Обсуждение: SGI Port of Postgresql 6.4 snapshot of 09/28/98
Gentlemen:
I have tested today's snapshot (retrieved this morning)
on an SGI Irix 6.2 system with the following results:
1) The int8 test still fails because the SGI C compiler does
not handle a variable argument declaration of "long long" correctly in
the definition of the function dopr in snprintf.c. According to the
SGI documentation, the handling of "long long" is undefined. I think
the best solution is to use sprintf in int8.c, because there is no
additional value to using snprintf in this context. There is no risk
over buffer overflow in int8.c because the maximum length of a
possible result is known in advance (21 characters + one for the null
termination). I've modified int8.c on my machine accordingly.
2) The definition of the slock_t type is incorrect in irix5.h.
3) The new locking mechanism using test_and_set subroutine
calls was tested by running 5 parallel processes, two creating indices
and three doing sequential scans. All queries completed normally.
4) The comment in s_lock.h on possible future SGI locking
mechanisms should be deleted, since this mechanism is fast, works, and
is supported by SGI.
Patches to make these corrections (except 4) are included below.
Also, please include the following Makefile.custom into
the SGI Irix FAQ:
CUSTOM_CC = cc -32 -g
LD += -32 -g
The "-32" switch is essential, but I haven't tested the need for "-g".
If I have time before the release, I will report the results. At least,
I know that this version passes the regression tests.
Would you please put my name on the list of people who
contributed to the 6.4 release? Thanks.
Here are the patches:
*** backend/utils/adt/int8.c.orig Tue Sep 1 03:01:35 1998
--- backend/utils/adt/int8.c Mon Sep 28 21:32:38 1998
***************
*** 66,74 ****
if (!PointerIsValid(val))
return NULL;
! if ((len = snprintf(buf, MAXINT8LEN, INT64_FORMAT, *val)) < 0)
elog(ERROR, "Unable to format int8", NULL);
-
result = palloc(len + 1);
strcpy(result, buf);
--- 66,73 ----
if (!PointerIsValid(val))
return NULL;
! if ((len = sprintf(buf, INT64_FORMAT, *val)) < 0)
elog(ERROR, "Unable to format int8", NULL);
result = palloc(len + 1);
strcpy(result, buf);
*** include/port/irix5.h.orig Sun Sep 7 00:59:54 1997
--- include/port/irix5.h Mon Sep 28 19:37:12 1998
***************
*** 3,7 ****
#define NO_EMPTY_STMTS
#define SYSV_DIRENT
#define HAS_TEST_AND_SET
! #include <abi_mutex.h>
! typedef abilock_t slock_t;
--- 3,6 ----
#define NO_EMPTY_STMTS
#define SYSV_DIRENT
#define HAS_TEST_AND_SET
! typedef unsigned long slock_t;
*** include/storage/s_lock.h.orig Mon Sep 21 03:00:17 1998
--- include/storage/s_lock.h Mon Sep 28 19:37:12 1998
***************
*** 259,266 ****
#if defined(__sgi)
/*
* SGI IRIX 5
! * slock_t is defined as a struct abilock_t, which has a single unsigned long
! * member.
*
* 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
--- 259,269 ----
#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
+------------------------------------------+------------------------------+
| Robert E. Bruccoleri, Ph.D. | Associate Research Professor |
| phone: 732 235 5796 | Center for Advanced |
| Fax: 732 235 4850 | Biotechnology and Medicine |
| email: bruc@acm.org | Rutgers University |
| URL: http://www.cabm.rutgers.edu/~bruc | 679 Hoes Lane |
| | Piscataway, NJ 08854-5638 |
+------------------------------------------+------------------------------+
> Gentlemen:
The test of the Irix port of PostgreSQL running with default
optimization passes all the regression tests the same as before
except that the random number test is now different. The new output is
included below. It appears to be OK, but I would like confirmation.
Thanks.
=============================random.out=================================
QUERY: SELECT count(*) FROM onek;
count
-----
1000
(1 row)
QUERY: SELECT count(*) AS random INTO RANDOM_TBL
FROM onek WHERE oidrand(onek.oid, 10);
QUERY: INSERT INTO RANDOM_TBL (random)
SELECT count(*)
FROM onek WHERE oidrand(onek.oid, 10);
QUERY: SELECT random, count(random) FROM RANDOM_TBL
GROUP BY random HAVING count(random) > 1;
random|count
------+-----
(0 rows)
QUERY: SELECT random FROM RANDOM_TBL
WHERE random NOT BETWEEN 80 AND 120;
random
------
74
(1 row)
+------------------------------------------+------------------------------+
| Robert E. Bruccoleri, Ph.D. | Associate Research Professor |
| phone: 732 235 5796 | Center for Advanced |
| Fax: 732 235 4850 | Biotechnology and Medicine |
| email: bruc@acm.org | Rutgers University |
| URL: http://www.cabm.rutgers.edu/~bruc | 679 Hoes Lane |
| | Piscataway, NJ 08854-5638 |
+------------------------------------------+------------------------------+
I have applied patches 2, 3, and 4. For patch 1, I changed the 'long long' in va_arg() to a typedef for 'long long'. Let me know if that fixes it. > Gentlemen: > I have tested today's snapshot (retrieved this morning) > on an SGI Irix 6.2 system with the following results: > > 1) The int8 test still fails because the SGI C compiler does > not handle a variable argument declaration of "long long" correctly in > the definition of the function dopr in snprintf.c. According to the > SGI documentation, the handling of "long long" is undefined. I think > the best solution is to use sprintf in int8.c, because there is no > additional value to using snprintf in this context. There is no risk > over buffer overflow in int8.c because the maximum length of a > possible result is known in advance (21 characters + one for the null > termination). I've modified int8.c on my machine accordingly. > 2) The definition of the slock_t type is incorrect in irix5.h. > 3) The new locking mechanism using test_and_set subroutine > calls was tested by running 5 parallel processes, two creating indices > and three doing sequential scans. All queries completed normally. > 4) The comment in s_lock.h on possible future SGI locking > mechanisms should be deleted, since this mechanism is fast, works, and > is supported by SGI. > Patches to make these corrections (except 4) are included below. > > Also, please include the following Makefile.custom into > the SGI Irix FAQ: > > CUSTOM_CC = cc -32 -g > LD += -32 -g > > The "-32" switch is essential, but I haven't tested the need for "-g". > If I have time before the release, I will report the results. At least, > I know that this version passes the regression tests. Please contact the IRIX FAQ maintainer directly. > > Would you please put my name on the list of people who > contributed to the 6.4 release? Thanks. Done. > > Here are the patches: > > *** backend/utils/adt/int8.c.orig Tue Sep 1 03:01:35 1998 > --- backend/utils/adt/int8.c Mon Sep 28 21:32:38 1998 > *************** > *** 66,74 **** > if (!PointerIsValid(val)) > return NULL; > > ! if ((len = snprintf(buf, MAXINT8LEN, INT64_FORMAT, *val)) < 0) > elog(ERROR, "Unable to format int8", NULL); > - > result = palloc(len + 1); > > strcpy(result, buf); > --- 66,73 ---- > if (!PointerIsValid(val)) > return NULL; > > ! if ((len = sprintf(buf, INT64_FORMAT, *val)) < 0) > elog(ERROR, "Unable to format int8", NULL); > result = palloc(len + 1); > > strcpy(result, buf); > *** include/port/irix5.h.orig Sun Sep 7 00:59:54 1997 > --- include/port/irix5.h Mon Sep 28 19:37:12 1998 > *************** > *** 3,7 **** > #define NO_EMPTY_STMTS > #define SYSV_DIRENT > #define HAS_TEST_AND_SET > ! #include <abi_mutex.h> > ! typedef abilock_t slock_t; > --- 3,6 ---- > #define NO_EMPTY_STMTS > #define SYSV_DIRENT > #define HAS_TEST_AND_SET > ! typedef unsigned long slock_t; > *** include/storage/s_lock.h.orig Mon Sep 21 03:00:17 1998 > --- include/storage/s_lock.h Mon Sep 28 19:37:12 1998 > *************** > *** 259,266 **** > #if defined(__sgi) > /* > * SGI IRIX 5 > ! * slock_t is defined as a struct abilock_t, which has a single unsigned long > ! * member. > * > * 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 > --- 259,269 ---- > #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 > > +------------------------------------------+------------------------------+ > | Robert E. Bruccoleri, Ph.D. | Associate Research Professor | > | phone: 732 235 5796 | Center for Advanced | > | Fax: 732 235 4850 | Biotechnology and Medicine | > | email: bruc@acm.org | Rutgers University | > | URL: http://www.cabm.rutgers.edu/~bruc | 679 Hoes Lane | > | | Piscataway, NJ 08854-5638 | > +------------------------------------------+------------------------------+ > -- Bruce Momjian | http://www.op.net/~candle maillist@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