Nobody using the missing_ok flag yet, but let's speculate that this will
be a better interface for future callers.
KaiGai Kohei, with some adjustments by me.
Relation rowtypes and automatically-generated array types do not need to
have their own extension membership dependency entries. If we create such
then it becomes more difficult to remove items from an extension, and it's
also harder for an extension upgrade script to make sure it duplicates the
dependencies created by the extension's regular installation script.
I changed the code in such a way that this happened in commit
988cccc620, I think because of worries about
the shell-type-replacement case; but that cure was worse than the disease.
It would only matter if one extension created a shell type that was
replaced with an auto-generated type in another extension, which seems
pretty far-fetched. Better to make this work unsurprisingly in normal
cases.
Report and patch by Robert Haas, comment adjustments by me.
We have seen one too many reports of people trying to use 9.1 extension
files in the old-fashioned way of sourcing them in psql. Not only does
that usually not work (due to failure to substitute for MODULE_PATHNAME
and/or @extschema@), but if it did work they'd get a collection of loose
objects not an extension. To prevent this, insert an \echo ... \quit
line that prints a suitable error message into each extension script file,
and teach commands/extension.c to ignore lines starting with \echo.
That should not only prevent any adverse consequences of loading a script
file the wrong way, but make it crystal clear to users that they need to
do it differently now.
Tom Lane, following an idea of Andrew Dunstan's. Back-patch into 9.1
... there is not going to be much value in this if we wait till 9.2.
It's been bothering me for several days that pretending that the cstring
data stored in a btree name_ops column is really a "name" Datum could lead
to reading past the end of memory. However, given the current memory
layout used for index-only scans in the btree code, a crash is in fact not
possible. Document that so we don't break it. I have not thought of any
other solutions that aren't fairly ugly too, and most of them lose the
functionality of index-only scans on name columns altogether, so this seems
like the way to go.
Dept. of second thoughts: as long as we've got that tlist hanging around
anyway, we can apply ExecTypeFromTL to it to get a suitable descriptor for
the ScanTupleSlot. This is a nicer solution than the previous one because
it eliminates some hard-wired knowledge about btree name_ops, and because
it avoids the somewhat shaky assumption that we needn't set up the scan
tuple descriptor in EXPLAIN_ONLY mode. It doesn't change what actually
happens at run-time though, and I'm still a bit nervous about that.
This commit changes index-only scans so that data is read directly from the
index tuple without first generating a faux heap tuple. The only immediate
benefit is that indexes on system columns (such as OID) can be used in
index-only scans, but this is necessary infrastructure if we are ever to
support index-only scans on expression indexes. The executor is now ready
for that, though the planner still needs substantial work to recognize
the possibility.
To do this, Vars in index-only plan nodes have to refer to index columns
not heap columns. I introduced a new special varno, INDEX_VAR, to mark
such Vars to avoid confusion. (In passing, this commit renames the two
existing special varnos to OUTER_VAR and INNER_VAR.) This allows
ruleutils.c to handle them with logic similar to what we use for subplan
reference Vars.
Since index-only scans are now fundamentally different from regular
indexscans so far as their expression subtrees are concerned, I also chose
to change them to have their own plan node type (and hence, their own
executor source file).
There's no particular advantage to this change on its face; indeed,
it's possible that this might be slightly slower than the old way.
But it makes this information more easily accessible to other
functions, and therefore paves the way for future code consolidation.
Performance isn't critical here, so there's no need to be smart about
how we do the search.
This is a heavily cut-down version of a patch from KaiGai Kohei,
with several fixes by me. Additional review from Dimitri Fontaine.
When I consolidated two copies of the HOT-chain search logic in commit
4da99ea423, I introduced a behavior
change: the old code wouldn't necessarily traverse the entire chain,
if the most recently returned tuple were updated while the HOT chain
traversal is in progress. The new behavior seems more correct, but
unfortunately, the code here relies on a scan with SnapshotNow failing
to see its own updates. That seems pretty shaky even with the old HOT
chain traversal behavior, since there's no guarantee that these
updates will always be HOT, but it's trivial to broke a failure with
the new HOT search logic. Fix by updating just the first matching
pg_constraint tuple, rather than all of them, since there should be
only one anyway. But since nobody has reproduced this failure on older
versions, no back-patch for now.
Report and test case by Alex Hunsaker; tablecmds.c changes by me.
The original idea of this patch was to make box picksplit run faster, by
eliminating unnecessary palloc() overhead, but that was obsoleted by the new
double-sorting split algorithm that doesn't call these functions so heavily
anymore. Nevertheless, the code looks better this way.
Original patch by me, reviewed and tidied up after the double-sorting patch
by Kevin Grittner.
We copy all the matched tuples off the page during _bt_readpage, instead of
expensively re-locking the page during each subsequent tuple fetch. This
costs a bit more local storage, but not more than 2*BLCKSZ worth, and the
reduction in LWLock traffic is certainly worth that. What's more, this
lets us get rid of the API wart in the original patch that said an index AM
could randomly decline to supply an index tuple despite having asserted
pg_am.amcanreturn. That will be important for future improvements in the
index-only-scan feature, since the executor will now be able to rely on
having the index data available.
transform_null_equals is only supposed to affect "foo = NULL" expressions
given directly by the user, not the internal "foo = NULL" expression
generated from CASE-WHEN.
This fixes bug #6242, reported by Sergey. Backpatch to all supported
branches.
When a btree index contains all columns required by the query, and the
visibility map shows that all tuples on a target heap page are
visible-to-all, we don't need to fetch that heap page. This patch depends
on the previous patches that made the visibility map reliable.
There's a fair amount left to do here, notably trying to figure out a less
chintzy way of estimating the cost of an index-only scan, but the core
functionality seems ready to commit.
Robert Haas and Ibrar Ahmed, with some previous work by Heikki Linnakangas.
In oder to exit on SIGTERM when in non-walsender code,
such as do_pg_stop_backup(), we need to set the interrupt
variables that are used there, and not just the walsender
local ones.
CREATE EXTENSION needs to transiently set search_path, as well as
client_min_messages and log_min_messages. We were doing this by the
expedient of saving the current string value of each variable, doing a
SET LOCAL, and then doing another SET LOCAL with the previous value at
the end of the command. This is a bit expensive though, and it also fails
badly if there is anything funny about the existing search_path value,
as seen in a recent report from Roger Niederland. Fortunately, there's a
much better way, which is to piggyback on the GUC infrastructure previously
developed for functions with SET options. We just open a new GUC nesting
level, do our assignments with GUC_ACTION_SAVE, and then close the nesting
level when done. This automatically restores the prior settings without a
re-parsing pass, so (in principle anyway) there can't be an error. And
guc.c still takes care of cleanup in event of an error abort.
The CREATE EXTENSION code for this was modeled on some much older code in
ri_triggers.c, which I also changed to use the better method, even though
there wasn't really much risk of failure there. Also improve the comments
in guc.c to reflect this additional usage.
Arrange for any problems with pre-existing settings to be reported as
WARNING not ERROR, so that we don't undesirably abort the loading of the
incoming add-on module. The bad setting is just discarded, as though it
had never been applied at all. (This requires a change in the API of
set_config_option. After some thought I decided the most potentially
useful addition was to allow callers to just pass in a desired elevel.)
Arrange to restore the complete stacked state of the variable, rather than
cheesily reinstalling only the active value. This ensures that custom GUCs
will behave unsurprisingly even when the module loading operation occurs
within nested subtransactions that have changed the active value. Since a
module load could occur as a result of, eg, a PL function call, this is not
an unlikely scenario.
This oversight meant that on Windows, the pg_settings view would not
display source file or line number information for values coming from
postgresql.conf, unless the backend had received a SIGHUP since starting.
In passing, also make the error detection in read_nondefault_variables a
tad more thorough, and fix it to not lose precision on float GUCs (these
changes are already in HEAD as of my previous commit).
We used to just remember the GucSource, but saving GucContext too provides
a little more information --- notably, whether a SET was done by a
superuser or regular user. This allows us to rip out the fairly dodgy code
that define_custom_variable used to use to try to infer the context to
re-install a pre-existing setting with. In particular, it now works for
a superuser to SET a extension's SUSET custom variable before loading the
associated extension, because GUC can remember whether the SET was done as
a superuser or not. The plperl regression tests contain an example where
this is useful.
Previously, the code assumed that the only possible action to take was
to delete files behind a certain cutoff point. The async notify code
was already a crock: it used a different "pagePrecedes" function for
truncation than for regular operation. By allowing it to pass a
callback to SlruScanDirectory it can do cleanly exactly what it needs to
do.
The clog.c code also had its own use for SlruScanDirectory, which is
made a bit simpler with this.
This variable provides only marginal error-prevention capability (since
it can only check the prefix of a qualified GUC name), and the consensus
is that that isn't worth the amount of hassle that maintaining the setting
creates for DBAs. So, let's just remove it.
With this commit, the system will silently accept a value for any qualified
GUC name at all, whether it has anything to do with any known extension or
not. (Unqualified names still have to match known built-in settings,
though; and you will get a WARNING at extension load time if there's an
unrecognized setting with that extension's prefix.)
There's still some discussion ongoing about whether to tighten that up and
if so how; but if we do come up with a solution, it's not likely to look
anything like custom_variable_classes.
Thus, an object referenced in a default expression could be dropped while
the function remained present. This was unaccountably missed in the
original patch to add default parameters for functions. Reported by
Pavel Stehule.
This patch has two distinct purposes: to report multiple problems in
postgresql.conf rather than always bailing out after the first one,
and to change the policy for whether changes are applied when there are
unrelated errors in postgresql.conf.
Formerly the policy was to apply no changes if any errors could be
detected, but that had a significant consistency problem, because in some
cases specific values might be seen as valid by some processes but invalid
by others. This meant that the latter processes would fail to adopt
changes in other parameters even though the former processes had done so.
The new policy is that during SIGHUP, the file is rejected as a whole
if there are any errors in the "name = value" syntax, or if any lines
attempt to set nonexistent built-in parameters, or if any lines attempt
to set custom parameters whose prefix is not listed in (the new value of)
custom_variable_classes. These tests should always give the same results
in all processes, and provide what seems a reasonably robust defense
against loading values from badly corrupted config files. If these tests
pass, all processes will apply all settings that they individually see as
good, ignoring (but logging) any they don't.
In addition, the postmaster does not abandon reading a configuration file
after the first syntax error, but continues to read the file and report
syntax errors (up to a maximum of 100 syntax errors per file).
The postmaster will still refuse to start up if the configuration file
contains any errors at startup time, but these changes allow multiple
errors to be detected and reported before quitting.
Alexey Klyukin, reviewed by Andy Colson and av (Alexander ?)
with some additional hacking by Tom Lane
We'll now use "exists" for EXISTS(SELECT ...), "array" for ARRAY(SELECT
...), or the sub-select's own result column name for a simple expression
sub-select. Previously, you usually got "?column?" in such cases.
Marti Raudsepp, reviewed by Kyotaro Horiugchi
pg_trgm was already doing this unofficially, but the implementation hadn't
been thought through very well and leaked memory. Restructure the core
GiST code so that it actually works, and document it. Ordinarily this
would have required an extra memory context creation/destruction for each
GiST index search, but I was able to avoid that in the normal case of a
non-rescanned search by finessing the handling of the RBTree. It used to
have its own context always, but now shares a context with the
scan-lifespan data structures, unless there is more than one rescan call.
This should make the added overhead unnoticeable in typical cases.
This code was looking at the sub-Query tree as seen in the parent query's
RangeTblEntry; but that's the pristine parser output, and what we need to
look at is the tree as it stands at the completion of planning. Otherwise
we might pick up a Var that references a subquery that got flattened and
hence has no RelOptInfo in the subroot. Per report from Peter Geoghegan.
If an indexable operator for a non-collatable indexed datatype has a
collatable right-hand input type, any OpExpr for it will be marked with a
nonzero inputcollid (since having one collatable input is sufficient to
make that happen). However, an index on a non-collatable column certainly
doesn't have any collation. This caused us to fail to match such operators
to their indexes, because indxpath.c required an exact match of index
collation and clause collation. It seems correct to allow a match when the
index is collation-less regardless of the clause's inputcollid: an operator
with both noncollatable and collatable inputs could perhaps depend on the
collation of the collatable input, but it could hardly expect the index for
the noncollatable input to have that same collation.
Per bug #6232 from Pierre Ducroquet. His example is specifically about
"hstore ? text" but the problem seems quite generic.
In hio.c, document how we avoid deadlock with respect to visibility map
buffer locks. In visibilitymap.c, update the LOCKING section of the
file header comment.
Both oversights noted by Heikki Linnakangas.
In commit c1d9579dd8, I changed things so
that the output of the Agg node that feeds the window functions would not
list any ungrouped Vars directly. Formerly, for example, the Agg tlist
might have included both "x" and "sum(x)", which is not really valid if
"x" isn't a grouping column. If we then had a window function ordering on
something like "sum(x) + 1", prepare_sort_from_pathkeys would find no exact
match for this in the Agg tlist, and would conclude that it must recompute
the expression. But it would break the expression down to just the Var
"x", which it would find in the tlist, and then rebuild the ORDER BY
expression using a reference to the subplan's "x" output. Now, after the
above-referenced changes, "x" isn't in the Agg tlist if it's not a grouping
column, so that prepare_sort_from_pathkeys fails with "could not find
pathkey item to sort", as reported by Bricklen Anderson.
The fix is to not break down Aggrefs into their component parts, but just
treat them as irreducible expressions to be sought in the subplan tlist.
This is definitely OK for the use with respect to window functions in
grouping_planner, since it just built the tlist being used on the same
basis. AFAICT it is safe for other uses too; most of the other call sites
couldn't encounter Aggrefs anyway.
In REPEATABLE READ (nee SERIALIZABLE) mode, an attempt to do
GetTransactionSnapshot() between AbortTransaction and CleanupTransaction
failed, because GetTransactionSnapshot would recompute the transaction
snapshot (which is already wrong, given the isolation mode) and then
re-register it in the TopTransactionResourceOwner, leading to an Assert
because the TopTransactionResourceOwner should be empty of resources after
AbortTransaction. This is the root cause of bug #6218 from Yamamoto
Takashi. While changing plancache.c to avoid requesting a snapshot when
handling a ROLLBACK masks the problem, I think this is really a snapmgr.c
bug: it's lower-level than the resource manager mechanism and should not be
shutting itself down before we unwind resource manager resources. However,
just postponing the release of the transaction snapshot until cleanup time
didn't work because of the circular dependency with
TopTransactionResourceOwner. Fix by managing the internal reference to
that snapshot manually instead of depending on TopTransactionResourceOwner.
This saves a few cycles as well as making the module layering more
straightforward. predicate.c's dependencies on TopTransactionResourceOwner
go away too.
I think this is a longstanding bug, but there's no evidence that it's more
than a latent bug, so it doesn't seem worth any risk of back-patching.
The code path that tried a generic plan, didn't like it, and then made a
custom plan was mistakenly passing the same copy of the query_list to the
planner both times. This doesn't work too well for nontrivial queries,
since the planner tends to scribble on its input. Diagnosis and fix by
Yamamoto Takashi.
I had copied-and-pasted a claim that we couldn't reach this point when
dealing with utility statements, but that was a leftover from when the
caller was required to supply a plan to start with. We now will go
through here at least once when handling a utility statement, so it
seems worth a check to see whether a snapshot is actually needed.
(Note that analyze_requires_snapshot is quite a cheap test.)
Per suggestion from Yamamoto Takashi. I don't think I believe that this
resolves his reported assertion failure; but it's worth changing anyway,
just to save a cycle or two.
The constraint exclusion feature checks for contradictions among scan
restriction clauses, as well as contradictions between those clauses and a
table's CHECK constraints. The first aspect of this testing can be useful
for non-table relations (such as subqueries or functions-in-FROM), but the
feature was coded with only the CHECK case in mind so we were applying it
only to plain-table RTEs. Move the relation_excluded_by_constraints call
so that it is applied to all RTEs not just plain tables. With the default
setting of constraint_exclusion this results in no extra work, but with
constraint_exclusion = ON we will detect optimizations that we missed
before (at the cost of more planner cycles than we expended before).
Per a gripe from Gunnlaugur Þór Briem. Experimentation with
his example also showed we were not being very bright about the case where
constraint exclusion is proven within a subquery within UNION ALL, so tweak
the code to allow set_append_rel_pathlist to recognize such cases.
This is not actually used anywhere yet, but it gets the basic
infrastructure in place. It is fairly likely that there are bugs, and
support for some important platforms may be missing, so we'll need to
refine this as we go along.
This provides information about the numbers of tuples that were visited
but not returned by table scans, as well as the numbers of join tuples
that were considered and discarded within a join plan node.
There is still some discussion going on about the best way to report counts
for outer-join situations, but I think most of what's in the patch would
not change if we revise that, so I'm going to go ahead and commit it as-is.
Documentation changes to follow (they weren't in the submitted patch
either).
Marko Tiikkaja, reviewed by Marc Cousin, somewhat revised by Tom
Per bug #6205, reported by Abel Abraham Camarillo Ojeda. This isn't a
particularly elegant fix, but I'm trying to minimize the chances of
causing yet another round of breakage.
Adjust regression tests to exercise this case.
The standardized errno code for "no such locale" failures is ENOENT, which
we were just reporting at face value, viz "No such file or directory".
Per gripe from Thom Brown, this might confuse users, so add an errdetail
message to clarify what it means. Also, report newlocale() failures as
ERRCODE_INVALID_PARAMETER_VALUE rather than using
errcode_for_file_access(), since newlocale()'s errno values aren't
necessarily tied directly to file access failures.
The regression tests were failing with CLOBBER_CACHE_ALWAYS enabled,
as reported by buildfarm member jaguar. There was an Assert in
BuildCachedPlan that asserted that the CachedPlanSource hadn't been
invalidated since we called RevalidateCachedQuery, which in theory can't
happen because we are holding locks on all the relevant database objects.
However, CLOBBER_CACHE_ALWAYS generates a false positive by making an
invalidation happen anyway; and on reflection, that could also occur as a
result of a badly-timed sinval reset due to queue overflow. We could just
remove the Assert and forge ahead with the not-really-stale querytree, but
it seems safer to do another RevalidateCachedQuery call just to make real
sure everything's OK.
This reverts commit 79b2ee20c8, which proved
to not be very informative; it looks like the "pgstat wait timeout"
warnings in the buildfarm are just a symptom of running on heavily loaded
machines, and there isn't any weird mechanism causing them to appear.
To try to reduce the frequency of buildfarm failures from this effect,
increase PGSTAT_MAX_WAIT_TIME from 5 seconds to 10.
Also, arrange to not send a fresh inquiry message every single time through
the loop, as that seems more likely to cause problems (by swamping the
collector) than fix them. We'll now send an inquiry the first time through
the delay loop, and every 640 msec thereafter.
As observed by Heikki, we need not conflict on heap page locks during an
insert; heap page locks are only aggregated tuple locks, they don't imply
locking "gaps" as index page locks do. So we can avoid some unnecessary
conflicts, and also do the SSI check while not holding exclusive lock on
the target buffer.
Kevin Grittner, reviewed by Jeff Davis. Back-patch to 9.1.
This oversight led to a massive memory leak --- upwards of 10KB per tuple
--- during creation-time verification of an exclusion constraint based on a
GIST index. In most other scenarios it'd just be a leak of 10KB that would
be recovered at end of query, so not too significant; though perhaps the
leak would be noticeable in a situation where a GIST index was being used
in a nestloop inner indexscan. In any case, it's a real leak of long
standing, so patch all supported branches. Per report from Harald Fuchs.
Rewrite plancache.c so that a "cached plan" (which is rather a misnomer
at this point) can support generation of custom, parameter-value-dependent
plans, and can make an intelligent choice between using custom plans and
the traditional generic-plan approach. The specific choice algorithm
implemented here can probably be improved in future, but this commit is
all about getting the mechanism in place, not the policy.
In addition, restructure the API to greatly reduce the amount of extraneous
data copying needed. The main compromise needed to make that possible was
to split the initial creation of a CachedPlanSource into two steps. It's
worth noting in particular that SPI_saveplan is now deprecated in favor of
SPI_keepplan, which accomplishes the same end result with zero data
copying, and no need to then spend even more cycles throwing away the
original SPIPlan. The risk of long-term memory leaks while manipulating
SPIPlans has also been greatly reduced. Most of this improvement is based
on use of the recently-added MemoryContextSetParent primitive.
queuedForEmptying flag correctly on buffer when adding it to the queue.
Also, don't add buffer to the queue if it's there already. These were
harmless oversights; failing to set the flag just means that a buffer might
get added to the queue twice if more tuples are added to it (although that
can't actually happen at this point because all the upper buffers have
already been emptied), and having the same buffer twice in the emptying
queue is harmless. But better be tidy.
This function will be useful for altering the lifespan of a context after
creation (for example, by creating it under a transient context and later
reparenting it to belong to a long-lived context). It costs almost no new
code, since we can refactor what was there. Per my proposal of yesterday.
This addresses only those cases that are easy to fix by adding or
moving a const qualifier or removing an unnecessary cast. There are
many more complicated cases remaining.
We were doing some amazingly complicated things in order to avoid running
the very expensive identify_system_timezone() procedure during GUC
initialization. But there is an obvious fix for that, which is to do it
once during initdb and have initdb install the system-specific default into
postgresql.conf, as it already does for most other GUC variables that need
system-environment-dependent defaults. This means that the timezone (and
log_timezone) settings no longer have any magic behavior in the server.
Per discussion.
As per my recent proposal, this refactors things so that these typedefs and
macros are available in a header that can be included in frontend-ish code.
I also changed various headers that were undesirably including
utils/timestamp.h to include datatype/timestamp.h instead. Unsurprisingly,
this showed that half the system was getting utils/timestamp.h by way of
xlog.h.
No actual code changes here, just header refactoring.
When building a GiST index that doesn't fit in cache, buffers are attached
to some internal nodes in the index. This speeds up the build by avoiding
random I/O that would otherwise be needed to traverse all the way down the
tree to the find right leaf page for tuple.
Alexander Korotkov
Trailing-zero stripping applied by the FM specifier could strip zeroes
to the left of the decimal point, for a format with no digit positions
after the decimal point (such as "FM999.").
Reported and diagnosed by Marti Raudsepp, though I didn't use his patch.
With 9.1's use of Params to pass down values from NestLoop join nodes
to their inner plans, it is possible for a Param to have type RECORD, in
which case the set of fields comprising the value isn't determinable by
inspection of the Param alone. However, just as with a Var of type RECORD,
we can find out what we need to know if we can locate the expression that
the Param represents. We already knew how to do this in get_parameter(),
but I'd overlooked the need to be able to cope in get_name_for_var_field(),
which led to EXPLAIN failing with "record type has not been registered".
To fix, refactor the search code in get_parameter() so it can be used by
both functions.
Per report from Marti Raudsepp.
Adds additional test for active walsenders and closes a race
condition for when we failover when a new walsender was connecting.
Reported and fixed bu Fujii Masao. Review by Heikki Linnakangas
The code in shift_jis_20042euc_jis_2004() would fetch two bytes even when
only one remained in the string. Since conversion functions aren't
supposed to assume null-terminated input, this poses a small risk of
fetching past the end of memory and incurring SIGSEGV. No such crash has
been identified in the field, but we've certainly seen the equivalent
happen in other code paths, so patch this one all the way back.
Report and patch by Noah Misch.
Since the last couple of columns of pg_type are often NULL,
sizeof(FormData_pg_type) can be an overestimate of the actual size of the
tuple data part. Therefore memcpy'ing that much out of the catalog cache,
as analyze.c was doing, poses a small risk of copying past the end of
memory and incurring SIGSEGV. No such crash has been identified in the
field, but we've certainly seen the equivalent happen in other code paths,
so patch this one all the way back.
Per valgrind testing by Noah Misch, though this is not his proposed patch.
I chose to use SearchSysCacheCopy1 rather than inventing special-purpose
infrastructure for copying only the minimal part of a pg_type tuple.
It used to say
ERROR: invalid byte sequence for encoding "UTF8": 0xdb24
Change this to
ERROR: invalid byte sequence for encoding "UTF8": 0xdb 0x24
to make it clear that this is a byte sequence and not a code point.
Also fix the adjacent "character has no equivalent" message that has
the same issue.
If a sub-select's output column is a simple Var, recursively look for
statistics applying to that Var, and use them if available. The need for
this was foreseen ages ago, but we didn't have enough infrastructure to do
it with reasonable speed until just now.
We punt and stick with default estimates if the subquery uses set
operations, GROUP BY, or DISTINCT, since those operations would change the
underlying column statistics (particularly, the relative frequencies of
different values) beyond recognition. This means that the types of
sub-selects for which this improvement applies are fairly limited, since
most subqueries satisfying those restrictions would have gotten flattened
into the parent query anyway. But it does help for some cases, such as
subqueries with ORDER BY or LIMIT.
Since the subroots will surely link back to the same glob struct, this
necessarily leads to infinite recursion. Doh. Found while trying to
debug some other code.
walsender.h should depend on xlog.h, not vice versa. (Actually, the
inclusion was circular until a couple hours ago, which was even sillier;
but Bruce broke it in the expedient rather than logically correct
direction.) Because of that poor decision, plus blind application of
pgrminclude, we had a situation where half the system was depending on
xlog.h to include such unrelated stuff as array.h and guc.h. Clean up
the header inclusion, and manually revert a lot of what pgrminclude had
done so things build again.
This episode reinforces my feeling that pgrminclude should not be run
without adult supervision. Inclusion changes in header files in particular
need to be reviewed with great care. More generally, it'd be good if we
had a clearer notion of module layering to dictate which headers can sanely
include which others ... but that's a big task for another day.
Formerly, set_subquery_pathlist and other creators of plans for subqueries
saved only the rangetable and rowMarks lists from the lower-level
PlannerInfo. But there's no reason not to remember the whole PlannerInfo,
and indeed this turns out to simplify matters in a number of places.
The immediate reason for doing this was so that the subroot will still be
accessible when we're trying to extract column statistics out of an
already-planned subquery. But now that I've done it, it seems like a good
code-beautification effort in its own right.
I also chose to get rid of the transient subrtable and subrowmark fields in
SubqueryScan nodes, in favor of having setrefs.c look up the subquery's
RelOptInfo. That required changing all the APIs in setrefs.c to pass
PlannerInfo not PlannerGlobal, which was a large but quite mechanical
transformation.
One side-effect not foreseen at the beginning is that this finally broke
inheritance_planner's assumption that replanning the same subquery RTE N
times would necessarily give interchangeable results each time. That
assumption was always pretty risky, but now we really have to make a
separate RTE for each instance so that there's a place to carry the
separate subroots.
In the past, relhassubclass always remained true if a relation had ever had
child relations, even if the last subclass was long gone. While this had
only marginal performance implications in most cases, it was annoying, and
I'm now considering some planner changes that would raise the cost of a
false positive. It was previously impractical to fix this because of race
condition concerns. However, given the recent change that made tablecmds.c
take ShareExclusiveLock on relations that are gaining a child (commit
fbcf4b92aa), we can now allow ANALYZE to
clear the flag when it's no longer relevant. There is no additional
locking cost to do so, since ANALYZE takes ShareExclusiveLock anyway.
dots. I previously worked around this in initdb, mapping the known
problematic locale names to aliases that work, but Hiroshi Inoue pointed
out that that's not enough because even if you use one of the aliases, like
"Chinese_HKG", setlocale(LC_CTYPE, NULL) returns back the long form, ie.
"Chinese_Hong Kong S.A.R.". When we try to restore an old locale value by
passing that value back to setlocale(), it fails. Note that you are affected
by this bug also if you use one of those short-form names manually, so just
reverting the hack in initdb won't fix it.
To work around that, move the locale name mapping from initdb to a wrapper
around setlocale(), so that the mapping is invoked on every setlocale() call.
Also, add a few checks for failed setlocale() calls in the backend. These
calls shouldn't fail, and if they do there isn't much we can do about it,
but at least you'll get a warning.
Backpatch to 9.1, where the initdb hack was introduced. The Windows bug
affects older versions too if you set locale manually to one of the aliases,
but given the lack of complaints from the field, I'm hesitent to backpatch.
Examination of examples provided by Mark Kirkwood and others has convinced
me that actually commit 7f3eba30c9 was quite
a few bricks shy of a load. The useful part of that patch was clamping
ndistinct for the inner side of a semi or anti join, and the reason why
that's needed is that it's the only way that restriction clauses
eliminating rows from the inner relation can affect the estimated size of
the join result. I had not clearly understood why the clamping was
appropriate, and so mis-extrapolated to conclude that we should clamp
ndistinct for the outer side too, as well as for both sides of regular
joins. These latter actions were all wrong, and are reverted with this
patch. In addition, the clamping logic is now made to affect the behavior
of both paths in eqjoinsel_semi, with or without MCV lists to compare.
When we have MCVs, we suppose that the most common values are the ones
that are most likely to survive the decimation resulting from a lower
restriction clause, so we think of the clamping as eliminating non-MCV
values, or potentially even the least-common MCVs for the inner relation.
Back-patch to 8.4, same as previous fixes in this area.
This patch fixes an oversight in my commit
7f3eba30c9 of 2008-10-23. That patch
accounted for baserel restriction clauses that reduced the number of rows
coming out of a table (and hence the number of possibly-distinct values of
a join variable), but not for join restriction clauses that might have been
applied at a lower level of join. To account for the latter, look up the
sizes of the min_lefthand and min_righthand inputs of the current join,
and clamp with those in the same way as for the base relations.
Noted while investigating a complaint from Ben Chobot, although this in
itself doesn't seem to explain his report.
Back-patch to 8.4; previous versions used different estimation methods
for which this heuristic isn't relevant.
It is possible for VACUUM to scan no pages at all, if the visibility map
shows that all pages are all-visible. In this situation VACUUM has no new
information to report about the relation's tuple density, so it wasn't
changing pg_class.reltuples ... but it updated pg_class.relpages anyway.
That's wrong in general, since there is no evidence to justify changing the
density ratio reltuples/relpages, but it's particularly bad if the previous
state was relpages=reltuples=0, which means "unknown tuple density".
We just replaced "unknown" with "zero". ANALYZE would eventually recover
from this, but it could take a lot of repetitions of ANALYZE to do so if
the relation size is much larger than the maximum number of pages ANALYZE
will scan, because of the moving-average behavior introduced by commit
b4b6923e03.
The only known situation where we could have relpages=reltuples=0 and yet
the visibility map asserts everything's visible is immediately following
a pg_upgrade. It might be advisable for pg_upgrade to try to preserve the
relpages/reltuples statistics; but in any case this code is wrong on its
own terms, so fix it. Per report from Sergey Koposov.
Back-patch to 8.4, where the visibility map was introduced, same as the
previous change.
Previously, 'yesterday 04:00:00'::timestamp didn't do the same thing as
'04:00:00 yesterday'::timestamp, and the return value from the latter
was midnight rather than the specified time.
Dean Rasheed, with some stylistic changes
At least on this architecture, it's very important to spin on a
non-atomic instruction and only retry the atomic once it appears
that it will succeed. To fix this, split TAS() into two macros:
TAS(), for trying to grab the lock the first time, and TAS_SPIN(),
for spinning until we get it. TAS_SPIN() defaults to same as TAS(),
but we can override it when we know there's a better way.
It's likely that some of the other cases in s_lock.h require
similar treatment, but this is the only one we've got conclusive
evidence for at present.
tsvector_concat() allocated its result workspace using the "conservative"
estimate of the sum of the two input tsvectors' sizes. Unfortunately that
wasn't so conservative as all that, because it supposed that the number of
pad bytes required could not grow. Which it can, as per test case from
Jesper Krogh, if there's a mix of lexemes with positions and lexemes
without them in the input data. The fix is to assume that we might add
a not-previously-present pad byte for each and every lexeme in the two
inputs; which really is conservative, but it doesn't seem worthwhile to
try to be more precise.
This is an aboriginal bug in tsvector_concat, so back-patch to all
versions containing it.
When we implemented extensions, we made findDependentObjects() treat
EXTENSION dependency links similarly to INTERNAL links. However, that
logic contained an implicit assumption that an object could have at most
one INTERNAL dependency, so it did not work correctly for objects having
both INTERNAL and DEPENDENCY links. This led to failure to drop some
extension member objects when dropping the extension. Furthermore, we'd
never actually exercised the case of recursing to an internally-referenced
(owning) object from anything other than a NORMAL dependency, and it turns
out that passing the incoming dependency's flags to the owning object is
the Wrong Thing. This led to sometimes dropping a whole extension silently
when we should have rejected the drop command for lack of CASCADE.
Since we obviously were under-testing extension drop scenarios, add some
regression test cases. Unfortunately, such test cases require some
extensions (duh), so we can't test for problems in the core regression
tests. I chose to add them to the earthdistance contrib module, which is
a good test case because it has a dependency on the cube contrib module.
Back-patch to 9.1. Arguably these are pre-existing bugs in INTERNAL
dependency handling, but since it appears that the cases can never arise
pre-9.1, I'll refrain from back-patching the logic changes further than
that.
When creating a new schema for a non-relocatable extension, we neglected
to check whether the calling user has permission to create schemas.
That didn't matter in the original coding, since we had already checked
superuserness, but in the new dispensation where users need not be
superusers, we should check it. Use CreateSchemaCommand() rather than
calling NamespaceCreate() directly, so that we also enforce the rules
about reserved schema names.
Per complaint from KaiGai Kohei, though this isn't the same as his patch.
set_append_rel_pathlist supposed that, while computing per-column width
estimates for the appendrel, it could ignore child rels for which the
translated reltargetlist entry wasn't a Var. This gave rise to completely
silly estimates in some common cases, such as constant outputs from some or
all of the arms of a UNION ALL. Instead, fall back on get_typavgwidth to
estimate from the value's datatype; which might be a poor estimate but at
least it's not completely wacko.
That problem was exposed by an Assert in set_subquery_size_estimates, which
unfortunately was still overoptimistic even with that fix, since we don't
compute attr_widths estimates for appendrels that are entirely excluded by
constraints. So remove the Assert; we'll just fall back on get_typavgwidth
in such cases.
Also, since set_subquery_size_estimates calls set_baserel_size_estimates
which calls set_rel_width, there's no need for set_subquery_size_estimates
to call get_typavgwidth; set_rel_width will handle it for us if we just
leave the estimate set to zero. Remove the unnecessary code.
Per report from Erik Rijkers and subsequent investigation.
The previous coding would result in deleting and not re-creating the
extension membership pg_depend rows, since there was no
CommandCounterIncrement that would allow recordDependencyOnCurrentExtension
to see that the deletion had happened. Make it work like the shell type
case, ie, keep the existing entries (and then throw an error if they're for
the wrong extension).
Per bug #6172 from Hitoshi Harada. Investigation and fix by Dimitri
Fontaine.
Due to tuple-slot mismanagement, evaluation of WHEN conditions for AFTER
ROW UPDATE triggers could crash if there had been a BEFORE ROW trigger
fired for the same update. Fix by not trying to overload the use of
estate->es_trig_tuple_slot. Per report from Yoran Heling.
Back-patch to 9.0, when trigger WHEN conditions were introduced.
As pointed out by Sergey Koposov, repeated invocations of tbm_lossify can
make building a large tidbitmap into an O(N^2) operation. To fix, make
sure we remove more than the minimum amount of information per call, and
add a fallback path to behave sanely if we're unable to fit the bitmap
within the requested amount of memory.
This has been wrong since the tidbitmap code was written, so back-patch
to all supported branches.
Perhaps we ought to add some other kind of documentation here instead,
but for now let's get rid of this woefully obsolete description of the
sinval machinery.
This requires adjusting the API for syscache callback functions: they now
get a hash value, not a TID, to identify the target tuple. Most of them
weren't paying any attention to that argument anyway, but plancache did
require a small amount of fixing.
Also, improve performance a trifle by avoiding sending duplicate inval
messages when a heap_update isn't changing the catcache lookup columns.
The TID isn't stable enough: we might queue an sinval event before a VACUUM
FULL, and then process it afterwards, when the target tuple no longer has
the same TID. So we must invalidate entries on the basis of hash value
only. The old coding can be shown to result in various bizarre,
hard-to-reproduce errors in the presence of concurrent VACUUM FULLs on
system catalogs, and could easily result in permanent catalog corruption,
up to and including complete loss of tables.
This commit is just a minimal fix that removes the unsafe comparison.
We should remove transmission of the tuple TID from sinval messages
altogether, and then arrange to suppress the extra message in the common
case of a heap_update that doesn't change the key hashvalue. But that's
going to be much more invasive, and will only produce a probably-marginal
performance gain, so it doesn't seem like material for a back-patch.
Back-patch to 9.0. Before that, VACUUM FULL refused to do any tuple moving
if it found any INSERT_IN_PROGRESS or DELETE_IN_PROGRESS tuples (and
CLUSTER would give up altogether), so there was no risk of moving a tuple
that might be the subject of an unsent sinval message.
We have to be sure that we have revalidated each nailed-in-cache relcache
entry before we try to use it to load data for some other relcache entry.
The introduction of "mapped relations" in 9.0 broke this, because although
we updated the state kept in relmapper.c early enough, we failed to
propagate that information into relcache entries soon enough; in
particular, we could try to fetch pg_class rows out of pg_class before
we'd updated its relcache entry's rd_node.relNode value from the map.
This bug accounts for Dave Gould's report of failures after "vacuum full
pg_class", and I believe that there is risk for other system catalogs
as well.
The core part of the fix is to copy relmapper data into the relcache
entries during "phase 1" in RelationCacheInvalidate(), before they'll be
used in "phase 2". To try to future-proof the code against other similar
bugs, I also rearranged the order in which nailed relations are visited
during phase 2: now it's pg_class first, then pg_class_oid_index, then
other nailed relations. This should ensure that RelationClearRelation can
apply RelationReloadIndexInfo to all nailed indexes without risking use
of not-yet-revalidated relcache entries.
Back-patch to 9.0 where the relation mapper was introduced.
This works around the problem that a catalog cache entry might contain a
toast pointer that we try to dereference just as a VACUUM FULL completes
on that catalog. We will see the sinval message on the cache entry when
we acquire lock on the toast table, but by that point we've already told
tuptoaster.c "here's the pointer to fetch", so it's difficult from a code
structural standpoint to update the pointer before we use it. Much less
painful to ensure that toast pointers are not invalidated in the first
place. We have to add a bit of code to deal with the case that a value
that previously wasn't toasted becomes so; but that should be a
seldom-exercised corner case, so the inefficiency shouldn't be significant.
Back-patch to 9.0. In prior versions, we didn't allow CLUSTER on system
catalogs, and VACUUM FULL didn't result in reassignment of toast OIDs, so
there was no problem.
The previous code tried to synchronize by unlinking the init file twice,
but that doesn't actually work: it leaves a window wherein a third process
could read the already-stale init file but miss the SI messages that would
tell it the data is stale. The result would be bizarre failures in catalog
accesses, typically "could not read block 0 in file ..." later during
startup.
Instead, hold RelCacheInitLock across both the unlink and the sending of
the SI messages. This is more straightforward, and might even be a bit
faster since only one unlink call is needed.
This has been wrong since it was put in (in 2002!), so back-patch to all
supported releases.
This makes it clearer that the error message is perhaps not supposed
to be understood by users, and it also makes it somewhat clearer that
it was not accidentally omitted from translation.
Idea from Heikki Linnakangas, except that we don't mark "Reason code"
for translation at this point, because that would make the
implementation too cumbersome.
When updating or deleting a system catalog tuple, it's necessary to acquire
RowExclusiveLock on the catalog before looking up the tuple; otherwise a
concurrent VACUUM FULL on the catalog might move the tuple to a different
TID before we can apply the update. Coding patterns that find the tuple
via a table scan aren't at risk here, but when obtaining the tuple from a
catalog cache, correct ordering is important; and several routines in
foreigncmds.c got it wrong. Noted while running the regression tests in
parallel with VACUUM FULL of assorted system catalogs.
For consistency I moved all the heap_open calls to the starts of their
functions, including a couple for which there was no actual bug.
Back-patch to 8.4 where foreigncmds.c was added.
The statement start timestamp was not set before initiating the transaction
that is used to look up client authentication information in pg_authid.
In consequence, enable_sig_alarm computed a wrong value (far in the past)
for statement_fin_time. That didn't have any immediate effect, because the
timeout alarm was set without reference to statement_fin_time; but if we
subsequently blocked on a lock for a short time, CheckStatementTimeout
would consult the bogus value when we cancelled the lock timeout wait,
and then conclude we'd timed out, leading to immediate failure of the
connection attempt. Thus an innocent "vacuum full pg_authid" would cause
failures of concurrent connection attempts. Noted while testing other,
more serious consequences of vacuum full on system catalogs.
We should set the statement timestamp before StartTransactionCommand(),
so that the transaction start timestamp is also valid. I'm not sure if
there are any non-cosmetic effects of it not being valid, but the xact
timestamp is at least sent to the statistics machinery.
Back-patch to 9.0. Before that, the client authentication timeout was done
outside any transaction and did not depend on this state to be valid.
poll() is preferred over select() on platforms where both are available,
because it tends to be a bit faster and it doesn't have an arbitrary limit
on the range of FD numbers that can be accessed. The FD range limit does
not appear to be a risk factor for any 9.1 usages, so this doesn't need to
be back-patched, but we need to have it in place if we keep on expanding
the uses of WaitLatch.
The latch infrastructure is now capable of detecting all cases where the
walsender loop needs to wake up, so there is no reason to have an arbitrary
timeout.
Also, modify the walsender loop logic to follow the standard pattern of
ResetLatch, test for work to do, WaitLatch. The previous coding was both
hard to follow and buggy: it would sometimes busy-loop despite having
nothing available to do, eg between receipt of a signal and the next time
it was caught up with new WAL, and it also had interesting choices like
deciding to update to WALSNDSTATE_STREAMING on the strength of information
known to be obsolete.
This is in hopes of learning more about what causes "pgstat wait timeout"
warnings in the buildfarm. This patch should probably be reverted once
we've learned what we can. As coded, it will result in regression test
"failures" at half the delay that the existing code does, so I expect
to see a few more than before.
In pursuit of this (and with the expectation that WaitLatch will be needed
in more places), convert the latch field that was already added to PGPROC
for sync rep into a generic latch that is activated for all PGPROC-owning
processes, and change many of the standard backend signal handlers to set
that latch when a signal happens. This will allow WaitLatch callers to be
wakened properly by these signals.
In passing, fix a whole bunch of signal handlers that had been hacked to do
things that might change errno, without adding the necessary save/restore
logic for errno. Also make some minor fixes in unix_latch.c, and clean
up bizarre and unsafe scheme for disowning the process's latch. Much of
this has to be back-patched into 9.1.
Peter Geoghegan, with additional work by Tom
streamed backup, throw an error and refuse to start up. The restore has not
finished correctly in that case and the data directory is possibly corrupt.
We already errored out in case of archive recovery, but could not during
crash recovery because we couldn't distinguish between the case that
pg_start_backup() was called and the database then crashed (must not error,
data is OK), and the case that we're restoring from a backup and not all
the needed WAL was replayed (data can be corrupt).
To distinguish those cases, add a line to backup_label to indicate
whether the backup was taken with pg_start/stop_backup(), or by streaming
(ie. pg_basebackup).
This requires re-initdb, because of a new field added to the control file.
The original definition had the problem that timeouts exceeding about 2100
seconds couldn't be specified on 32-bit machines. Milliseconds seem like
sufficient resolution, and finer grain than that would be fantasy anyway
on many platforms.
Back-patch to 9.1 so that this aspect of the latch API won't change between
9.1 and later releases.
Peter Geoghegan
Improve the documentation around weak-memory-ordering risks, and do a pass
of general editorialization on the comments in the latch code. Make the
Windows latch code more like the Unix latch code where feasible; in
particular provide the same Assert checks in both implementations.
Fix poorly-placed WaitLatch call in syncrep.c.
This patch resolves, for the moment, concerns around weak-memory-ordering
bugs in latch-related code: we have documented the restrictions and checked
that existing calls meet them. In 9.2 I hope that we will install suitable
memory barrier instructions in SetLatch/ResetLatch, so that their callers
don't need to be quite so careful.
Such a construction is useless since the lower PlaceHolderVar is already
nullable; no need to make it more so. Noted while pursuing bug #6154.
This is just a minor planner efficiency improvement, since the final plan
will come out the same anyway after PHVs are flattened. So not worth the
risk of back-patching.
Don't try to allocate the default value for a string relopt in the same
palloc chunk as the relopt_string struct. That didn't work too well if you
added a built-in string relopt in the stringRelOpts array, as it's not
possible to have an initializer for a variable length struct in C. This
makes the code slightly simpler too.
While we're at it, move the call to validator function in
add_string_reloption to before the allocation, so that if someone does pass
a bogus default value, we don't leak memory.
A PlaceHolderVar's expression might contain another, lower-level
PlaceHolderVar. If the outer PlaceHolderVar is used, the inner one
certainly will be also, and so we have to make sure that both of them get
into the placeholder_list with correct ph_may_need values during the
initial pre-scan of the query (before deconstruct_jointree starts).
We did this correctly for PlaceHolderVars appearing in the query quals,
but overlooked the issue for those appearing in the top-level targetlist;
with the result that nested placeholders referenced only in the targetlist
did not work correctly, as illustrated in bug #6154.
While at it, add some error checking to find_placeholder_info to ensure
that we don't try to create new placeholders after it's too late to do so;
they have to all be created before deconstruct_jointree starts.
Back-patch to 8.4 where the PlaceHolderVar mechanism was introduced.
Somebody thought it'd be cute to invent a set of Node tag numbers that were
defined independently of, and indeed conflicting with, the main tag-number
list. While this accidentally failed to fail so far, it would certainly
lead to trouble as soon as anyone wanted to, say, apply copyObject to these
node types. Clang was already complaining about the use of makeNode on
these tags, and I think quite rightly so. Fix by pushing these node
definitions into the mainstream, including putting replnodes.h where it
belongs.
The previous limit of 1024 was set on the assumption that all modern syslog
implementations have line length limits of 2KB or so. However, this is
false, as at least Solaris and sysklogd truncate at only 1KB. 900 seems
to leave enough room for the max likely length of the tacked-on prefixes,
so let's go with that.
As with the previous change, it doesn't seem wise to back-patch this into
already-released branches; but it should be OK to sneak it into 9.1.
Noah Misch
Instead of entering them on transaction startup, we materialize them
only when someone wants to wait, which will occur only during CREATE
INDEX CONCURRENTLY. In Hot Standby mode, the startup process must also
be able to probe for conflicting VXID locks, but the lock need never be
fully materialized, because the startup process does not use the normal
lock wait mechanism. Since most VXID locks never need to touch the
lock manager partition locks, this can significantly reduce blocking
contention on read-heavy workloads.
Patch by me. Review by Jeff Davis.
glibc renders random() thread-safe by wrapping a futex lock around it;
testing reveals that this limits the performance of pgbench on machines
with many CPU cores. Rather than switching to random_r(), which is
only available on GNU systems and crashes unless you use undocumented
alchemy to initialize the random state properly, switch to our built-in
implementation of erand48(), which is both thread-safe and concurrent.
Since the list of reasons not to use the operating system's erand48()
is getting rather long, rename ours to pg_erand48() (and similarly
for our implementations of lrand48() and srand48()) and just always
use those. We were already doing this on Cygwin anyway, and the
glibc implementation is not quite thread-safe, so pgbench wouldn't
be able to use that either.
Per discussion with Tom Lane.
This kluge was inserted in a spot apparently chosen at random: the lock
manager's state is not yet fully set up for the wait, and in particular
LockWaitCancel hasn't been armed by setting lockAwaited, so the ProcLock
will not get cleaned up if the ereport is thrown. This seems to not cause
any observable problem in trivial test cases, because LockReleaseAll will
silently clean up the debris; but I was able to cause failures with tests
involving subtransactions.
Fixes breakage induced by commit c85c941470.
Back-patch to all affected branches.
It was initialized in the wrong place and to the wrong value. With bad
luck this could result in incorrect query-cancellation failures in hot
standby sessions, should a HS backend be holding pin on buffer number 1
while trying to acquire a lock.
Testing shows that the overhead of acquiring and releasing
SInvalReadLock and msgNumLock on high-core count boxes can waste a lot
of CPU time and hurt performance. This patch adds a per-backend flag
that allows us to skip all that locking in most cases. Further
testing shows that this improves performance even when sinval traffic
is very high.
Patch by me. Review and testing by Noah Misch.