[BUGS] BUG #2506: hash index
The following bug has been logged online: Bug reference: 2506 Logged by: Wagner Cipriano Email address: [EMAIL PROTECTED] PostgreSQL version: 8.0 Operating system: Linus Suse 10 Description:hash index Details: Error: index "xxx" is not a hash index. Always this error happen with me. Can You help me ? Thanks! ---(end of broadcast)--- TIP 6: explain analyze is your friend
Re: [BUGS] BUG #2503: Query results differ based on record count
Alvaro Herrera wrote: > Russell Francis wrote: > >> If a SELECT statement contains a table in it's FROM clause which has 0 >> records in it, the query will not return any results. If a record is added >> to the table even though it shouldn't modify the results of query, the query >> will return the expected result. >> >> The SQL statements below are a trimmed down test case which duplicates this >> issue. The identical SELECT statements return different results depending >> on the record count of the unrelated table_b. >> >> SELECT table_a.* FROM table_a, table_b WHERE table_a.ta_col_a > 0; > > table_b is certainly not unrelated, as it appears in the FROM clause. > So the query is correct to not show any result, per definition of > cartesian product. > Alvaro, Yes, you are correct. Sorry. Thanks, Russ signature.asc Description: OpenPGP digital signature
[BUGS] use-after-free in psql
There's a minor bug in the ON_ERROR_ROLLBACK code in psql. In HEAD, at line 878 the storage pointed to by "results" is released by a PQclear(), but is referenced by the PQcmdStatus() calls on lines 898, 899, and 900. I'm busy at the moment -- if someone wants to fix this (backport to 8.1 please!), have at it. Otherwise I'll fix it this weekend. -Neil ---(end of broadcast)--- TIP 2: Don't 'kill -9' the postmaster
Re: [BUGS] BUG #2506: hash index
"Wagner Cipriano" <[EMAIL PROTECTED]> writes: > Error: index "xxx" is not a hash index. > Always this error happen with me. You haven't provided enough information to let anyone help you ... but if you mean that you've got a corrupt index, REINDEXing it should fix that. regards, tom lane ---(end of broadcast)--- TIP 5: don't forget to increase your free space map settings
Re: [BUGS] use-after-free in psql
> There's a minor bug in the ON_ERROR_ROLLBACK code in psql. In
> HEAD, at line 878 the storage pointed to by "results" is
> released by a PQclear(), but is referenced by the
> PQcmdStatus() calls on lines 898, 899, and 900.
>
> I'm busy at the moment -- if someone wants to fix this
> (backport to 8.1 please!), have at it.
Attached is a quick patch for HEAD and 8.1, which should do the job.
Thanks for finding this.
--
Greg Sabino Mullane [EMAIL PROTECTED]
PGP Key: 0x14964AC8 200606301039
http://biglumber.com/x/web?pk=2529DF6AB8F79407E94445B4BC9B906714964AC8
Index: common.c
===
RCS file: /projects/cvsroot/pgsql/src/bin/psql/common.c,v
retrieving revision 1.119
diff -u -u -r1.119 common.c
--- common.c 14 Jun 2006 16:49:02 - 1.119
+++ common.c 30 Jun 2006 13:33:31 -
@@ -875,8 +875,6 @@
if (OK)
OK = PrintQueryResults(results);
- PQclear(results);
-
/* If we made a temporary savepoint, possibly release/rollback */
if (on_error_rollback_savepoint)
{
@@ -884,23 +882,35 @@
/* We always rollback on an error */
if (transaction_status == PQTRANS_INERROR)
+ {
+ PQclear(results);
results = PQexec(pset.db, "ROLLBACK TO pg_psql_temporary_savepoint");
+ }
/* If they are no longer in a transaction, then do nothing */
else if (transaction_status != PQTRANS_INTRANS)
+ {
+ PQclear(results);
results = NULL;
+ }
else
{
/*
- * Do nothing if they are messing with savepoints themselves: If
+ * Do nothing if they are messing with savepoints themselves: if
* the user did RELEASE or ROLLBACK, our savepoint is gone. If
* they issued a SAVEPOINT, releasing ours would remove theirs.
*/
if (strcmp(PQcmdStatus(results), "SAVEPOINT") == 0 ||
strcmp(PQcmdStatus(results), "RELEASE") == 0 ||
strcmp(PQcmdStatus(results), "ROLLBACK") == 0)
+ {
+PQclear(results);
results = NULL;
+ }
else
+ {
+PQclear(results);
results = PQexec(pset.db, "RELEASE pg_psql_temporary_savepoint");
+ }
}
if (PQresultStatus(results) != PGRES_COMMAND_OK)
{
@@ -909,8 +919,8 @@
ResetCancelConn();
return false;
}
- PQclear(results);
}
+ PQclear(results);
/* Possible microtiming output */
if (OK && pset.timing)
Index: common.c
===
RCS file: /projects/cvsroot/pgsql/src/bin/psql/common.c,v
retrieving revision 1.110.2.1
diff -u -u -r1.110.2.1 common.c
--- common.c 22 Nov 2005 18:23:27 - 1.110.2.1
+++ common.c 30 Jun 2006 13:42:29 -
@@ -1067,8 +1067,6 @@
if (OK)
OK = PrintQueryResults(results);
- PQclear(results);
-
/* If we made a temporary savepoint, possibly release/rollback */
if (on_error_rollback_savepoint)
{
@@ -1076,23 +1074,35 @@
/* We always rollback on an error */
if (transaction_status == PQTRANS_INERROR)
+ {
+ PQclear(results);
results = PQexec(pset.db, "ROLLBACK TO pg_psql_temporary_savepoint");
+ }
/* If they are no longer in a transaction, then do nothing */
else if (transaction_status != PQTRANS_INTRANS)
+ {
+ PQclear(results);
results = NULL;
+ }
else
{
/*
- * Do nothing if they are messing with savepoints themselves: If
+ * Do nothing if they are messing with savepoints themselves: if
* the user did RELEASE or ROLLBACK, our savepoint is gone. If
* they issued a SAVEPOINT, releasing ours would remove theirs.
*/
if (strcmp(PQcmdStatus(results), "SAVEPOINT") == 0 ||
strcmp(PQcmdStatus(results), "RELEASE") == 0 ||
strcmp(PQcmdStatus(results), "ROLLBACK") == 0)
+ {
+PQclear(results);
results = NULL;
+ }
else
+ {
+PQclear(results);
results = PQexec(pset.db, "RELEASE pg_psql_temporary_savepoint");
+ }
}
if (PQresultStatus(results) != PGRES_COMMAND_OK)
{
@@ -1101,8 +,8 @@
ResetCancelConn();
return false;
}
- PQclear(results);
}
+ PQclear(results);
/* Possible microtiming output */
if (OK && pset.timing)
signature.asc
Description: This is a digitally signed message part
Re: [BUGS] use-after-free in psql
Neil Conway wrote:
> There's a minor bug in the ON_ERROR_ROLLBACK code in psql. In HEAD, at
> line 878 the storage pointed to by "results" is released by a PQclear(),
> but is referenced by the PQcmdStatus() calls on lines 898, 899, and 900.
I think this is the fix for HEAD. I'll apply after testing it a bit
more.
--
Alvaro Herrerahttp://www.CommandPrompt.com/
The PostgreSQL Company - Command Prompt, Inc.
? common.c.fix
Index: common.c
===
RCS file: /cvsroot/pgsql/src/bin/psql/common.c,v
retrieving revision 1.119
diff -c -p -r1.119 common.c
*** common.c14 Jun 2006 16:49:02 - 1.119
--- common.c30 Jun 2006 14:40:18 -
*** SendQuery(const char *query)
*** 875,893
if (OK)
OK = PrintQueryResults(results);
- PQclear(results);
-
/* If we made a temporary savepoint, possibly release/rollback */
if (on_error_rollback_savepoint)
{
transaction_status = PQtransactionStatus(pset.db);
/* We always rollback on an error */
if (transaction_status == PQTRANS_INERROR)
! results = PQexec(pset.db, "ROLLBACK TO
pg_psql_temporary_savepoint");
/* If they are no longer in a transaction, then do nothing */
else if (transaction_status != PQTRANS_INTRANS)
! results = NULL;
else
{
/*
--- 875,893
if (OK)
OK = PrintQueryResults(results);
/* If we made a temporary savepoint, possibly release/rollback */
if (on_error_rollback_savepoint)
{
+ PGresult *svptres;
+
transaction_status = PQtransactionStatus(pset.db);
/* We always rollback on an error */
if (transaction_status == PQTRANS_INERROR)
! svptres = PQexec(pset.db, "ROLLBACK TO
pg_psql_temporary_savepoint");
/* If they are no longer in a transaction, then do nothing */
else if (transaction_status != PQTRANS_INTRANS)
! svptres = NULL;
else
{
/*
*** SendQuery(const char *query)
*** 898,917
if (strcmp(PQcmdStatus(results), "SAVEPOINT") == 0 ||
strcmp(PQcmdStatus(results), "RELEASE") == 0 ||
strcmp(PQcmdStatus(results), "ROLLBACK") == 0)
! results = NULL;
else
! results = PQexec(pset.db, "RELEASE
pg_psql_temporary_savepoint");
}
! if (PQresultStatus(results) != PGRES_COMMAND_OK)
{
psql_error("%s", PQerrorMessage(pset.db));
PQclear(results);
ResetCancelConn();
return false;
}
- PQclear(results);
}
/* Possible microtiming output */
if (OK && pset.timing)
printf(_("Time: %.3f ms\n"), DIFF_MSEC(&after, &before));
--- 898,919
if (strcmp(PQcmdStatus(results), "SAVEPOINT") == 0 ||
strcmp(PQcmdStatus(results), "RELEASE") == 0 ||
strcmp(PQcmdStatus(results), "ROLLBACK") == 0)
! svptres = NULL;
else
! svptres = PQexec(pset.db, "RELEASE
pg_psql_temporary_savepoint");
}
! if (svptres && PQresultStatus(svptres) != PGRES_COMMAND_OK)
{
psql_error("%s", PQerrorMessage(pset.db));
PQclear(results);
+ PQclear(svptres);
ResetCancelConn();
return false;
}
}
+ PQclear(results);
+
/* Possible microtiming output */
if (OK && pset.timing)
printf(_("Time: %.3f ms\n"), DIFF_MSEC(&after, &before));
---(end of broadcast)---
TIP 3: Have you checked our extensive FAQ?
http://www.postgresql.org/docs/faq
