Instead, add a function pg_tablespace_location(oid) used to return
the same information, and do this by reading the symbolic link.
Doing it this way makes it possible to relocate a tablespace when the
database is down by simply changing the symbolic link.
This patch creates an API whereby a btree index opclass can optionally
provide non-SQL-callable support functions for sorting. In the initial
patch, we only use this to provide a directly-callable comparator function,
which can be invoked with a bit less overhead than the traditional
SQL-callable comparator. While that should be of value in itself, the real
reason for doing this is to provide a datatype-extensible framework for
more aggressive optimizations, as in Peter Geoghegan's recent work.
Robert Haas and Tom Lane
invalid-page hash table, PANIC immediately. Immediate PANIC is much better
than waiting for end-of-recovery, which is what we did before, because the
end-of-recovery might not come until months later if this is a standby
server.
Also refrain from creating a restartpoint if there are invalid-page entries
in the hash table. Restarting recovery from such a restartpoint would not
see the invalid references, and wouldn't be able to cross-check them when
consistency is reached. That wouldn't matter when things are going smoothly,
but the more sanity checks you have the better.
Fujii Masao
In the previous coding, callers were faced with an awkward choice:
look up the name, do permissions checks, and then lock the table; or
look up the name, lock the table, and then do permissions checks.
The first choice was wrong because the results of the name lookup
and permissions checks might be out-of-date by the time the table
lock was acquired, while the second allowed a user with no privileges
to interfere with access to a table by users who do have privileges
(e.g. if a malicious backend queues up for an AccessExclusiveLock on
a table on which AccessShareLock is already held, further attempts
to access the table will be blocked until the AccessExclusiveLock
is obtained and the malicious backend's transaction rolls back).
To fix, allow callers of RangeVarGetRelid() to pass a callback which
gets executed after performing the name lookup but before acquiring
the relation lock. If the name lookup is retried (because
invalidation messages are received), the callback will be re-executed
as well, so we get the best of both worlds. RangeVarGetRelid() is
renamed to RangeVarGetRelidExtended(); callers not wishing to supply
a callback can continue to invoke it as RangeVarGetRelid(), which is
now a macro. Since the only one caller that uses nowait = true now
passes a callback anyway, the RangeVarGetRelid() macro defaults nowait
as well. The callback can also be used for supplemental locking - for
example, REINDEX INDEX needs to acquire the table lock before the index
lock to reduce deadlock possibilities.
There's a lot more work to be done here to fix all the cases where this
can be a problem, but this commit provides the general infrastructure
and fixes the following specific cases: REINDEX INDEX, REINDEX TABLE,
LOCK TABLE, and and DROP TABLE/INDEX/SEQUENCE/VIEW/FOREIGN TABLE.
Per discussion with Noah Misch and Alvaro Herrera.
The EvalPlanQual machinery assumes that whole-row Vars generated for the
outputs of non-table RTEs will be of composite types. However, for the
case where the RTE is a function call returning a scalar type, we were
doing the wrong thing, as a result of sharing code with a parser case
where the function's scalar output is wanted. (Or at least, that's what
that case has done historically; it does seem a bit inconsistent.)
To fix, extend makeWholeRowVar's API so that it can support both use-cases.
This fixes Belinda Cussen's report of crashes during concurrent execution
of UPDATEs involving joins to the result of UNNEST() --- in READ COMMITTED
mode, we'd run the EvalPlanQual machinery after a conflicting row update
commits, and it was expecting to get a HeapTuple not a scalar datum from
the "wholerowN" variable referencing the function RTE.
Back-patch to 9.0 where the current EvalPlanQual implementation appeared.
In 9.1 and up, this patch also fixes failure to attach the correct
collation to the Var generated for a scalar-result case. An example:
regression=# select upper(x.*) from textcat('ab', 'cd') x;
ERROR: could not determine which collation to use for upper() function
In the original implementation, a range-contained-by search had to scan
the entire index because an empty range could be lurking anywhere.
Improve that by adding a flag to upper GiST entries that says whether the
represented subtree contains any empty ranges.
Also, make a simple mod to the penalty function to discourage empty ranges
from getting pushed into subtrees without any. This needs more work, and
the picksplit function should be taught about it too, but that code can be
improved without causing an on-disk compatibility break; so we'll leave it
for another day.
Since we're breaking on-disk compatibility of range values anyway, I took
the opportunity to reorganize the range flags bits; the unused
RANGE_xB_NULL bits are now adjacent, which might open the door for using
them in some other way later.
In passing, remove the GiST range opclass entry for <>, which doesn't seem
like it can really be indexed usefully.
Alexander Korotkov, with some editorializing by Tom
This adds some I/O stats to the logging of autovacuum (when the
operation takes long enough that log_autovacuum_min_duration causes it
to be logged), so that it is easier to tune. Notably, it adds buffer
I/O counts (hits, misses, dirtied) and read and write rate.
Authors: Greg Smith and Noah Misch
This speeds up snapshot-taking and reduces ProcArrayLock contention.
Also, the PGPROC (and PGXACT) structures used by two-phase commit are
now allocated as part of the main array, rather than in a separate
array, and we keep ProcArray sorted in pointer order. These changes
are intended to minimize the number of cache lines that must be pulled
in to take a snapshot, and testing shows a substantial increase in
performance on both read and write workloads at high concurrencies.
Pavan Deolasee, Heikki Linnakangas, Robert Haas
The WITH [NO] DATA option was not supported, nor the ability to specify
replacement column names; the former limitation wasn't even documented, as
per recent complaint from Naoya Anzai. Fix by moving the responsibility
for supporting these options into the executor. It actually takes less
code this way ...
catversion bump due to change in representation of IntoClause, which might
affect stored rules.
It's not clear that a per-datatype typanalyze function would be any more
useful than a generic typanalyze for ranges. What *is* clear is that
letting unprivileged users select typanalyze functions is a crash risk or
worse. So remove the option from CREATE TYPE AS RANGE, and instead put in
a generic typanalyze function for ranges. The generic function does
nothing as yet, but hopefully we'll improve that before 9.2 release.
Per discussion, the zero-argument forms aren't really worth the catalog
space (just write 'empty' instead). The one-argument forms have some use,
but they also have a serious problem with looking too much like functional
cast notation; to the point where in many real use-cases, the parser would
misinterpret what was wanted.
Committing this as a separate patch, with the thought that we might want
to revert part or all of it if we can think of some way around the cast
ambiguity.
Implement these tests directly instead of constructing a singleton range
and then applying range-contains. This saves a range serialize/deserialize
cycle as well as a couple of redundant bound-comparison steps, and adds
very little code on net.
Remove elem_contained_by_range from the GiST opclass: it doesn't belong
there because there is no way to use it in an index clause (where the
indexed column would have to be on the left). Its commutator is in the
opclass, and that's what counts.
Per discussion, relax the range input/construction rules so that the
only hard error is lower bound > upper bound. Cases where the lower
bound is <= upper bound, but the range nonetheless normalizes to empty,
are now permitted.
Fix core dump in range_adjacent when bounds are infinite. Marginal
cleanup of regression test cases, some more code commenting.
Fix up some infelicitous coding in DefineRange, and add some missing error
checks. Rearrange operator strategy number assignments for GiST anyrange
opclass so that they don't make such a mess of opr_sanity's table of
operator names associated with different strategy numbers. Assign
hopefully-temporary selectivity estimators to range operators that didn't
have one --- poor as the estimates are, they're still a lot better than the
default 0.5 estimate, and they'll shut up the opr_sanity test that wants to
see selectivity estimators on all built-in operators.
This gets rid of an impressive amount of duplicative code, with only
minimal behavior changes. DROP FOREIGN DATA WRAPPER now requires object
ownership rather than superuser privileges, matching the documentation
we already have. We also eliminate the historical warning about dropping
a built-in function as unuseful. All operations are now performed in the
same order for all object types handled by dropcmds.c.
KaiGai Kohei, with minor revisions by me
Use of anynonarray was a crude hack to get around ambiguity versus the
array inclusion operators of the same names. My previous patch to extend
the parser's type resolution heuristics makes that unnecessary, so use
the more general declaration instead. This eliminates a wart that these
operators couldn't be used with ranges over arrays, which are otherwise
supported just fine.
Also, mark range_before and range_after as commutator operators,
per discussion with Jeff Davis.
A very long time ago, language names were specified as literals rather
than identifiers, so this code was added to do case-folding. But that
style has ben deprecated for many years so this isn't needed any more.
Language names will still be downcased when specified as unquoted
identifiers, but quoted identifiers or the old style using string
literals will be left as-is.
Fix assorted infelicities, such as dependency on OIDs that aren't
hardwired, as well as outright misdeclaration of daterange_canonical(),
which resulted in crashes if you invoked it directly. Add some more
regression tests to try to catch similar mistakes in future.
Move the responsibility for caching specialized information about range
types into the type cache, so that the catalog lookups only have to occur
once per session. Rearrange APIs a bit so that fn_extra caching is
actually effective in the GiST support code. (Use of OidFunctionCallN is
bad enough for performance in itself, but it also prevents the function
from exploiting fn_extra caching.)
The range I/O functions are still not very bright about caching repeated
lookups, but that seems like material for a separate patch.
Also, avoid unnecessary use of memcpy to fetch/store the range type OID and
flags, and don't use the full range_deserialize machinery when all we need
to see is the flags value.
Also fix API error in range_gist_penalty --- it was failing to set *penalty
for any case involving an empty range.
A range type whose element type has 'd' alignment must have 'd' alignment
itself, else there is no guarantee that the element value can be used
in-place. (Because range_deserialize uses att_align_pointer which forcibly
aligns the given pointer, violations of this rule did not lead to SIGBUS
but rather to garbage data being extracted, as in one of the added
regression test cases.)
Also, you can't put a toast pointer inside a range datum, since the
referenced value could disappear with the range datum still present.
For consistency with the handling of arrays and records, I also forced
decompression of in-line-compressed bound values. It would work to store
them as-is, but our policy is to avoid situations that might result in
double compression.
Add assorted regression tests for this, and bump catversion because of
fixes to built-in pg_type entries.
Also some marginal cleanup of inconsistent/unnecessary error checks.
No functional changes in this commit (except I could not resist the
temptation to re-word a couple of error messages). This is just manual
cleanup after pgindent to make the code look reasonably like other PG
code, in preparation for more detailed code review to come.
Previously we waited for wal_writer_delay before flushing WAL. Now
we also wake WALWriter as soon as a WAL buffer page has filled.
Significant effect observed on performance of asynchronous commits
by Robert Haas, attributed to the ability to set hint bits on tuples
earlier and so reducing contention caused by clog lookups.
This reverts commit 0180bd6180511875db046bf8ddcaa633a2952dfd.
contrib/userlock is gone, but user-level locking still exists,
and is exposed via the pg_advisory* family of functions.
This greatly reduces the WAL volume, especially when the table is narrow.
The overhead of locking the heap page is also reduced. Reduced WAL traffic
also makes it scale a lot better, if you run multiple COPY processes at
the same time.
Add PlaceHolderVar wrappers as needed to make UNION ALL sub-select output
expressions appear non-constant and distinct from each other. This makes
the world safe for add_child_rel_equivalences to do what it does. Before,
it was possible for that function to add identical expressions to different
EquivalenceClasses, which logically should imply merging such ECs, which
would be wrong; or to improperly add a constant to an EquivalenceClass,
drastically changing its behavior. Per report from Teodor Sigaev.
The only currently known consequence of this bug is "MergeAppend child's
targetlist doesn't match MergeAppend" planner failures in 9.1 and later.
I am suspicious that there may be other failure modes that could affect
older release branches; but in the absence of any hard evidence, I'll
refrain from back-patching further than 9.1.
a new macro, DatumGetInetPP(), that does not. This brings these macros
in line with other DatumGet*P() macros.
Backpatch to 8.3, where 1-byte header varlenas were introduced.
In a regular VACUUM, it's OK to skip pages for which a cleanup lock
isn't immediately available; the next VACUUM will deal with them. If
we're scanning the entire relation to advance relfrozenxid, we might
need to wait, but only if there are tuples on the page that actually
require freezing. These changes should greatly reduce the incidence
of of vacuum processes getting "stuck".
Simon Riggs and Robert Haas
It was inadvertently changed to 201111111, which is a wrong date. Change it
to current date, and remove the comment that was supposed to remind me to
fix it before committing.
If we use a PlaceHolderVar from the outer relation in an inner indexscan,
we need to reference the PlaceHolderVar as such as the value to be passed
in from the outer relation. The previous code effectively tried to
reconstruct the PHV from its component expression, which doesn't work since
(a) the Vars therein aren't necessarily bubbled up far enough, and (b) it
would be the wrong semantics anyway because of the possibility that the PHV
is supposed to have gone to null at some point before the current join.
Point (a) led to "variable not found in subplan target list" planner
errors, but point (b) would have led to silently wrong answers.
Per report from Roger Niederland.
There was a timing window between when oldestActiveXid was derived
and when it should have been derived that only shows itself under
heavy load. Move code around to ensure correct timing of derivation.
No change to StartupSUBTRANS() code, which is where this failed.
Bug report by Chris Redekop
If a tuple in a syscache contains an out-of-line toasted field, and we
try to fetch that field shortly after some other transaction has committed
an update or deletion of the tuple, there is a race condition: vacuum
could come along and remove the toast tuples before we can fetch them.
This leads to transient failures like "missing chunk number 0 for toast
value NNNNN in pg_toast_2619", as seen in recent reports from Andrew
Hammond and Tim Uckun.
The design idea of syscache is that access to stale syscache entries
should be prevented by relation-level locks, but that fails for at least
two cases where toasted fields are possible: ANALYZE updates pg_statistic
rows without locking out sessions that might want to plan queries on the
same table, and CREATE OR REPLACE FUNCTION updates pg_proc rows without
any meaningful lock at all.
The least risky fix seems to be an idea that Heikki suggested when we
were dealing with a related problem back in August: forcibly detoast any
out-of-line fields before putting a tuple into syscache in the first place.
This avoids the problem because at the time we fetch the parent tuple from
the catalog, we should be holding an MVCC snapshot that will prevent
removal of the toast tuples, even if the parent tuple is outdated
immediately after we fetch it. (Note: I'm not convinced that this
statement holds true at every instant where we could be fetching a syscache
entry at all, but it does appear to hold true at the times where we could
fetch an entry that could have a toasted field. We will need to be a bit
wary of adding toast tables to low-level catalogs that don't have them
already.) An additional benefit is that subsequent uses of the syscache
entry should be faster, since they won't have to detoast the field.
Back-patch to all supported versions. The problem is significantly harder
to reproduce in pre-9.0 releases, because of their willingness to flush
every entry in a syscache whenever the underlying catalog is vacuumed
(cf CatalogCacheFlushRelation); but there is still a window for trouble.
bgwriter is now a much less important process, responsible for page
cleaning duties only. checkpointer is now responsible for checkpoints
and so has a key role in shutdown. Later patches will correct doc
references to the now old idea that bgwriter performs checkpoints.
Has beneficial effect on performance at high write rates, but mainly
refactoring to more easily allow changes for power reduction by
simplifying previously tortuous code around required to allow page
cleaning and checkpointing to time slice in the same process.
Patch by me, Review by Dickson Guedes
This infrastructure doesn't in any way guarantee that the character
we produce will sort before the one we incremented; but it does at least
make it much more likely that we'll end up with something that is a valid
character, which improves our chances.
Kyotaro Horiguchi, with various adjustments by me.
We need not wait until the commit record is durably on disk, because
in the event of a crash the page we're updating with hint bits will
be gone anyway. Per off-list report from Heikki Linnakangas, this
can significantly degrade the performance of unlogged tables; I was
able to show a 2x speedup from this patch on a pgbench run with scale
factor 15. In practice, this will mostly help small, heavily updated
tables, because on larger tables you're unlikely to run into the same
row again before the commit record makes it out to disk.
If the right-hand side of a semijoin is unique, then we can treat it like a
normal join (or another way to say that is: we don't need to explicitly
unique-ify the data before doing it as a normal join). We were recognizing
such cases when the RHS was a sub-query with appropriate DISTINCT or GROUP
BY decoration, but there's another way: if the RHS is a plain relation with
unique indexes, we can check if any of the indexes prove the output is
unique. Most of the infrastructure for that was there already in the join
removal code, though I had to rearrange it a bit. Per reflection about a
recent example in pgsql-performance.
The uniqueness condition might fail to hold intra-transaction, and assuming
it does can give incorrect query results. Per report from Marti Raudsepp,
though this is not his proposed patch.
Back-patch to 9.0, where both these features were introduced. In the
released branches, add the new IndexOptInfo field to the end of the struct,
to try to minimize ABI breakage for third-party code that may be examining
that struct.
A transaction can export a snapshot with pg_export_snapshot(), and then
others can import it with SET TRANSACTION SNAPSHOT. The data does not
leave the server so there are not security issues. A snapshot can only
be imported while the exporting transaction is still running, and there
are some other restrictions.
I'm not totally convinced that we've covered all the bases for SSI (true
serializable) mode, but it works fine for lesser isolation modes.
Joachim Wieland, reviewed by Marko Tiikkaja, and rather heavily modified
by Tom Lane
Avoid possibly dumping core when pgstat_track_activity_query_size has a
less-than-default value; avoid uselessly searching for the query string
of a successfully-exited backend; don't bother putting out an ERRDETAIL if
we don't have a query to show; some other minor stylistic improvements.
To avoid minimize risk inside the postmaster, we subject this feature
to a number of significant limitations. We very much wish to avoid
doing any complex processing inside the postmaster, due to the
posssibility that the crashed backend has completely corrupted shared
memory. To that end, no encoding conversion is done; instead, we just
replace anything that doesn't look like an ASCII character with a
question mark. We limit the amount of data copied to 1024 characters,
and carefully sanity check the source of that data. While these
restrictions would doubtless be unacceptable in a general-purpose
logging facility, even this limited facility seems like an improvement
over the status quo ante.
Marti Raudsepp, reviewed by PDXPUG and myself
