Tom Lane [Sat, 12 Jul 2008 02:28:43 +0000 (02:28 +0000)]
Don't make --enable-cassert turn on RANDOMIZE_ALLOCATED_MEMORY automatically;
it's just too dang expensive. Per recent discussion, but I just got my
nose rubbed in it again while doing some performance checking.
Tom Lane [Sat, 12 Jul 2008 00:44:38 +0000 (00:44 +0000)]
Const-ify the arguments of str_tolower() and friends to suppress compile
warnings. Clean up various unneeded cruft that was left behind after
creating those routines. Introduce some convenience functions str_tolower_z
etc to eliminate tedious and error-prone double arguments in formatting.c.
(Currently there seems no need to export the latter, but maybe reconsider
this later.)
Make sure we only try to free snapshots that have been passed through
CopySnapshot, per Neil Conway. Also add a comment about the assumption in
GetSnapshotData that the argument is statically allocated.
Also, fix some more typos in comments in snapmgr.c.
Tom Lane [Thu, 10 Jul 2008 22:08:17 +0000 (22:08 +0000)]
Add unchangeable GUC "variables" segment_size, wal_block_size, and
wal_segment_size to make those configuration parameters available to clients,
in the same way that block_size was previously exposed. Bernd Helmle, with
comments from Abhijit Menon-Sen and some further tweaking by me.
Tom Lane [Thu, 10 Jul 2008 02:14:03 +0000 (02:14 +0000)]
Tighten up SS_finalize_plan's computation of valid_params to exclude Params of
the current query level that aren't in fact output parameters of the current
initPlans. (This means, for example, output parameters of regular subplans.)
To make this work correctly for output parameters coming from sibling
initplans requires rejiggering the API of SS_finalize_plan just a bit:
we need the siblings to be visible to it, rather than hidden as
SS_make_initplan_from_plan had been doing. This is really part of my response
to bug #4290, but I concluded this part probably shouldn't be back-patched,
since all that it's doing is to make a debugging cross-check tighter.
Tom Lane [Thu, 10 Jul 2008 01:17:29 +0000 (01:17 +0000)]
Fix mis-calculation of extParam/allParam sets for plan nodes, as seen in
bug #4290. The fundamental bug is that masking extParam by outer_params,
as finalize_plan had been doing, caused us to lose the information that
an initPlan depended on the output of a sibling initPlan. On reflection
the best thing to do seemed to be not to try to adjust outer_params for
this case but get rid of it entirely. The only thing it was really doing
for us was to filter out param IDs associated with SubPlan nodes, and that
can be done (with greater accuracy) while processing individual SubPlan
nodes in finalize_primnode. This approach was vindicated by the discovery
that the masking method was hiding a second bug: SS_finalize_plan failed to
remove extParam bits for initPlan output params that were referenced in the
main plan tree (it only got rid of those referenced by other initPlans).
It's not clear that this caused any real problems, given the limited use
of extParam by the executor, but it's certainly not what was intended.
I originally thought that there was also a problem with needing to include
indirect dependencies on external params in initPlans' param sets, but it
turns out that the executor handles this correctly so long as the depended-on
initPlan is earlier in the initPlans list than the one using its output.
That seems a bit of a fragile assumption, but it is true at the moment,
so I just documented it in some code comments rather than making what would
be rather invasive changes to remove the assumption.
Back-patch to 8.1. Previous versions don't have the case of initPlans
referring to other initPlans' outputs, so while the existing logic is still
questionable for them, there are not any known bugs to be fixed. So I'll
refrain from changing them for now.
Tom Lane [Wed, 9 Jul 2008 15:56:49 +0000 (15:56 +0000)]
Increase PG_SYSLOG_LIMIT (the max line length sent to syslog()) from 128 to
1024 to improve performance when sending large elog messages. Also add a
comment about why we use that number.
Since this represents an externally visible behavior change, and might
possibly result in portability issues, it seems best not to back-patch it.
Tom Lane [Tue, 8 Jul 2008 22:17:41 +0000 (22:17 +0000)]
Fix performance bug in write_syslog(): the code to preferentially break the
log message at newlines cost O(N^2) for very long messages with few or no
newlines. For messages in the megabyte range this became the dominant cost.
Per gripe from Achilleas Mantzios.
Patch all the way back, since this is a safe change with no portability
risks. I am also thinking of increasing PG_SYSLOG_LIMIT, but that should
be done separately.
Tom Lane [Mon, 7 Jul 2008 20:24:55 +0000 (20:24 +0000)]
Fix estimate_num_groups() to assume that GROUP BY expressions yielding boolean
results always contribute two groups, regardless of the expression contents.
This is very substantially more accurate than the regular heuristic for
certain boolean tests like "col IS NULL". Per gripe from Sam Mason.
Back-patch to all supported releases, since the behavior of
estimate_num_groups() hasn't changed all that much since 7.4.
Tom Lane [Mon, 7 Jul 2008 18:09:46 +0000 (18:09 +0000)]
Fix AT TIME ZONE (in all three variants) so that we first try to interpret
the timezone argument as a timezone abbreviation, and only try it as a full
timezone name if that fails. The zic database has four zones (CET, EET, MET,
WET) that are full daylight-savings zones and yet have names that are the
same as their abbreviations for standard time, resulting in ambiguity.
In the timestamp input functions we resolve the ambiguity by preferring the
abbreviation, and AT TIME ZONE should work the same way. (No functionality
is lost because the zic database also has other names for these zones, eg
Europe/Zurich.) Per gripe from Jaromir Talir.
Backpatch to 8.1. Older releases did not have the issue because AT TIME ZONE
only accepted abbreviations not zone names. (Thus, this patch also arguably
fixes a compatibility botch introduced at 8.1: in ambiguous cases we now
behave the same as 8.0 did.)
Tom Lane [Sun, 6 Jul 2008 19:48:45 +0000 (19:48 +0000)]
Prevent integer overflows during units conversion when displaying a GUC
variable that has units. Per report from Stefan Kaltenbrunner.
Backport to 8.2. I also backported my patch of 2007-06-21 that prevented
comparable overflows on the input side, since that now seems to have enough
field track record to be back-patched safely. That patch included addition
of hints listing the available unit names, which I did not bother to strip
out of it --- this will make a little more work for the translators, but
they can copy the translation from 8.3, and anyway an untranslated hint
is better than no hint.
Magnus Hagander [Fri, 4 Jul 2008 10:50:18 +0000 (10:50 +0000)]
Fix a couple of bugs in win32 shmem name generation:
* Don't cut off the prefix. With this fix, it's again readable.
* Properly store it in the Global namespace as intended.
Joe Conway [Thu, 3 Jul 2008 03:56:57 +0000 (03:56 +0000)]
When an ERROR happens on a dblink remote connection, take
pains to pass the ERROR message components locally, including
using the passed SQLSTATE. Also wrap the passed info in an
appropriate CONTEXT message. Addresses complaint by Henry
Combrinck. Joe Conway, with much good advice from Tom Lane.
Move volatility, language, etc. modifiers before function body in the pg_dump
output for CREATE FUNCTION. This makes it easier to read especially if the
function body is long.
Original idea and patch by Greg Sabino Mullane, though this is a stripped
down version of that.
Extend VacAttrStats to allow typanalyze functions to store statistic values
of different types than the underlying column. The capability isn't yet
used for anything, but will be required by upcoming patch to analyze
tsvector columns.
Magnus Hagander [Tue, 1 Jul 2008 06:08:31 +0000 (06:08 +0000)]
Split apart message_level_options into one set for server-side settings and
one for client-side, restoring the previous behaviour with different
sort order for the 'log' level. Also, remove redundant list of available
options, since the enum code will output it automatically.
Tom Lane [Tue, 1 Jul 2008 03:40:55 +0000 (03:40 +0000)]
Fix identify_system_timezone() so that it tests the behavior of the system
timezone setting in the current year and for 100 years back, rather than
always examining years 1904-2004. The original coding would have problems
distinguishing zones whose behavior diverged only after 2004; which is a
situation we will surely face sometime, if it's not out there already.
In passing, also prevent selection of the dummy "Factory" timezone, even
if that's exactly what the system is using. Reporting time as GMT seems
better than that.
Tom Lane [Tue, 1 Jul 2008 02:09:34 +0000 (02:09 +0000)]
Teach autovacuum how to determine whether a temp table belongs to a crashed
backend. If so, send a LOG message to the postmaster log, and if the table
is beyond the vacuum-for-wraparound horizon, forcibly drop it. Per recent
discussions. Perhaps we ought to back-patch this, but it probably needs
to age a bit in HEAD first.
Bruce Momjian [Mon, 30 Jun 2008 19:45:15 +0000 (19:45 +0000)]
Add psql TODO:
> o Add "auto" expanded mode that outputs in expanded format if
> "wrapped" mode can't wrap the output to the screen width
>
> http://archives.postgresql.org/pgsql-hackers/2008-05/msg00417.php
>
Bruce Momjian [Mon, 30 Jun 2008 19:41:37 +0000 (19:41 +0000)]
Add psql TODO item:
> o Add option to wrap column values at whitespace boundaries,
> rather than chopping them at a fixed width.
> Currently, "wrapped" format chops values into fixed
> widths. Perhaps the word wrapping could use the same
> algorithm documented in the W3C specification.
> http://archives.postgresql.org/pgsql-hackers/2008-05/msg00404.php
> http://www.w3.org/TR/CSS21/tables.html#auto-table-layout
Turn PGBE_ACTIVITY_SIZE into a GUC variable, track_activity_query_size.
As the buffer could now be a lot larger than before, and copying it could
thus be a lot more expensive than before, use strcpy instead of memcpy to
copy the query string, as was already suggested in comments. Also, only copy
the PgBackendStatus struct and string if the slot is in use.
Tom Lane [Sun, 29 Jun 2008 21:04:01 +0000 (21:04 +0000)]
Remove unnecessary coziness of GIN code with datum copying. Now that
space is tracked via GetMemoryChunkSpace, there's really no advantage
to duplicating datumCopy's innards here. This is one bit of my toast
indirection patch that should go in anyway.
Tom Lane [Fri, 27 Jun 2008 20:54:37 +0000 (20:54 +0000)]
Consider a clause to be outerjoin_delayed if it references the nullable side
of any lower outer join, even if it also references the non-nullable side and
so could not get pushed below the outer join anyway. We need this in case
the clause is an OR clause: if it doesn't get marked outerjoin_delayed,
create_or_index_quals() could pull an indexable restriction for the nullable
side out of it, leading to wrong results as demonstrated by today's bug
report from toruvinn. (See added regression test case for an example.)
In principle this has been wrong for quite a while. In practice I don't
think any branch before 8.3 can really show the failure, because
create_or_index_quals() will only pull out indexable conditions, and before
8.3 those were always strict. So though we might have improperly generated
null-extended rows in the outer join, they'd get discarded from the result
anyway. The gating factor that makes the failure visible is that 8.3
considers "col IS NULL" to be indexable. Hence I'm not going to risk
back-patching further than 8.3.
Tom Lane [Fri, 27 Jun 2008 03:56:55 +0000 (03:56 +0000)]
Improve planner's estimation of the size of an append relation: rather than
taking the maximum of any child rel's width, we should weight the widths
proportionally to the number of rows expected from each child. In hindsight
this is obviously correct because row width is really a proxy for the total
physical size of the relation. Per discussion with Scott Carey (bug #4264).
Tom Lane [Fri, 27 Jun 2008 00:36:16 +0000 (00:36 +0000)]
Modify the recently-added probe for -Wl,--as-needed some more, because RHEL-4
vintage Linux is even more broken than we realized: a link to libreadline
will succeed, and fail only at runtime. It seems that an AC_TRY_RUN test
is the only reliable way to check whether this is really safe. Per report
from Tatsuo.
Teodor Sigaev [Thu, 26 Jun 2008 16:06:37 +0000 (16:06 +0000)]
Fix bug "select lower('asd') = 'asd'" returns false with multibyte encoding
and non-C locale. Fix is just to use correct source's length for char2wchar
call.
Tom Lane [Tue, 24 Jun 2008 17:58:27 +0000 (17:58 +0000)]
Reduce the alignment requirement of type "name" from int to char, and arrange
to suppress zero-padding of "name" entries in indexes.
The alignment change is unlikely to save any space, but it is really needed
anyway to make the world safe for our widespread practice of passing plain
old C strings to functions that are declared as taking Name. In the previous
coding, the C compiler was entitled to assume that a Name pointer was
word-aligned; but we were failing to guarantee that. I think the reason
we'd not seen failures is that usually the only thing that gets done with
such a pointer is strcmp(), which is hard to optimize in a way that exploits
word-alignment. Still, some enterprising compiler guy will probably think
of a way eventually, or we might change our code in a way that exposes
more-obvious optimization opportunities.
The padding change is accomplished in one-liner fashion by declaring the
"name" index opclasses to use storage type "cstring" in pg_opclass.h.
Normally btree and hash don't allow a nondefault storage type, because they
don't have any provisions for converting the input datum to another type.
However, because name and cstring are effectively the same thing except for
padding, no conversion is needed --- we only need index_form_tuple() to treat
the datum as being cstring not name, and this is sufficient. This seems to
make for about a one-third reduction in the typical sizes of system catalog
indexes that involve "name" columns, of which we have many.
These two changes are only weakly related, but the alignment change makes
me feel safer that the padding change won't introduce problems, so I'm
committing them together.
Bruce Momjian [Mon, 23 Jun 2008 19:27:19 +0000 (19:27 +0000)]
Merge duplicate upper/lower/initcap() routines in oracle_compat.c and
formatting.c to use common code; remove duplicate functions and support
routines that are no longer needed.
Tom Lane [Mon, 23 Jun 2008 17:54:30 +0000 (17:54 +0000)]
Fix Gen_fmgrtab.sh to not rely on hard-wired knowledge of the column numbers
in pg_proc. Also make it not emit duplicate extern declarations, and make it
a bit more bulletproof in some other small ways. Likewise fix the equally
hard-wired, and utterly undocumented, knowledge in the MSVC build scripts.
For testing purposes and perhaps other uses in future, pull out that portion
of the MSVC scripts into a standalone perl script equivalent to
Gen_fmgrtab.sh, and make it generate actually identical output, rather than
just more-or-less-the-same output.
Motivated by looking at Pavel's variadic function patch. Whether or not
that gets accepted, we can be sure that pg_proc's column set will change
again in the future; it's time to not have to deal with this gotcha.
Tom Lane [Fri, 20 Jun 2008 00:24:53 +0000 (00:24 +0000)]
Seems I was too optimistic in supposing that sinval's maxMsgNum could be
read and written without a lock. The value itself is atomic, sure, but on
processors with weak memory ordering it's possible for a reader to see the
value change before it sees the associated message written into the buffer
array. Fix by introducing a spinlock that's used just to read and write
maxMsgNum. (We could do this with less overhead if we recognized a concept
of "memory access barrier"; is it worth introducing such a thing? At the
moment probably not --- I can't measure any clear slowdown from adding the
spinlock, so this solution is probably fine.) Per buildfarm results.
Tom Lane [Thu, 19 Jun 2008 21:32:56 +0000 (21:32 +0000)]
Rewrite the sinval messaging mechanism to reduce contention and avoid
unnecessary cache resets. The major changes are:
* When the queue overflows, we only issue a cache reset to the specific
backend or backends that still haven't read the oldest message, rather
than resetting everyone as in the original coding.
* When we observe backend(s) falling well behind, we signal SIGUSR1
to only one backend, the one that is furthest behind and doesn't already
have a signal outstanding for it. When it finishes catching up, it will
in turn signal SIGUSR1 to the next-furthest-back guy, if there is one that
is far enough behind to justify a signal. The PMSIGNAL_WAKEN_CHILDREN
mechanism is removed.
* We don't attempt to clean out dead messages after every message-receipt
operation; rather, we do it on the insertion side, and only when the queue
fullness passes certain thresholds.
* Split SInvalLock into SInvalReadLock and SInvalWriteLock so that readers
don't block writers nor vice versa (except during the infrequent queue
cleanout operations).
* Transfer multiple sinval messages for each acquisition of a read or
write lock.
Alvaro Herrera [Thu, 19 Jun 2008 00:46:06 +0000 (00:46 +0000)]
Improve our #include situation by moving pointer types away from the
corresponding struct definitions. This allows other headers to avoid including
certain highly-loaded headers such as rel.h and relscan.h, instead using just
relcache.h, heapam.h or genam.h, which are more lightweight and thus cause less
unnecessary dependencies.
Tom Lane [Wed, 18 Jun 2008 20:55:42 +0000 (20:55 +0000)]
Improve error reporting for problems in text search configuration files
by installing an error context subroutine that will provide the file name
and line number for all errors detected while reading a config file.
Some of the reader routines were already doing that in an ad-hoc way for
errors detected directly in the reader, but it didn't help for problems
detected in subroutines, such as encoding violations.
Back-patch to 8.3 because 8.3 is where people will be trying to debug
configuration files.
Tom Lane [Tue, 17 Jun 2008 20:07:08 +0000 (20:07 +0000)]
Remove freeBackends counter from the sinval shared memory area. We used to
use it to help enforce superuser_reserved_backends, but since 8.1 it's
just been dead weight.
Tom Lane [Tue, 17 Jun 2008 19:10:56 +0000 (19:10 +0000)]
Clean up some problems with redundant cross-type arithmetic operators. Add
int2-and-int8 implementations of the basic arithmetic operators +, -, *, /.
This doesn't really add any new functionality, but it avoids "operator is not
unique" failures that formerly occurred in these cases because the parser
couldn't decide whether to promote the int2 to int4 or int8. We could
alternatively have removed the existing cross-type operators, but
experimentation shows that the cost of an additional type coercion expression
node is noticeable compared to such cheap operators; so let's not give up any
performance here. On the other hand, I removed the int2-and-int4 modulo (%)
operators since they didn't seem as important from a performance standpoint.
Per a complaint last January from ykhuang.
Tom Lane [Tue, 17 Jun 2008 14:51:32 +0000 (14:51 +0000)]
Fix the code that adds regclass constants to a plan's list of relation OIDs
that it depends on for replan-forcing purposes. We need to consider plain OID
constants too, because eval_const_expressions folds a RelabelType atop a Const
to just a Const. This change could result in OID values that aren't really
for tables getting added to the dependency list, but the worst-case
consequence would be occasional useless replans. Per report from Gabriele
Messineo.
Tom Lane [Tue, 17 Jun 2008 00:52:43 +0000 (00:52 +0000)]
Clean up a number of bogosities around pltcl's handling of the Tcl "result":
1. Directly reading interp->result is deprecated in Tcl 8.0 and later;
you're supposed to use Tcl_GetStringResult. This code finally broke with
Tcl 8.5, because Tcl_GetVar can now have side-effects on interp->result even
though it preserves the logical state of the result. (There's arguably a
Tcl issue here, because Tcl_GetVar could invalidate the pointer result of a
just-preceding Tcl_GetStringResult, but I doubt the Tcl guys will see it as
a bug.)
2. We were being sloppy about the encoding of the result: some places would
push database-encoding data into the Tcl result, which should not happen,
and we were assuming that any error result coming back from Tcl was in the
database encoding, which is not a good assumption.
3. There were a lot of calls of Tcl_SetResult that uselessly specified
TCL_VOLATILE for constant strings. This is only a minor performance issue,
but I fixed it in passing since I had to look at all the calls anyway.
#2 is a live bug regardless of which Tcl version you are interested in,
so back-patch even to branches that are unlikely to be used with Tcl 8.5.
I went back as far as 8.0, which is as far as the patch applied easily;
7.4 was using a different error processing scheme that has got its own
problems :-(
Tom Lane [Mon, 16 Jun 2008 03:13:14 +0000 (03:13 +0000)]
Rewrite docs section about routine vacuuming --- it's gotten rather mangled of
late, with lots of redundancy, bad grammar, and just plain poor exposition.
Make it clear that autovacuum is now considered the normal solution.
Andrew Dunstan [Sun, 15 Jun 2008 21:46:02 +0000 (21:46 +0000)]
Add script to find .c and .h files that are missing CVS PostgreSQL markers
and add them. Avoids third party files or those that would cause regression
failures.
Tom Lane [Sun, 15 Jun 2008 16:29:05 +0000 (16:29 +0000)]
Make DROP INDEX lock the parent table before locking the index. This behavior
is necessary to avoid deadlock against ordinary queries, but we'd broken it
with recent changes that made the DROP machinery lock the index before
arriving at index_drop. Per intermittent buildfarm failures.