1
0
mirror of https://github.com/postgres/postgres.git synced 2025-08-21 10:42:50 +03:00
Commit Graph

8555 Commits

Author SHA1 Message Date
Tom Lane
36e012e727 Remove temporary Windows-specific debugging code; it seems the problem
with fopen() not using FILE_SHARE_DELETE was indeed the bug we were after,
given lack of recent reports.
2006-11-06 17:10:22 +00:00
Tom Lane
74686b6de7 Get rid of some unnecessary dependencies on DataDir: wherever possible,
the backend should rely on its working-directory setting instead.
Also do some message-style police work in contrib/adminpack.
2006-11-06 03:06:41 +00:00
Neil Conway
62fe410ec6 Minor fix for LDAP authentication: if an error occurs, we need to
manually release the LDAP handle via ldap_unbind(). This isn't a
significant problem in practice because an error eventually results
in exiting the process, but we can cleanup correctly without too
much pain.

In passing, fix an error in snprintf() usage: the "size" parameter
to snprintf() is the size of the destination buffer, including space
for the NUL terminator. Also, depending on the value of NAMEDATALEN,
the old coding could have allowed for a buffer overflow.
2006-11-06 01:27:52 +00:00
Tom Lane
76d5667ba8 Fix recently-identified PITR recovery hazard: the base backup could contain
stale relcache init files (pg_internal.init), and there is no mechanism for
updating them during WAL replay.  Easiest solution is just to delete the init
files at conclusion of startup, and let the first backend started in each
database take care of rebuilding the init file.  Simon Riggs and Tom Lane.

Back-patched to 8.1.  Arguably this should be fixed in 8.0 too, but it would
require significantly more code since 8.0 has no handy startup-time scan of
pg_database to piggyback on.  Manual solution of the problem is possible
in 8.0 (just delete the pg_internal.init files before starting WAL replay),
so that may be a sufficient answer.
2006-11-05 23:40:31 +00:00
Tom Lane
48188e1621 Fix recently-understood problems with handling of XID freezing, particularly
in PITR scenarios.  We now WAL-log the replacement of old XIDs with
FrozenTransactionId, so that such replacement is guaranteed to propagate to
PITR slave databases.  Also, rather than relying on hint-bit updates to be
preserved, pg_clog is not truncated until all instances of an XID are known to
have been replaced by FrozenTransactionId.  Add new GUC variables and
pg_autovacuum columns to allow management of the freezing policy, so that
users can trade off the size of pg_clog against the amount of freezing work
done.  Revise the already-existing code that forces autovacuum of tables
approaching the wraparound point to make it more bulletproof; also, revise the
autovacuum logic so that anti-wraparound vacuuming is done per-table rather
than per-database.  initdb forced because of changes in pg_class, pg_database,
and pg_autovacuum catalogs.  Heikki Linnakangas, Simon Riggs, and Tom Lane.
2006-11-05 22:42:10 +00:00
Tom Lane
70ce5c9082 Fix "failed to re-find parent key" btree VACUUM failure by revising page
deletion code to avoid the case where an upper-level btree page remains "half
dead" for a significant period of time, and to block insertions into a key
range that is in process of being re-assigned to the right sibling of the
deleted page's parent.  This prevents the scenario reported by Ed L. wherein
index keys could become out-of-order in the grandparent index level.

Since this is a moderately invasive fix, I'm applying it only to HEAD.
The bug exists back to 7.4, but the back branches will get a different patch.
2006-11-01 19:43:17 +00:00
Tom Lane
76d5f6f035 expression_tree_walker failed to let walker function see the immediate child
node of a SubLink or SubPlan testexpr field.  Bug resulted from replacing
the old lefthand/exprs list fields with a simple expression field, and not
remembering that expression_tree_walker is coded to save a few cycles by
recursing directly to self on list fields (on the assumption the walker
isn't interested in List nodes per se).  On non-list fields it must of
course call the walker.  Possibly that hack isn't worth the risk of more
such bugs, but I'll leave it be for now.  Per bug report from James Robinson.
2006-10-25 22:11:32 +00:00
Tom Lane
4df8de7a68 Fix check for whether a clauseless join has to be forced in the presence of
outer joins.  Originally it was only looking for overlap of the righthand
side of a left join, but we have to do it on the lefthand side too.
Per example from Jean-Pierre Pelletier.
2006-10-24 17:50:22 +00:00
Tom Lane
954c1813ac Remove an unnecessary HOLD_INTERRUPTS/RESUME_INTERRUPTS pair.
This was required back when RESUME_INTERRUPTS could actually
execute ProcessInterrupts, but that hasn't been true since 2001...
2006-10-22 20:34:54 +00:00
Peter Eisentraut
9c2c41646a Translations update 2006-10-21 21:03:04 +00:00
Tom Lane
33af087695 Try to fix the AIX getaddrinfo mess in a way that works on all versions.
Going to wait for buildfarm results before backpatching, this time.
2006-10-19 23:17:39 +00:00
Tom Lane
1b84441a49 Marginal improvement in logging: include the function name when logging
a fastpath function call.
2006-10-19 22:44:11 +00:00
Tom Lane
2315df21f8 Fix a couple of places that were assuming debug_query_string couldn't
be NULL ... seems an unsafe assumption.
2006-10-19 19:53:03 +00:00
Tom Lane
681892208f Make sure that debug_query_string contains the original query text,
if available (which it usually should be), during processing of Bind
and Execute protocol messages.  This improves usefulness of
log_min_error_statement logging for extended query protocol.
2006-10-19 19:52:22 +00:00
Tom Lane
def651f48f Clean up local redeclarations of variables with DLLIMPORT, per report
from Magnus that MSVC complains about this.
2006-10-19 18:32:48 +00:00
Tom Lane
dbb397f30f Work around reported problem that AIX's getaddrinfo() doesn't seem to zero
sin_port in the returned IP address struct when servname is NULL.  This has
been observed to cause failure to bind the stats collection socket, and
could perhaps cause other issues too.  Per reports from Brad Nicholson
and Chris Browne.
2006-10-19 17:26:32 +00:00
Tom Lane
1e758d5263 Add some code to CREATE DATABASE to check for pre-existing subdirectories
that conflict with the OID that we want to use for the new database.
This avoids the risk of trying to remove files that maybe we shouldn't
remove.  Per gripe from Jon Lapham and subsequent discussion of 27-Sep.
2006-10-18 22:44:12 +00:00
Tom Lane
877f08da14 Fix up timetz input so that a date is required only when the specified
timezone actually has a daylight-savings rule.  This avoids breaking
cases that used to work because they went through the DecodePosixTimezone
code path.  Per contrib regression failures (mea culpa for not running
those yesterday...).  Also document the already-applied change to allow
GMT offsets up to 14 hours.
2006-10-18 16:43:14 +00:00
Tom Lane
022fd99668 Fix up some problems in handling of zic-style time zone names in datetime
input routines.  Remove the former "DecodePosixTimezone" function in favor of
letting the zic code handle POSIX-style zone specs (see tzparse()).  In
particular this means that "PST+3" now means the same as "-03", whereas it
used to mean "-11" --- the zone abbreviation is effectively just a noise word
in this syntax.  Make sure that all named and POSIX-style zone names will be
parsed as a single token.  Fix long-standing bogosities in printing and input
of fractional-hour timezone offsets (since the tzparse() code will accept
these, we'd better make 'em work).  Also correct an error in the original
coding of the zic-zone-name patch: in "timestamp without time zone" input,
zone names are supposed to be allowed but ignored, but the coding was such
that the zone changed the interpretation anyway.
2006-10-17 21:03:21 +00:00
Tom Lane
e0dece127d Redesign the patch for allocation of shmem space and LWLocks for add-on
modules; the first try was not usable in EXEC_BACKEND builds (e.g.,
Windows).  Instead, just provide some entry points to increase the
allocation requests during postmaster start, and provide a dedicated
LWLock that can be used to synchronize allocation operations performed
by backends.  Per discussion with Marc Munro.
2006-10-15 22:04:08 +00:00
Tom Lane
f58eac82ee Code and docs review for ALTER TABLE INHERIT/NO INHERIT patch. 2006-10-13 21:43:19 +00:00
Teodor Sigaev
47df6e7e2d Fix infinite sleep and failes of send in Win32.
1) pgwin32_waitforsinglesocket(): WaitForMultipleObjectsEx now called with
finite timeout (100ms) in case of FP_WRITE and UDP socket. If timeout occurs
then pgwin32_waitforsinglesocket() tries to write empty packet goes to
WaitForMultipleObjectsEx again.

2) pgwin32_send(): add loop around WSASend and pgwin32_waitforsinglesocket().
The reason is: for overlapped socket, 'ok' result from
pgwin32_waitforsinglesocket() isn't guarantee that socket is still free,
it can become busy again and following WSASend call will fail with
WSAEWOULDBLOCK error.

See http://archives.postgresql.org/pgsql-hackers/2006-10/msg00561.php
2006-10-13 13:59:47 +00:00
Tom Lane
d2e17e1ddc Fix mishandling of after-trigger state when a SQL function returns multiple
rows --- if the surrounding query queued any trigger events between the rows,
the events would be fired at the wrong time, leading to bizarre behavior.
Per report from Merlin Moncure.

This is a simple patch that should solve the problem fully in the back
branches, but in HEAD we also need to consider the possibility of queries
with RETURNING clauses.  Will look into a fix for that separately.
2006-10-12 17:02:24 +00:00
Bruce Momjian
638cc9abf8 Update pg_hba.conf comment about documentation section. 2006-10-11 23:01:46 +00:00
Tom Lane
772c5ba31f Repair incorrect check for coercion of unknown literal to ANYARRAY, a bug
I introduced in 7.4.1 :-(.  It's correct to allow unknown to be coerced to
ANY or ANYELEMENT, since it's a real-enough data type, but it most certainly
isn't an array datatype.  This can cause a backend crash but AFAICT is not
exploitable as a security hole.  Per report from Michael Fuhr.

Note: as fixed in HEAD, this changes a constant in the pg_stats view,
resulting in a change in the expected regression outputs.  The back-branch
patches have been hacked to avoid that, so that pre-existing installations
won't start failing their regression tests.
2006-10-11 20:21:04 +00:00
Tom Lane
8f2f180ff1 Code review for LIKE INCLUDING CONSTRAINTS patch --- improve comments,
don't cheat on the raw-vs-cooked status of a constraint.
2006-10-11 16:42:59 +00:00
Tom Lane
c50b36d215 Update Darwin dlopen() support to avoid deprecation warnings with latest
Apple developer tools.  We now use dlopen directly if available, and fall
back to the older code if not.  Chris Campbell
2006-10-08 19:31:03 +00:00
Bruce Momjian
7efbf6d7b6 Add include needed for new getrusage() call. 2006-10-08 17:45:50 +00:00
Tom Lane
c7611f99d6 On Windows, we know the backend stack size limit because we have to
specify it explicitly in backend/Makefile.  Arrange for this value to
be known by get_stack_depth_rlimit() too.  Per suggestion from Magnus.
2006-10-08 17:15:34 +00:00
Peter Eisentraut
5a05d18e59 Added missing entry (CASCADED) in keywords table. 2006-10-07 21:51:02 +00:00
Tom Lane
dc9142f406 When planning a query at Bind time, be careful to pass the correct
query_list into the Portal, ie, the one seen and possibly modified by
the planner.  My fault :-(  Per report from Sergey Koposov.
2006-10-07 20:16:57 +00:00
Tom Lane
71a6f8b85b On platforms that have getrlimit(RLIMIT_STACK), use it to ensure that
max_stack_depth is not set to an unsafe value.

This commit also provides configure-time checking for <sys/resource.h>,
and cleans up some perhaps-unportable code associated with use of that
include file and getrlimit().
2006-10-07 19:25:29 +00:00
Tom Lane
1c160291ef Adjust HINT for stack depth limit to mention checking the underlying
platform limit, rather than just blindly raising max_stack_depth.
Also, tweak the code to work properly if someone sets max_stack_depth
to more than 2Gb, which guc.c will allow on a 64-bit machine.
2006-10-07 16:43:28 +00:00
Tom Lane
452fa214e5 Fix string_to_array() to correctly handle the case where there are
overlapping possible matches for the separator string, such as
string_to_array('123xx456xxx789', 'xx').
Also, revise the logic of replace(), split_part(), and string_to_array()
to avoid O(N^2) work from redundant searches and conversions to pg_wchar
format when there are N matches to the separator string.
Backpatched the full patch as far as 8.0.  7.4 also has the bug, but the
code has diverged a lot, so I just went for a quick-and-dirty fix of the
bug itself in that branch.
2006-10-07 00:11:53 +00:00
Tom Lane
6ea8f49573 Fix SysCacheGetAttr() to handle the case where the specified syscache has not
been initialized yet.  This can happen because there are code paths that call
SysCacheGetAttr() on a tuple originally fetched from a different syscache
(hopefully on the same catalog) than the one specified in the call.  It
doesn't seem useful or robust to try to prevent that from happening, so just
improve the function to cope instead.  Per bug#2678 from Jeff Trout.  The
specific example shown by Jeff is new in 8.1, but to be on the safe side
I'm backpatching 8.0 as well.  We could patch 7.x similarly but I think
that's probably overkill, given the lack of evidence of old bugs of this ilk.
2006-10-06 18:23:35 +00:00
Peter Eisentraut
b9b4f10b5b Message style improvements 2006-10-06 17:14:01 +00:00
Tom Lane
378c79dc78 Cleanup for pglz_compress code: remove dead code, const-ify API of
remaining functions, simplify pglz_compress's API to not require a useless
data copy when compression fails.  Also add a check in pglz_decompress that
the expected amount of data was decompressed.
2006-10-05 23:33:33 +00:00
Tom Lane
e378f82e00 Make use of qsort_arg in several places that were formerly using klugy
static variables.  This avoids any risk of potential non-reentrancy,
and in particular offers a much cleaner workaround for the Intel compiler
bug that was affecting ginutil.c.
2006-10-05 17:57:40 +00:00
Tom Lane
b2c04d5e6b Work around buggy strtod on (some versions of?) IRIX. Combination of
proposed patches from John Jorgensen and Steve Singer.
2006-10-05 01:40:45 +00:00
Tom Lane
af0af42f85 Fix comment to agree with presentation. 2006-10-04 18:45:52 +00:00
Bruce Momjian
f99a569a2e pgindent run for 8.2. 2006-10-04 00:30:14 +00:00
Tom Lane
6edd2b4a91 Switch over to using our own qsort() all the time, as has been proposed
repeatedly.  Now that we don't have to worry about memory leaks from
glibc's qsort, we can safely put CHECK_FOR_INTERRUPTS into the tuplesort
comparators, as was requested a couple months ago.  Also, get rid of
non-reentrancy and an extra level of function call in tuplesort.c by
providing a variant qsort_arg() API that passes an extra void * argument
through to the comparison routine.  (We might want to use that in other
places too, I didn't look yet.)
2006-10-03 22:18:23 +00:00
Bruce Momjian
0a42adcc32 Improve numeric overflow error message.
David Fetter
2006-10-03 21:25:56 +00:00
Bruce Momjian
45c8ed96b9 Make some sentences consistent with similar ones.
Euler Taveira de Oliveira
2006-10-03 21:21:36 +00:00
Bruce Momjian
5441a64164 The attached patch changes units of the some default values in
postgresql.conf.

- shared_buffers = 32000kB => 32MB
- temp_buffers = 8000kB => 8MB
- wal_buffers = 8 => 64kB

The code of initdb was a bit modified to write MB-unit values.
Values greater than 8000kB are rounded out to MB.

GUC_UNIT_XBLOCKS is added for wal_buffers. It is like GUC_UNIT_BLOCKS,
but uses XLOG_BLCKSZ instead of BLCKSZ.

Also, I cleaned up the test of GUC_UNIT_* flags in preparation to
add more unit flags in less bits.

ITAGAKI Takahiro
2006-10-03 21:11:55 +00:00
Bruce Momjian
e3f1607cf8 Mention in postgresql.conf that autovacuum also needs
stats_start_collector and stats_row_level to also be on

David Wheeler
2006-10-03 00:39:26 +00:00
Tom Lane
8e4fe3ba1e Do a CHECK_FOR_INTERRUPTS after emitting a message of less than ERROR
severity.  This is to ensure the user can cancel a query that's spitting
out lots of notice/warning messages, even if they're coming from a loop
that doesn't otherwise contain a CHECK_FOR_INTERRUPTS.  Per gripe from
Stephen Frost.
2006-10-01 22:08:18 +00:00
Tom Lane
514ce7a331 Fix overly enthusiastic Assert introduced in 8.1: it's expecting a
CaseTestExpr, but forgot that the optimizer is sometimes able to replace
CaseTestExpr by Const.
2006-10-01 17:23:38 +00:00
Tom Lane
352a56ba68 Allow assignment to array elements not contiguous with those already
present; intervening positions are filled with nulls.  This behavior
is required by SQL99 but was not implementable before 8.2 due to lack
of support for nulls in arrays.  I have only made it work for the
one-dimensional case, which is all that SQL99 requires.  It seems quite
complex to get it right in higher dimensions, and since we never allowed
extension at all in higher dimensions, I think that must count as a
future feature addition not a bug fix.
2006-09-29 21:22:21 +00:00
Tom Lane
f213131f20 Fix IS NULL and IS NOT NULL tests on row-valued expressions to conform to
the SQL spec, viz IS NULL is true if all the row's fields are null, IS NOT
NULL is true if all the row's fields are not null.  The former coding got
this right for a limited number of cases with IS NULL (ie, those where it
could disassemble a ROW constructor at parse time), but was entirely wrong
for IS NOT NULL.  Per report from Teodor.

I desisted from changing the behavior for arrays, since on closer inspection
it's not clear that there's any support for that in the SQL spec.  This
probably needs more consideration.
2006-09-28 20:51:43 +00:00