEMMAX	(PROC_NSEMS_PER_SET)
  #define OPMAX	8
  
  #define MODE	0700
--- 26,32 ----
  
  
  #define SETMAX	((MAXBACKENDS + PROC_NSEMS_PER_SET - 1) / PROC_NSEMS_PER_SET)
! #define SEMMAX	(PROC_NSEMS_PER_SET+1)
  #define OPMAX	8
  
  #define MODE	0700
Index: src/include/storage/s_lock.h
===================================================================
RCS file: /home/projects/pgsql/cvsroot/pgsql/src/include/storage/s_lock.h,v
retrieving revision 1.86
diff -c -r1.86 s_lock.h
*** src/include/storage/s_lock.h	2001/01/24 19:43:28	1.86
--- src/include/storage/s_lock.h	2001/02/02 18:13:13
***************
*** 112,118 ****
   */
  
  
! #if defined(__i386__)
  #define TAS(lock) tas(lock)
  
  static __inline__ int
--- 112,118 ----
   */
  
  
! #if defined(__i386__) && !defined(__QNX__)
  #define TAS(lock) tas(lock)
  
  static __inline__ int
Index: src/interfaces/ecpg/preproc/pgc.l
===================================================================
RCS file: /home/projects/pgsql/cvsroot/pgsql/src/interfaces/ecpg/preproc/pgc.l,v
retrieving revision 1.73
diff -c -r1.73 pgc.l
*** src/interfaces/ecpg/preproc/pgc.l	2001/01/24 19:43:29	1.73
--- src/interfaces/ecpg/preproc/pgc.l	2001/02/02 18:13:17
***************
*** 35,40 ****
--- 35,45 ----
  #undef yywrap
  #endif /* yywrap */
  
+ #ifdef __QNX__
+ /* For some reason, QNX needs this, 2001-02-02 */
+ #define ECHO (void) fwrite( yytext, yyleng, 1, yyout )
+ #endif
+ 
  #define YY_NO_UNPUT
  
  extern YYSTYPE yylval;

Maurizio

maurizio.c@libero.it

almost 25 years ago

In reply to: Bruce Momjian (#2)

Re: 7.1 beta 3 CHANGES FOR QNX

I have tested and works well.

.
----- Original Message -----
From: "Bruce Momjian" <pgman@candle.pha.pa.us>
To: "Maurizio" <maurizio.c@libero.it>
Cc: <pgsql-hackers@postgresql.org>
Sent: Friday, February 02, 2001 7:21 PM
Subject: Re: [HACKERS] 7.1 beta 3 CHANGES FOR QNX

I have applied the following patch for QNX. I had to do the ECHO test
in pgc.l because pgc.c is generated from pgc.l. Can you test this to
see it fixes the problem?

in [ Charset ISO-8859-1 unsupported, converting... ]

I have compiled 7.1 b3 in QNX but to let postgresql works I changed

something.

In src/backend/port/qnx4/sem.c
#define SEMMAX

SEMS_PER_SET) ----------------------OLD

#define SEMMAX (PROC_NSEMS_PER_SET +

----------------------NEW

in src/include/storage/s_lock.h
#if

----------------------OLD

#if (defined(__i386__) &&

----------------------NEW

in src/interfaces/ecpg/preproc/pgc.c
#ifndef

---------------------OLD

#ifndef (ECHO ||

----------------------NEW

Attached are the files

Maurizio Cauci
DREAMTECH di Cauci Maurizio
Via Ronchetti, 2 - 21013 Gallarate (VA)
www.dreamtech-it.com

[ Attachment, skipping... ]

[ Attachment, skipping... ]

[ Attachment, skipping... ]
--
Bruce Momjian                        |  http://candle.pha.pa.us
pgman@candle.pha.pa.us               |  (610) 853-3000
+  If your life is a hard drive,     |  830 Blythe Avenue
+  Christ can be your backup.        |  Drexel Hill, Pennsylvania 19026

----------------------------------------------------------------------------
----

? config.log
? config.cache
? config.status
? GNUmakefile
? src/Makefile.custom
? src/GNUmakefile
? src/Makefile.global
? src/log
? src/crtags
? src/backend/port/Makefile
? src/bin/pg_dump/pg_dump
? src/bin/pg_dump/pg_restore
? src/bin/pg_dump/pg_dumpall
? src/include/config.h
? src/include/stamp-h
? src/interfaces/libpq/libpq.so.2.1
Index: src/backend/port/qnx4/sem.c
===================================================================
RCS file: /home/projects/pgsql/cvsroot/pgsql/src/backend/port/qnx4/sem.c,v
retrieving revision 1.3
diff -c -r1.3 sem.c
*** src/backend/port/qnx4/sem.c 2000/04/12 17:15:30 1.3
--- src/backend/port/qnx4/sem.c 2001/02/02 18:13:12
***************
*** 26,32 ****

#define SETMAX ((MAXBACKENDS + PROC_NSEMS_PER_SET - 1) /

PROC_NSEMS_PER_SET)

! #define SEMMAX (PROC_NSEMS_PER_SET)
#define OPMAX 8
#define MODE 0700
--- 26,32 ----
#define SETMAX ((MAXBACKENDS + PROC_NSEMS_PER_SET - 1) /

PROC_NSEMS_PER_SET)

! #define SEMMAX (PROC_NSEMS_PER_SET+1)
#define OPMAX 8

#define MODE 0700
Index: src/include/storage/s_lock.h
===================================================================
RCS file:

/home/projects/pgsql/cvsroot/pgsql/src/include/storage/s_lock.h,v

retrieving revision 1.86
diff -c -r1.86 s_lock.h
*** src/include/storage/s_lock.h 2001/01/24 19:43:28 1.86
--- src/include/storage/s_lock.h 2001/02/02 18:13:13
***************
*** 112,118 ****
*/

! #if defined(__i386__)
#define TAS(lock) tas(lock)

static __inline__ int
--- 112,118 ----
*/

! #if defined(__i386__) && !defined(__QNX__)
#define TAS(lock) tas(lock)

static __inline__ int
Index: src/interfaces/ecpg/preproc/pgc.l
===================================================================
RCS file:

/home/projects/pgsql/cvsroot/pgsql/src/interfaces/ecpg/preproc/pgc.l,v

Show quoted text

retrieving revision 1.73
diff -c -r1.73 pgc.l
*** src/interfaces/ecpg/preproc/pgc.l 2001/01/24 19:43:29 1.73
--- src/interfaces/ecpg/preproc/pgc.l 2001/02/02 18:13:17
***************
*** 35,40 ****
--- 35,45 ----
#undef yywrap
#endif /* yywrap */

+ #ifdef __QNX__
+ /* For some reason, QNX needs this, 2001-02-02 */
+ #define ECHO (void) fwrite( yytext, yyleng, 1, yyout )
+ #endif
+
#define YY_NO_UNPUT

extern YYSTYPE yylval;

Tom Lane

tgl@sss.pgh.pa.us

almost 25 years ago

In reply to: Bruce Momjian (#2)

Re: 7.1 beta 3 CHANGES FOR QNX

Bruce Momjian <pgman@candle.pha.pa.us> writes:

*** src/interfaces/ecpg/preproc/pgc.l	2001/01/24 19:43:29	1.73
--- src/interfaces/ecpg/preproc/pgc.l	2001/02/02 18:13:17
***************
*** 35,40 ****
--- 35,45 ----
#undef yywrap
#endif /* yywrap */

+ #ifdef __QNX__
+ /* For some reason, QNX needs this, 2001-02-02 */
+ #define ECHO (void) fwrite( yytext, yyleng, 1, yyout )
+ #endif
+ 
#define YY_NO_UNPUT

extern YYSTYPE yylval;

I do not believe this patch is correct or necessary. What is it trying
to fix, and if there's a problem here then why is there not a problem in
every other one of our flex outputs?

If there is a real problem --- presumably of the form "some QNX header
file pollutes the namespace with a definition of ECHO" --- then a more
appropriate fix would be

+ #ifdef __QNX__
+ /* Get rid of conflicting definition of ECHO from QNX's <something.h> */
+ #undef ECHO
+ #endif

so as not to tie ourselves to the exact definition of ECHO that flex is
currently using. However, I would first like to know where the conflict
is and why it doesn't break every other flex output in existence.

regards, tom lane

Maurizio

maurizio.c@libero.it

almost 25 years ago

In reply to: Bruce Momjian (#2)

Re: 7.1 beta 3 CHANGES FOR QNX

ECHO is defined in the following QNX gcc include files :
termio.h
termios.h

If ECHO was not redefined in pgc.l you can't compile in embedded SQL C.

I am also checking for another problem.
I have some errors if I compile pgsql without change the typedef Size in
c.h.
To succesfully compile pgsql I have changed typedef Size in int insteed
size_t.

regards
Maurizio Cauci

.
----- Original Message -----
From: "Tom Lane" <tgl@sss.pgh.pa.us>
To: "Bruce Momjian" <pgman@candle.pha.pa.us>
Cc: "Maurizio" <maurizio.c@libero.it>; <pgsql-hackers@postgresql.org>
Sent: Saturday, February 03, 2001 9:49 PM
Subject: Re: [HACKERS] 7.1 beta 3 CHANGES FOR QNX

Show quoted text

Bruce Momjian <pgman@candle.pha.pa.us> writes:
*** src/interfaces/ecpg/preproc/pgc.l 2001/01/24 19:43:29 1.73
--- src/interfaces/ecpg/preproc/pgc.l 2001/02/02 18:13:17
***************
*** 35,40 ****
--- 35,45 ----
#undef yywrap
#endif /* yywrap */
+ #ifdef __QNX__
+ /* For some reason, QNX needs this, 2001-02-02 */
+ #define ECHO (void) fwrite( yytext, yyleng, 1, yyout )
+ #endif
+
#define YY_NO_UNPUT
extern YYSTYPE yylval;

I do not believe this patch is correct or necessary. What is it trying
to fix, and if there's a problem here then why is there not a problem in
every other one of our flex outputs?

If there is a real problem --- presumably of the form "some QNX header
file pollutes the namespace with a definition of ECHO" --- then a more
appropriate fix would be
+ #ifdef __QNX__
+ /* Get rid of conflicting definition of ECHO from QNX's <something.h> */
+ #undef ECHO
+ #endif
so as not to tie ourselves to the exact definition of ECHO that flex is
currently using. However, I would first like to know where the conflict
is and why it doesn't break every other flex output in existence.

regards, tom lane

Tom Lane

tgl@sss.pgh.pa.us

almost 25 years ago

In reply to: Maurizio (#5)

Re: 7.1 beta 3 CHANGES FOR QNX

"Maurizio" <maurizio.c@libero.it> writes:

ECHO is defined in the following QNX gcc include files :
termio.h
termios.h
If ECHO was not redefined in pgc.l you can't compile in embedded SQL C.

Well, my question still stands: why aren't the other four flex outputs
also broken? They all use ECHO.

In any case, I'd prefer to see this fixed by not including <termios.h>
rather than hacking up the .l files. Surely it doesn't need to be
included everywhere, as src/include/port/qnx4.h is now causing to
happen. In fact, it looks to me like qnx4.h probably includes and
defines a lot more than it needs to; would you experiment with stripping
it down?

I am also checking for another problem.
I have some errors if I compile pgsql without change the typedef Size in
c.h.
To succesfully compile pgsql I have changed typedef Size in int insteed
size_t.

That strikes me as *horribly* dangerous. There is too much code whose
behavior might change in unpleasant ways if Size becomes a signed type.
Please explain what problems you are seeing that make you think this is
a good idea.

Andreas, the QNX port is largely your work IIRC. What do you think of
these issues? Have you tried 7.1beta on QNX?

regards, tom lane

Maurizio

maurizio@dreamtech-it.com

almost 25 years ago

In reply to: Bruce Momjian (#2)

Re: 7.1 beta 3 CHANGES FOR QNX

"Tom Lane" <tgl@sss.pgh.pa.us> writes:

Well, my question still stands: why aren't the other four flex outputs
also broken? They all use ECHO.

I don't know why, but probably you are right. I only know that if ECHO was
not redefined, when I compile with ecpg the output c file has all the
original lines on the same row (without an LF).

In any case, I'd prefer to see this fixed by not including <termios.h>
rather than hacking up the .l files. Surely it doesn't need to be
included everywhere, as src/include/port/qnx4.h is now causing to
happen. In fact, it looks to me like qnx4.h probably includes and
defines a lot more than it needs to; would you experiment with stripping
it down?

If You want I can experiment on qnx4.h. On Saturday I will post the risults.

That strikes me as *horribly* dangerous. There is too much code whose
behavior might change in unpleasant ways if Size becomes a signed type.
Please explain what problems you are seeing that make you think this is
a good idea.

In 7.0.2 and 7.0.3 release I have errors about some parameters in TCP/IP
functions.
The compiler tells me that I have a long int where an int was expected. When
I changed Size in int I compiled successfuly PGSQL. In 7.1 I changed
immediatly the size type and all seems works.
After your message I modified again Size type in size_t and recompiled 7.1
release. I compiled successfuly this version the only warnings are about
elog lines in wich there are Size variables. The compiler tells me I have a
long int where an unsigned was expected (only the format, there is a %u).
e.g. readfuncs.c at 2089 and 2113.
However PGSQL works right.

Andreas, the QNX port is largely your work IIRC. What do you think of
these issues? Have you tried 7.1beta on QNX?

I also would like to know Andreas Kardos is still out there and what he
think. When, some month ago,
I starded looking for POSTGRESQL and I had a lot of problems compiling 7.0.1
version (the major problem was what I have to do for the Size type?) I send
some e-mails to Dr. Kardos and he tells me that his version working fine.
After the first replay he never didn't replay to me.
Also other QNX users has the same problems I had and send me e-mails to know
if I have compiled successfully PGSQL for QNX. Nobody has had a reply from
Dr. Kardos.

thanks

regards
Maurizio Cauci

----- Original Message -----
From: "Tom Lane" <tgl@sss.pgh.pa.us>
To: "Maurizio" <maurizio.c@libero.it>
Cc: <pgsql-hackers@postgresql.org>; "Bruce Momjian"
<pgman@candle.pha.pa.us>; "Kardos, Dr. Andreas" <
Sent: Tuesday, February 06, 2001 4:08 AM
Subject: Re: [HACKERS] 7.1 beta 3 CHANGES FOR QNX

Show quoted text

"Maurizio" <maurizio.c@libero.it> writes:

ECHO is defined in the following QNX gcc include files :
termio.h
termios.h
If ECHO was not redefined in pgc.l you can't compile in embedded SQL C.

Well, my question still stands: why aren't the other four flex outputs
also broken? They all use ECHO.

In any case, I'd prefer to see this fixed by not including <termios.h>
rather than hacking up the .l files. Surely it doesn't need to be
included everywhere, as src/include/port/qnx4.h is now causing to
happen. In fact, it looks to me like qnx4.h probably includes and
defines a lot more than it needs to; would you experiment with stripping
it down?

I am also checking for another problem.
I have some errors if I compile pgsql without change the typedef Size in
c.h.
To succesfully compile pgsql I have changed typedef Size in int insteed
size_t.

That strikes me as *horribly* dangerous. There is too much code whose
behavior might change in unpleasant ways if Size becomes a signed type.
Please explain what problems you are seeing that make you think this is
a good idea.

Andreas, the QNX port is largely your work IIRC. What do you think of
these issues? Have you tried 7.1beta on QNX?

regards, tom lane

Maurizio

maurizio.c@libero.it

almost 25 years ago

In reply to: Bruce Momjian (#2)

Re: 7.1 beta 3 CHANGES FOR QNX

"Tom Lane" <tgl@sss.pgh.pa.us> writes:

Well, my question still stands: why aren't the other four flex outputs
also broken? They all use ECHO.

I don't know why, but probably you are right. I only know that if ECHO was
not redefined, when I compile with ecpg the output c file has all the
original lines on the same row (without an LF).

In any case, I'd prefer to see this fixed by not including <termios.h>
rather than hacking up the .l files. Surely it doesn't need to be
included everywhere, as src/include/port/qnx4.h is now causing to
happen. In fact, it looks to me like qnx4.h probably includes and
defines a lot more than it needs to; would you experiment with stripping
it down?

If You want I can experiment on qnx4.h. On Saturday I will post the risults.

That strikes me as *horribly* dangerous. There is too much code whose
behavior might change in unpleasant ways if Size becomes a signed type.
Please explain what problems you are seeing that make you think this is
a good idea.

Andreas, the QNX port is largely your work IIRC. What do you think of
these issues? Have you tried 7.1beta on QNX?

I also would like to know what Andreas Kardos think. When, some month ago,
I starded looking for POSTGRESQL and I had a lot of problems compiling 7.0.1
version (the major problem was what I have to do for the Size type?) I send
some e-mails to Dr. Kardos and he tells me that his version working fine.
After the first replay he didn't replay to me.
Also other QNX users has the same problems I had and send me e-mails to know
if I have compiled successfully PGSQl for QNX. Nobody has had a reply from
Dr. Kardos.

regards
Maurizio Cauci

Show quoted text

"Maurizio" <maurizio.c@libero.it> writes:

ECHO is defined in the following QNX gcc include files :
termio.h
termios.h
If ECHO was not redefined in pgc.l you can't compile in embedded SQL C.

Well, my question still stands: why aren't the other four flex outputs
also broken? They all use ECHO.

In any case, I'd prefer to see this fixed by not including <termios.h>
rather than hacking up the .l files. Surely it doesn't need to be
included everywhere, as src/include/port/qnx4.h is now causing to
happen. In fact, it looks to me like qnx4.h probably includes and
defines a lot more than it needs to; would you experiment with stripping
it down?

I am also checking for another problem.
I have some errors if I compile pgsql without change the typedef Size in
c.h.
To succesfully compile pgsql I have changed typedef Size in int insteed
size_t.

That strikes me as *horribly* dangerous. There is too much code whose
behavior might change in unpleasant ways if Size becomes a signed type.
Please explain what problems you are seeing that make you think this is
a good idea.

Andreas, the QNX port is largely your work IIRC. What do you think of
these issues? Have you tried 7.1beta on QNX?

regards, tom lane

Maurizio

maurizio.c@libero.it

almost 25 years ago

In reply to: Bruce Momjian (#2)

Re: 7.1 beta 3 CHANGES FOR QNX

Hi,
I have deleted the include of termios.h in include/port/qnx4.h.
Then I recompiled pgsql and I have compiled a program with ecpg.

All seem to work right.

Thanks
Maurizio

.
----- Original Message -----
From: "Tom Lane" <tgl@sss.pgh.pa.us>
To: "Maurizio" <maurizio.c@libero.it>
Cc: <pgsql-hackers@postgresql.org>; "Bruce Momjian"
<pgman@candle.pha.pa.us>; "Kardos, Dr. Andreas" <kardos@repas-aeg.de>
Sent: Tuesday, February 06, 2001 4:08 AM
Subject: Re: [HACKERS] 7.1 beta 3 CHANGES FOR QNX

Show quoted text

"Maurizio" <maurizio.c@libero.it> writes:

ECHO is defined in the following QNX gcc include files :
termio.h
termios.h
If ECHO was not redefined in pgc.l you can't compile in embedded SQL C.

Well, my question still stands: why aren't the other four flex outputs
also broken? They all use ECHO.

In any case, I'd prefer to see this fixed by not including <termios.h>
rather than hacking up the .l files. Surely it doesn't need to be
included everywhere, as src/include/port/qnx4.h is now causing to
happen. In fact, it looks to me like qnx4.h probably includes and
defines a lot more than it needs to; would you experiment with stripping
it down?

I am also checking for another problem.
I have some errors if I compile pgsql without change the typedef Size in
c.h.
To succesfully compile pgsql I have changed typedef Size in int insteed
size_t.

That strikes me as *horribly* dangerous. There is too much code whose
behavior might change in unpleasant ways if Size becomes a signed type.
Please explain what problems you are seeing that make you think this is
a good idea.

Andreas, the QNX port is largely your work IIRC. What do you think of
these issues? Have you tried 7.1beta on QNX?

regards, tom lane

#10

Bruce Momjian

pgman@candle.pha.pa.us

almost 25 years ago

In reply to: Maurizio (#9)

Re: 7.1 beta 3 CHANGES FOR QNX

OK, I have removed the termios.h, and thc ECHO hack from pgc.l.

[ Charset ISO-8859-1 unsupported, converting... ]

Hi,
I have deleted the include of termios.h in include/port/qnx4.h.
Then I recompiled pgsql and I have compiled a program with ecpg.

All seem to work right.

Thanks
Maurizio

.
----- Original Message -----
From: "Tom Lane" <tgl@sss.pgh.pa.us>
To: "Maurizio" <maurizio.c@libero.it>
Cc: <pgsql-hackers@postgresql.org>; "Bruce Momjian"
<pgman@candle.pha.pa.us>; "Kardos, Dr. Andreas" <kardos@repas-aeg.de>
Sent: Tuesday, February 06, 2001 4:08 AM
Subject: Re: [HACKERS] 7.1 beta 3 CHANGES FOR QNX

"Maurizio" <maurizio.c@libero.it> writes:

ECHO is defined in the following QNX gcc include files :
termio.h
termios.h
If ECHO was not redefined in pgc.l you can't compile in embedded SQL C.

Well, my question still stands: why aren't the other four flex outputs
also broken? They all use ECHO.

In any case, I'd prefer to see this fixed by not including <termios.h>
rather than hacking up the .l files. Surely it doesn't need to be
included everywhere, as src/include/port/qnx4.h is now causing to
happen. In fact, it looks to me like qnx4.h probably includes and
defines a lot more than it needs to; would you experiment with stripping
it down?

I am also checking for another problem.
I have some errors if I compile pgsql without change the typedef Size in
c.h.
To succesfully compile pgsql I have changed typedef Size in int insteed
size_t.

That strikes me as *horribly* dangerous. There is too much code whose
behavior might change in unpleasant ways if Size becomes a signed type.
Please explain what problems you are seeing that make you think this is
a good idea.

Andreas, the QNX port is largely your work IIRC. What do you think of
these issues? Have you tried 7.1beta on QNX?

regards, tom lane

-- 
  Bruce Momjian                        |  http://candle.pha.pa.us
  pgman@candle.pha.pa.us               |  (610) 853-3000
  +  If your life is a hard drive,     |  830 Blythe Avenue
  +  Christ can be your backup.        |  Drexel Hill, Pennsylvania 19026