Originally, Postgres Lists were a more or less exact reimplementation of
Lisp lists, which consist of chains of separately-allocated cons cells,
each having a value and a next-cell link. We'd hacked that once before
(commit d0b4399d8) to add a separate List header, but the data was still
in cons cells. That makes some operations -- notably list_nth() -- O(N),
and it's bulky because of the next-cell pointers and per-cell palloc
overhead, and it's very cache-unfriendly if the cons cells end up
scattered around rather than being adjacent.
In this rewrite, we still have List headers, but the data is in a
resizable array of values, with no next-cell links. Now we need at
most two palloc's per List, and often only one, since we can allocate
some values in the same palloc call as the List header. (Of course,
extending an existing List may require repalloc's to enlarge the array.
But this involves just O(log N) allocations not O(N).)
Of course this is not without downsides. The key difficulty is that
addition or deletion of a list entry may now cause other entries to
move, which it did not before.
For example, that breaks foreach() and sister macros, which historically
used a pointer to the current cons-cell as loop state. We can repair
those macros transparently by making their actual loop state be an
integer list index; the exposed "ListCell *" pointer is no longer state
carried across loop iterations, but is just a derived value. (In
practice, modern compilers can optimize things back to having just one
loop state value, at least for simple cases with inline loop bodies.)
In principle, this is a semantics change for cases where the loop body
inserts or deletes list entries ahead of the current loop index; but
I found no such cases in the Postgres code.
The change is not at all transparent for code that doesn't use foreach()
but chases lists "by hand" using lnext(). The largest share of such
code in the backend is in loops that were maintaining "prev" and "next"
variables in addition to the current-cell pointer, in order to delete
list cells efficiently using list_delete_cell(). However, we no longer
need a previous-cell pointer to delete a list cell efficiently. Keeping
a next-cell pointer doesn't work, as explained above, but we can improve
matters by changing such code to use a regular foreach() loop and then
using the new macro foreach_delete_current() to delete the current cell.
(This macro knows how to update the associated foreach loop's state so
that no cells will be missed in the traversal.)
There remains a nontrivial risk of code assuming that a ListCell *
pointer will remain good over an operation that could now move the list
contents. To help catch such errors, list.c can be compiled with a new
define symbol DEBUG_LIST_MEMORY_USAGE that forcibly moves list contents
whenever that could possibly happen. This makes list operations
significantly more expensive so it's not normally turned on (though it
is on by default if USE_VALGRIND is on).
There are two notable API differences from the previous code:
* lnext() now requires the List's header pointer in addition to the
current cell's address.
* list_delete_cell() no longer requires a previous-cell argument.
These changes are somewhat unfortunate, but on the other hand code using
either function needs inspection to see if it is assuming anything
it shouldn't, so it's not all bad.
Programmers should be aware of these significant performance changes:
* list_nth() and related functions are now O(1); so there's no
major access-speed difference between a list and an array.
* Inserting or deleting a list element now takes time proportional to
the distance to the end of the list, due to moving the array elements.
(However, it typically *doesn't* require palloc or pfree, so except in
long lists it's probably still faster than before.) Notably, lcons()
used to be about the same cost as lappend(), but that's no longer true
if the list is long. Code that uses lcons() and list_delete_first()
to maintain a stack might usefully be rewritten to push and pop at the
end of the list rather than the beginning.
* There are now list_insert_nth...() and list_delete_nth...() functions
that add or remove a list cell identified by index. These have the
data-movement penalty explained above, but there's no search penalty.
* list_concat() and variants now copy the second list's data into
storage belonging to the first list, so there is no longer any
sharing of cells between the input lists. The second argument is
now declared "const List *" to reflect that it isn't changed.
This patch just does the minimum needed to get the new implementation
in place and fix bugs exposed by the regression tests. As suggested
by the foregoing, there's a fair amount of followup work remaining to
do.
Also, the ENABLE_LIST_COMPAT macros are finally removed in this
commit. Code using those should have been gone a dozen years ago.
Patch by me; thanks to David Rowley, Jesper Pedersen, and others
for review.
Discussion: https://postgr.es/m/11587.1550975080@sss.pgh.pa.us
This addresses a couple of issues in the code:
- Typos and inconsistencies in comments and function declarations.
- Removal of unreferenced function declarations.
- Removal of unnecessary compile flags.
- A cleanup error in regressplans.sh.
Author: Alexander Lakhin
Discussion: https://postgr.es/m/0c991fdf-2670-1997-c027-772a420c4604@gmail.com
There were a number of issues in the recent commits which include typos,
code and comments mismatch, leftover function declarations. Fix them.
Reported-by: Alexander Lakhin
Author: Alexander Lakhin, Amit Kapila and Amit Langote
Reviewed-by: Amit Kapila
Discussion: https://postgr.es/m/ef0c0232-0c1d-3a35-63d4-0ebd06e31387@gmail.com
We used the same slot to store a tuple from the index, and to store a
tuple from the table. That's not OK. It worked with the heap, because
heapam_getnextslot() stores a HeapTuple to the slot, and doesn't care how
large the tts_values/nulls arrays are. But when I played with a toy table
AM implementation that used a virtual tuple, it caused memory overruns.
In the passing, tidy up comments on the ioss_PscanLen fields.
86b85044e rewrote how COPY FROM works to allow multiple tuple buffers to
exist to once thus allowing multi-inserts to be used in more cases with
partitioned tables. That commit neglected to update the estate's
es_result_relation_info when flushing the insert buffer to the partition
making it possible for the index tuples to be added into an index on the
wrong partition.
Fix this and also add an Assert in ExecInsertIndexTuples to help ensure
that we never make this mistake again.
Reported-by: Haruka Takatsuka
Author: Ashutosh Sharma
Discussion: https://postgr.es/m/15832-b1bf336a4ee246b5@postgresql.org
Some of the wrapper functions didn't match the callback names. Many of
them due to staying "consistent" with historic naming of the wrapped
functionality. We decided that for most cases it's more important to
be for tableam to be consistent going forward, than with the past.
The one exception is beginscan/endscan/... because it'd have looked
odd to have systable_beginscan/endscan/... with a different naming
scheme, and changing the systable_* APIs would have caused way too
much churn (including breaking a lot of external users).
Author: Ashwin Agrawal, with some small additions by Andres Freund
Reviewed-By: Andres Freund
Discussion: https://postgr.es/m/CALfoeiugyrXZfX7n0ORCa4L-m834dzmaE8eFdbNR6PMpetU4Ww@mail.gmail.com
When there were duplicate columns in the hash key list, the array
sizes could be miscomputed, resulting in access off the end of the
array. Adjust the computation to ensure the array is always large
enough.
(I considered whether the duplicates could be removed in planning, but
I can't rule out the possibility that duplicate columns might have
different hash functions assigned. Simpler to just make sure it works
at execution time regardless.)
Bug apparently introduced in fc4b3dea2 as part of narrowing down the
tuples stored in the hashtable. Reported by Colm McHugh of Salesforce,
though I didn't use their patch. Backpatch back to version 10 where
the bug was introduced.
Discussion: https://postgr.es/m/CAFeeJoKKu0u+A_A9R9316djW-YW3-+Gtgvy3ju655qRHR3jtdA@mail.gmail.com
This is still using the 2.0 version of pg_bsd_indent.
I thought it would be good to commit this separately,
so as to document the differences between 2.0 and 2.1 behavior.
Discussion: https://postgr.es/m/16296.1558103386@sss.pgh.pa.us
This code was still using the old style of forming a heap tuple rather
than using tuple slots. This would be less efficient if a non-heap
access method was used. And using tuple slots is actually quite a bit
faster when using heap as well.
Also add some test cases for generated columns with null values and
with varlena values. This lack of coverage was discovered while
working on this patch.
Discussion: https://www.postgresql.org/message-id/flat/20190331025744.ugbsyks7czfcoksd%40alap3.anarazel.de
For partial aggregation combine steps,
AggStatePerTrans->numTransInputs was set to the transition function's
number of inputs, rather than the combine function's number of
inputs (always 1).
That lead to partial aggregates with strict combine functions to
wrongly check for NOT NULL input as required by strictness. When the
aggregate wasn't exactly passed one argument, the strictness check was
either omitted (in the 0 args case) or too many arguments were
checked. In the latter case we'd read beyond the end of
FunctionCallInfoData->args (only in master).
AggStatePerTrans->numTransInputs actually has been wrong since since
9.6, where partial aggregates were added. But it turns out to not be
an active problem in 9.6 and 10, because numTransInputs wasn't used at
all for combine functions: Before c253b722f6 there simply was no NULL
check for the input to strict trans functions, and after that the
check was simply hardcoded for the right offset in fcinfo, as it's
done by code specific to combine functions.
In bf6c614a2f (11) the strictness check was generalized, with common
code doing the strictness checks for both plain and combine transition
functions, based on numTransInputs. For combine functions this lead to
not emitting an expression step to check for strict input in the 0
arguments case, and in the > 1 arguments case, we'd check too many
arguments.Due to the fact that the relevant fcinfo->isnull[2..] was
always zero-initialized (more or less by accident, by being part of
the AggStatePerTrans struct, which is palloc0'ed), there was no
observable damage in the latter case before a9c35cf85c, we just
checked too many array elements.
Due to the changes in a9c35cf85c, > 1 argument bug became visible,
because these days fcinfo is a) dynamically allocated without being
zeroed b) exactly the length required for the number of specified
arguments (hardcoded to 2 in this case).
This commit only contains a fairly minimal fix, setting numTransInputs
to a hardcoded 1 when building a pertrans for a combine function. It
seems likely that we'll want to clean this up further (e.g. the
arguments build_pertrans_for_aggref() aren't particularly meaningful
for combine functions). But the wrap date for 12 beta1 is coming up
fast, so it seems good to have a minimal fix in place.
Backpatch to 11. While AggStatePerTrans->numTransInputs was set
wrongly before that, the value was not used for combine functions.
Reported-By: Rajkumar Raghuwanshi
Diagnosed-By: Kyotaro Horiguchi, Jeevan Chalke, Andres Freund, David Rowley
Author: David Rowley, Kyotaro Horiguchi, Andres Freund
Discussion: https://postgr.es/m/CAKcux6=uZEyWyLw0N7HtR9OBc-sWEFeByEZC7t-KDf15FKxVew@mail.gmail.com
Instead add a tableam callback to do so. To avoid adding per
validation overhead, pass a scan to tuple_tid_valid. In heap's case
we'd otherwise incurred a RelationGetNumberOfBlocks() call for each
tid - which'd have added noticable overhead to nodeTidscan.c.
Author: Andres Freund
Reviewed-By: Ashwin Agrawal
Discussion: https://postgr.es/m/20190515185447.gno2jtqxyktylyvs@alap3.anarazel.de
Previously, gen_partprune_steps() always built executor pruning steps
using all suitable clauses, including those containing PARAM_EXEC
Params. This meant that the pruning steps were only completely safe
for executor run-time (scan start) pruning. To prune at executor
startup, we had to ignore the steps involving exec Params. But this
doesn't really work in general, since there may be logic changes
needed as well --- for example, pruning according to the last operator's
btree strategy is the wrong thing if we're not applying that operator.
The rules embodied in gen_partprune_steps() and its minions are
sufficiently complicated that tracking their incremental effects in
other logic seems quite impractical.
Short of a complete redesign, the only safe fix seems to be to run
gen_partprune_steps() twice, once to create executor startup pruning
steps and then again for run-time pruning steps. We can save a few
cycles however by noting during the first scan whether we rejected
any clauses because they involved exec Params --- if not, we don't
need to do the second scan.
In support of this, refactor the internal APIs in partprune.c to make
more use of passing information in the GeneratePruningStepsContext
struct, rather than as separate arguments.
This is, I hope, the last piece of our response to a bug report from
Alan Jackson. Back-patch to v11 where this code came in.
Discussion: https://postgr.es/m/FAD28A83-AC73-489E-A058-2681FA31D648@tvsquared.com
For some reason both callsite and the implementation for heapam had
the meaning inverted (i.e. succeeded == true was passed in case of
conflict). That's confusing.
I (Andres) briefly pondered whether it'd be better to rename
table_complete_speculative's argument to 'bool specConflict' or such,
but decided not to. The 'complete' in the function name for me makes
`succeeded` sound a bit better.
Reported-By: Ashwin Agrawal, Melanie Plageman, Heikki Linnakangas
Discussion:
https://postgr.es/m/CALfoeitk7-TACwYv3hCw45FNPjkA86RfXg4iQ5kAOPhR+F1Y4w@mail.gmail.comhttps://postgr.es/m/97673451-339f-b21e-a781-998d06b1067c@iki.fi
For amcanreorderby scans the nodeIndexscan.c's reorder queue holds
heap tuples, but the underlying table likely does not. Before this fix
we'd return different types of slots, depending on whether the tuple
came from the reorder queue, or from the index + table.
While that could be fixed by signalling that the node doesn't return a
fixed type of slot, it seems better to instead remove the separate
slot for the reorder queue, and use ExecForceStoreHeapTuple() to store
tuples from the queue. It's not particularly common to need
reordering, after all.
This reverts most of the iss_ReorderQueueSlot related changes to
nodeIndexscan.c made in 1a0586de36, except that now
ExecForceStoreHeapTuple() is used instead of ExecStoreHeapTuple().
Noticed when testing zheap against the in-core version of tableam.
Author: Andres Freund
As reported by Tom, when ExecStoreMinimalTuple() had to perform a
conversion to store the minimal tuple in the slot, it forgot to
respect the shouldFree flag, and leaked the tuple into the current
memory context if true. Fix that by freeing the tuple in that case.
Looking at the relevant code made me (Andres) realize that not having
the shouldFree parameter to ExecForceStoreHeapTuple() was a bad
idea. Some callers had to locally implement the necessary logic, and
in one case it was missing, creating a potential per-group leak in
non-hashed aggregation.
The choice to not free the tuple in ExecComputeStoredGenerated() is
not pretty, but not introduced by this commit - I'll start a separate
discussion about it.
Reported-By: Tom Lane
Discussion: https://postgr.es/m/366.1555382816@sss.pgh.pa.us
HoldPinnedPortals() did things in the wrong order: it must not mark
a portal autoHeld until it's been successfully held. Otherwise,
a failure while persisting the portal results in a server crash
because we think the portal is in a good state when it's not.
Also add a check that portal->status is READY before attempting to
hold a pinned portal. We have such a check before the only other
use of HoldPortal(), so it seems unwise not to check it here.
Lastly, rethink the responsibility for where to call HoldPinnedPortals.
The comment for it imagined that it was optional for any individual PL
to call it or not, but that cannot be the case: if some outer level of
procedure has a pinned portal, failing to persist it when an inner
procedure commits is going to be trouble. Let's have SPI do it instead
of the individual PLs. That's not a complete solution, since in theory
a PL might not be using SPI to perform commit/rollback, but such a PL
is going to have to be aware of lots of related requirements anyway.
(This change doesn't cause an API break for any external PLs that might
be calling HoldPinnedPortals per the old regime, because calling it
twice during a commit or rollback sequence won't hurt.)
Per bug #15703 from Julian Schauder. Back-patch to v11 where this code
came in.
Discussion: https://postgr.es/m/15703-c12c5bc0ea34ba26@postgresql.org
Commit 3f2393edef changed ExecCleanupTupleRouting() so that it skipped
cleaning up subplan resultrels before calling EndForeignInsert(), but
that would cause an issue: when those resultrels were foreign tables,
the FDWs would fail to shut down. Repair by skipping it after calling
EndForeignInsert() as before.
Author: Etsuro Fujita
Reviewed-by: David Rowley and Amit Langote
Discussion: https://postgr.es/m/5CAF3B8F.2090905@lab.ntt.co.jp
This commit fixes three, unfortunately related, issues:
1) Since 5db6df0c01, the introduction of DML via tableam, it was
possible to trigger "ERROR: unexpected table_lock_tuple status: 1"
when updating a row that was previously updated in the same
transaction - but only when the previously updated row was before
updated in a concurrent transaction (and READ COMMITTED was
used). The reason for that was that that case simply wasn't
expected. Fixing that lead to:
2) Even before the above commit, there were error checks (introduced
in 6868ed7491) preventing a row being updated by different
commands within the same statement (say in a function called by an
UPDATE) - but that check wasn't performed when the row was first
updated in a concurrent transaction - instead the second update was
silently skipped in that case. After this change we throw the same
error as we'd without the concurrent transaction.
3) The error messages (introduced in 6868ed7491) preventing such
updates emitted the same error message for both DELETE and
UPDATE ("tuple to be updated was already modified by an operation
triggered by the current command"). While that could be changed
separately, it made it hard to write tests that verify the correct
correct behavior of the code.
This commit changes heap's implementation of table_lock_tuple() to
return TM_SelfModified instead of TM_Invisible (previously loosely
modeled after EvalPlanQualFetch), and teaches nodeModifyTable.c to
handle that in response to table_lock_tuple() and not just in response
to table_(delete|update).
Additionally it fixes the wrong error message (see 3 above). The
comment for table_lock_tuple() is also adjusted to state that
TM_Deleted won't return information in TM_FailureData - it'll not
always be available.
This also adds tests to ensure that DELETE/UPDATE correctly error out
when affecting a row that concurrently was modified by another
transaction.
Author: Andres Freund
Reported-By: Tom Lane, when investigating a bug bug fix to another bug
by Amit Langote
Discussion: https://postgr.es/m/19321.1554567786@sss.pgh.pa.us
If we need ordered output from a scan of a partitioned table, but
the ordering matches the partition ordering, then we don't need to
use a MergeAppend to combine the pre-ordered per-partition scan
results: a plain Append will produce the same results. This
both saves useless comparison work inside the MergeAppend proper,
and allows us to start returning tuples after istarting up just
the first child node not all of them.
However, all is not peaches and cream, because if some of the
child nodes have high startup costs then there will be big
discontinuities in the tuples-returned-versus-elapsed-time curve.
The planner's cost model cannot handle that (yet, anyway).
If we model the Append's startup cost as being just the first
child's startup cost, we may drastically underestimate the cost
of fetching slightly more tuples than are available from the first
child. Since we've had bad experiences with over-optimistic choices
of "fast start" plans for ORDER BY LIMIT queries, that seems scary.
As a klugy workaround, set the startup cost estimate for an ordered
Append to be the sum of its children's startup costs (as MergeAppend
would). This doesn't really describe reality, but it's less likely
to cause a bad plan choice than an underestimated startup cost would.
In practice, the cases where we really care about this optimization
will have child plans that are IndexScans with zero startup cost,
so that the overly conservative estimate is still just zero.
David Rowley, reviewed by Julien Rouhaud and Antonin Houska
Discussion: https://postgr.es/m/CAKJS1f-hAqhPLRk_RaSFTgYxd=Tz5hA7kQ2h4-DhJufQk8TGuw@mail.gmail.com
This adds table_multi_insert(), and converts COPY FROM, the only user
of heap_multi_insert, to it.
A simple conversion of COPY FROM use slots would have yielded a
slowdown when inserting into a partitioned table for some
workloads. Different partitions might need different slots (both slot
types and their descriptors), and dropping / creating slots when
there's constant partition changes is measurable.
Thus instead revamp the COPY FROM buffering for partitioned tables to
allow to buffer inserts into multiple tables, flushing only when
limits are reached across all partition buffers. By only dropping
slots when there've been inserts into too many different partitions,
the aforementioned overhead is gone. By allowing larger batches, even
when there are frequent partition changes, we actuall speed such cases
up significantly.
By using slots COPY of very narrow rows into unlogged / temporary
might slow down very slightly (due to the indirect function calls).
Author: David Rowley, Andres Freund, Haribabu Kommi
Discussion:
https://postgr.es/m/20180703070645.wchpu5muyto5n647@alap3.anarazel.dehttps://postgr.es/m/20190327054923.t3epfuewxfqdt22e@alap3.anarazel.de
The assertions added by commit b04aeb0a0 exposed that there are some
code paths wherein the executor will try to open an index without
holding any lock on it. We do have some lock on the index's table,
so it seems likely that there's no fatal problem with this (for
instance, the index couldn't get dropped from under us). Still,
it's bad practice and we should fix it.
To do so, remove the optimizations in ExecInitIndexScan and friends
that tried to avoid taking a lock on an index belonging to a target
relation, and just take the lock always. In non-bug cases, this
will result in no additional shared-memory access, since we'll find
in the local lock table that we already have a lock of the desired
type; hence, no significant performance degradation should occur.
Also, adjust the planner and executor so that the type of lock taken
on an index is always identical to the type of lock taken for its table,
by relying on the recently added RangeTblEntry.rellockmode field.
This avoids some corner cases where that might not have been true
before (possibly resulting in extra locking overhead), and prevents
future maintenance issues from having multiple bits of logic that
all needed to be in sync. In addition, this change removes all core
calls to ExecRelationIsTargetRelation, which avoids a possible O(N^2)
startup penalty for queries with large numbers of target relations.
(We'd probably remove that function altogether, were it not that we
advertise it as something that FDWs might want to use.)
Also adjust some places in selfuncs.c to not take any lock on indexes
they are transiently opening, since we can assume that plancat.c
did that already.
In passing, change gin_clean_pending_list() to take RowExclusiveLock
not AccessShareLock on its target index. Although it's not clear that
that's actually a bug, it seemed very strange for a function that's
explicitly going to modify the index to use only AccessShareLock.
David Rowley, reviewed by Julien Rouhaud and Amit Langote,
a bit of further tweaking by me
Discussion: https://postgr.es/m/19465.1541636036@sss.pgh.pa.us
This moves bitmap heap scan support to below an optional tableam
callback. It's optional as the whole concept of bitmap heapscans is
fairly block specific.
This basically moves the work previously done in bitgetpage() into the
new scan_bitmap_next_block callback, and the direct poking into the
buffer done in BitmapHeapNext() into the new scan_bitmap_next_tuple()
callback.
The abstraction is currently somewhat leaky because
nodeBitmapHeapscan.c's prefetching and visibilitymap based logic
remains - it's likely that we'll later have to move more into the
AM. But it's not trivial to do so without introducing a significant
amount of code duplication between the AMs, so that's a project for
later.
Note that now nodeBitmapHeapscan.c and the associated node types are a
bit misnamed. But it's not clear whether renaming wouldn't be a cure
worse than the disease. Either way, that'd be best done in a separate
commit.
Author: Andres Freund
Reviewed-By: Robert Haas (in an older version)
Discussion: https://postgr.es/m/20180703070645.wchpu5muyto5n647@alap3.anarazel.de
This moves sample scan support to below tableam. It's not optional as
there is, in contrast to e.g. bitmap heap scans, no alternative way to
perform tablesample queries. If an AM can't deal with the block based
API, it will have to throw an ERROR.
The tableam callbacks for this are block based, but given the current
TsmRoutine interface, that seems to be required.
The new interface doesn't require TsmRoutines to perform visibility
checks anymore - that requires the TsmRoutine to know details about
the AM, which we want to avoid. To continue to allow taking the
returned number of tuples account SampleScanState now has a donetuples
field (which previously e.g. existed in SystemRowsSamplerData), which
is only incremented after the visibility check succeeds.
Author: Andres Freund
Discussion: https://postgr.es/m/20180703070645.wchpu5muyto5n647@alap3.anarazel.de
Previously, the planner created RangeTblEntry and RelOptInfo structs
for every partition of a partitioned table, even though many of them
might later be deemed uninteresting thanks to partition pruning logic.
This incurred significant overhead when there are many partitions.
Arrange to postpone creation of these data structures until after
we've processed the query enough to identify restriction quals for
the partitioned table, and then apply partition pruning before not
after creation of each partition's data structures. In this way
we need not open the partition relations at all for partitions that
the planner has no real interest in.
For queries that can be proven at plan time to access only a small
number of partitions, this patch improves the practical maximum
number of partitions from under 100 to perhaps a few thousand.
Amit Langote, reviewed at various times by Dilip Kumar, Jesper Pedersen,
Yoshikazu Imai, and David Rowley
Discussion: https://postgr.es/m/9d7c5112-cb99-6a47-d3be-cf1ee6862a1d@lab.ntt.co.jp
This is an SQL-standard feature that allows creating columns that are
computed from expressions rather than assigned, similar to a view or
materialized view but on a column basis.
This implements one kind of generated column: stored (computed on
write). Another kind, virtual (computed on read), is planned for the
future, and some room is left for it.
Reviewed-by: Michael Paquier <michael@paquier.xyz>
Reviewed-by: Pavel Stehule <pavel.stehule@gmail.com>
Discussion: https://www.postgresql.org/message-id/flat/b151f851-4019-bdb1-699e-ebab07d2f40a@2ndquadrant.com
This primarily is to allow WHERE CURRENT OF to continue to work as it
currently does. It's not clear to me that these semantics make sense
for every AM, but it works for the in-core heap, and the out of core
zheap. We can refine it further at a later point if necessary.
Author: Andres Freund
Discussion: https://postgr.es/m/20180703070645.wchpu5muyto5n647@alap3.anarazel.de
If there's only one child relation, the Append or MergeAppend isn't
doing anything useful, and can be elided. It does have a purpose
during planning though, which is to serve as a buffer between parent
and child Var numbering. Therefore we keep it all the way through
to setrefs.c, and get rid of it only after fixing references in the
plan level(s) above it. This works largely the same as setrefs.c's
ancient hack to get rid of no-op SubqueryScan nodes, and can even
share some code with that.
Note the change to make setrefs.c use apply_tlist_labeling rather than
ad-hoc code. This has the effect of propagating the child's resjunk
and ressortgroupref labels, which formerly weren't propagated when
removing a SubqueryScan. Doing that is demonstrably necessary for
the [Merge]Append cases, and seems harmless for SubqueryScan, if only
because trivial_subqueryscan is afraid to collapse cases where the
resjunk marking differs. (I suspect that restriction could now be
removed, though it's unclear that it'd make any new matches possible,
since the outer query can't have references to a child resjunk column.)
David Rowley, reviewed by Alvaro Herrera and Tomas Vondra
Discussion: https://postgr.es/m/CAKJS1f_7u8ATyJ1JGTMHFoKDvZdeF-iEBhs+sM_SXowOr9cArg@mail.gmail.com
This is essentially the tableam version of heapam_fetch(),
i.e. fetching a tuple identified by a tid, performing visibility
checks.
Note that this different from table_index_fetch_tuple(), which is for
index lookups. It therefore has to handle a tid pointing to an earlier
version of a tuple if the AM uses an optimization like heap's HOT. Add
comments to that end.
This commit removes the stats_relation argument from heap_fetch, as
it's been unused for a long time.
Author: Andres Freund
Reviewed-By: Haribabu Kommi
Discussion: https://postgr.es/m/20180703070645.wchpu5muyto5n647@alap3.anarazel.de
Add command variants COMMIT AND CHAIN and ROLLBACK AND CHAIN, which
start new transactions with the same transaction characteristics as the
just finished one, per SQL standard.
Support for transaction chaining in PL/pgSQL is also added. This
functionality is especially useful when running COMMIT in a loop in
PL/pgSQL.
Reviewed-by: Fabien COELHO <coelho@cri.ensmp.fr>
Discussion: https://www.postgresql.org/message-id/flat/28536681-324b-10dc-ade8-ab46f7645a5a@2ndquadrant.com
This adds new, required, table AM callbacks for insert/delete/update
and lock_tuple. To be able to reasonably use those, the EvalPlanQual
mechanism had to be adapted, moving more logic into the AM.
Previously both delete/update/lock call-sites and the EPQ mechanism had
to have awareness of the specific tuple format to be able to fetch the
latest version of a tuple. Obviously that needs to be abstracted
away. To do so, move the logic that find the latest row version into
the AM. lock_tuple has a new flag argument,
TUPLE_LOCK_FLAG_FIND_LAST_VERSION, that forces it to lock the last
version, rather than the current one. It'd have been possible to do
so via a separate callback as well, but finding the last version
usually also necessitates locking the newest version, making it
sensible to combine the two. This replaces the previous use of
EvalPlanQualFetch(). Additionally HeapTupleUpdated, which previously
signaled either a concurrent update or delete, is now split into two,
to avoid callers needing AM specific knowledge to differentiate.
The move of finding the latest row version into tuple_lock means that
encountering a row concurrently moved into another partition will now
raise an error about "tuple to be locked" rather than "tuple to be
updated/deleted" - which is accurate, as that always happens when
locking rows. While possible slightly less helpful for users, it seems
like an acceptable trade-off.
As part of this commit HTSU_Result has been renamed to TM_Result, and
its members been expanded to differentiated between updating and
deleting. HeapUpdateFailureData has been renamed to TM_FailureData.
The interface to speculative insertion is changed so nodeModifyTable.c
does not have to set the speculative token itself anymore. Instead
there's a version of tuple_insert, tuple_insert_speculative, that
performs the speculative insertion (without requiring a flag to signal
that fact), and the speculative insertion is either made permanent
with table_complete_speculative(succeeded = true) or aborted with
succeeded = false).
Note that multi_insert is not yet routed through tableam, nor is
COPY. Changing multi_insert requires changes to copy.c that are large
enough to better be done separately.
Similarly, although simpler, CREATE TABLE AS and CREATE MATERIALIZED
VIEW are also only going to be adjusted in a later commit.
Author: Andres Freund and Haribabu Kommi
Discussion:
https://postgr.es/m/20180703070645.wchpu5muyto5n647@alap3.anarazel.dehttps://postgr.es/m/20190313003903.nwvrxi7rw3ywhdel@alap3.anarazel.dehttps://postgr.es/m/20160812231527.GA690404@alvherre.pgsql
This adds a flag "deterministic" to collations. If that is false,
such a collation disables various optimizations that assume that
strings are equal only if they are byte-wise equal. That then allows
use cases such as case-insensitive or accent-insensitive comparisons
or handling of strings with different Unicode normal forms.
This functionality is only supported with the ICU provider. At least
glibc doesn't appear to have any locales that work in a
nondeterministic way, so it's not worth supporting this for the libc
provider.
The term "deterministic comparison" in this context is from Unicode
Technical Standard #10
(https://unicode.org/reports/tr10/#Deterministic_Comparison).
This patch makes changes in three areas:
- CREATE COLLATION DDL changes and system catalog changes to support
this new flag.
- Many executor nodes and auxiliary code are extended to track
collations. Previously, this code would just throw away collation
information, because the eventually-called user-defined functions
didn't use it since they only cared about equality, which didn't
need collation information.
- String data type functions that do equality comparisons and hashing
are changed to take the (non-)deterministic flag into account. For
comparison, this just means skipping various shortcuts and tie
breakers that use byte-wise comparison. For hashing, we first need
to convert the input string to a canonical "sort key" using the ICU
analogue of strxfrm().
Reviewed-by: Daniel Verite <daniel@manitou-mail.org>
Reviewed-by: Peter Geoghegan <pg@bowt.ie>
Discussion: https://www.postgresql.org/message-id/flat/1ccc668f-4cbc-0bef-af67-450b47cdfee7@2ndquadrant.com
Previously, the SERIALIZABLE isolation level prevented parallel query
from being used. Allow the two features to be used together by
sharing the leader's SERIALIZABLEXACT with parallel workers.
An extra per-SERIALIZABLEXACT LWLock is introduced to make it safe to
share, and new logic is introduced to coordinate the early release
of the SERIALIZABLEXACT required for the SXACT_FLAG_RO_SAFE
optimization, as follows:
The first backend to observe the SXACT_FLAG_RO_SAFE flag (set by
some other transaction) will 'partially release' the SERIALIZABLEXACT,
meaning that the conflicts and locks it holds are released, but the
SERIALIZABLEXACT itself will remain active because other backends
might still have a pointer to it.
Whenever any backend notices the SXACT_FLAG_RO_SAFE flag, it clears
its own MySerializableXact variable and frees local resources so that
it can skip SSI checks for the rest of the transaction. In the
special case of the leader process, it transfers the SERIALIZABLEXACT
to a new variable SavedSerializableXact, so that it can be completely
released at the end of the transaction after all workers have exited.
Remove the serializable_okay flag added to CreateParallelContext() by
commit 9da0cc35, because it's now redundant.
Author: Thomas Munro
Reviewed-by: Haribabu Kommi, Robert Haas, Masahiko Sawada, Kevin Grittner
Discussion: https://postgr.es/m/CAEepm=0gXGYhtrVDWOTHS8SQQy_=S9xo+8oCxGLWZAOoeJ=yzQ@mail.gmail.com
Too allow table accesses to be not directly dependent on heap, several
new abstractions are needed. Specifically:
1) Heap scans need to be generalized into table scans. Do this by
introducing TableScanDesc, which will be the "base class" for
individual AMs. This contains the AM independent fields from
HeapScanDesc.
The previous heap_{beginscan,rescan,endscan} et al. have been
replaced with a table_ version.
There's no direct replacement for heap_getnext(), as that returned
a HeapTuple, which is undesirable for a other AMs. Instead there's
table_scan_getnextslot(). But note that heap_getnext() lives on,
it's still used widely to access catalog tables.
This is achieved by new scan_begin, scan_end, scan_rescan,
scan_getnextslot callbacks.
2) The portion of parallel scans that's shared between backends need
to be able to do so without the user doing per-AM work. To achieve
that new parallelscan_{estimate, initialize, reinitialize}
callbacks are introduced, which operate on a new
ParallelTableScanDesc, which again can be subclassed by AMs.
As it is likely that several AMs are going to be block oriented,
block oriented callbacks that can be shared between such AMs are
provided and used by heap. table_block_parallelscan_{estimate,
intiialize, reinitialize} as callbacks, and
table_block_parallelscan_{nextpage, init} for use in AMs. These
operate on a ParallelBlockTableScanDesc.
3) Index scans need to be able to access tables to return a tuple, and
there needs to be state across individual accesses to the heap to
store state like buffers. That's now handled by introducing a
sort-of-scan IndexFetchTable, which again is intended to be
subclassed by individual AMs (for heap IndexFetchHeap).
The relevant callbacks for an AM are index_fetch_{end, begin,
reset} to create the necessary state, and index_fetch_tuple to
retrieve an indexed tuple. Note that index_fetch_tuple
implementations need to be smarter than just blindly fetching the
tuples for AMs that have optimizations similar to heap's HOT - the
currently alive tuple in the update chain needs to be fetched if
appropriate.
Similar to table_scan_getnextslot(), it's undesirable to continue
to return HeapTuples. Thus index_fetch_heap (might want to rename
that later) now accepts a slot as an argument. Core code doesn't
have a lot of call sites performing index scans without going
through the systable_* API (in contrast to loads of heap_getnext
calls and working directly with HeapTuples).
Index scans now store the result of a search in
IndexScanDesc->xs_heaptid, rather than xs_ctup->t_self. As the
target is not generally a HeapTuple anymore that seems cleaner.
To be able to sensible adapt code to use the above, two further
callbacks have been introduced:
a) slot_callbacks returns a TupleTableSlotOps* suitable for creating
slots capable of holding a tuple of the AMs
type. table_slot_callbacks() and table_slot_create() are based
upon that, but have additional logic to deal with views, foreign
tables, etc.
While this change could have been done separately, nearly all the
call sites that needed to be adapted for the rest of this commit
also would have been needed to be adapted for
table_slot_callbacks(), making separation not worthwhile.
b) tuple_satisfies_snapshot checks whether the tuple in a slot is
currently visible according to a snapshot. That's required as a few
places now don't have a buffer + HeapTuple around, but a
slot (which in heap's case internally has that information).
Additionally a few infrastructure changes were needed:
I) SysScanDesc, as used by systable_{beginscan, getnext} et al. now
internally uses a slot to keep track of tuples. While
systable_getnext() still returns HeapTuples, and will so for the
foreseeable future, the index API (see 1) above) now only deals with
slots.
The remainder, and largest part, of this commit is then adjusting all
scans in postgres to use the new APIs.
Author: Andres Freund, Haribabu Kommi, Alvaro Herrera
Discussion:
https://postgr.es/m/20180703070645.wchpu5muyto5n647@alap3.anarazel.dehttps://postgr.es/m/20160812231527.GA690404@alvherre.pgsql
... as well as its implementation from backend/access/hash/hashfunc.c to
backend/utils/hash/hashfn.c.
access/hash is the place for the hash index AM, not really appropriate
for generic facilities, which is what hash_any is; having things the old
way meant that anything using hash_any had to include the AM's include
file, pointlessly polluting its namespace with unrelated, unnecessary
cruft.
Also move the HTEqual strategy number to access/stratnum.h from
access/hash.h.
To avoid breaking third-party extension code, add an #include
"utils/hashutils.h" to access/hash.h. (An easily removed line by
committers who enjoy their asbestos suits to protect them from angry
extension authors.)
Discussion: https://postgr.es/m/201901251935.ser5e4h6djt2@alvherre.pgsql
We still require AccessExclusiveLock on the partition itself, because
otherwise an insert that violates the newly-imposed partition
constraint could be in progress at the same time that we're changing
that constraint; only the lock level on the parent relation is
weakened.
To make this safe, we have to cope with (at least) three separate
problems. First, relevant DDL might commit while we're in the process
of building a PartitionDesc. If so, find_inheritance_children() might
see a new partition while the RELOID system cache still has the old
partition bound cached, and even before invalidation messages have
been queued. To fix that, if we see that the pg_class tuple seems to
be missing or to have a null relpartbound, refetch the value directly
from the table. We can't get the wrong value, because DETACH PARTITION
still requires AccessExclusiveLock throughout; if we ever want to
change that, this will need more thought. In testing, I found it quite
difficult to hit even the null-relpartbound case; the race condition
is extremely tight, but the theoretical risk is there.
Second, successive calls to RelationGetPartitionDesc might not return
the same answer. The query planner will get confused if lookup up the
PartitionDesc for a particular relation does not return a consistent
answer for the entire duration of query planning. Likewise, query
execution will get confused if the same relation seems to have a
different PartitionDesc at different times. Invent a new
PartitionDirectory concept and use it to ensure consistency. This
ensures that a single invocation of either the planner or the executor
sees the same view of the PartitionDesc from beginning to end, but it
does not guarantee that the planner and the executor see the same
view. Since this allows pointers to old PartitionDesc entries to
survive even after a relcache rebuild, also postpone removing the old
PartitionDesc entry until we're certain no one is using it.
For the most part, it seems to be OK for the planner and executor to
have different views of the PartitionDesc, because the executor will
just ignore any concurrently added partitions which were unknown at
plan time; those partitions won't be part of the inheritance
expansion, but invalidation messages will trigger replanning at some
point. Normally, this happens by the time the very next command is
executed, but if the next command acquires no locks and executes a
prepared query, it can manage not to notice until a new transaction is
started. We might want to tighten that up, but it's material for a
separate patch. There would still be a small window where a query
that started just after an ATTACH PARTITION command committed might
fail to notice its results -- but only if the command starts before
the commit has been acknowledged to the user. All in all, the warts
here around serializability seem small enough to be worth accepting
for the considerable advantage of being able to add partitions without
a full table lock.
Although in general the consequences of new partitions showing up
between planning and execution are limited to the query not noticing
the new partitions, run-time partition pruning will get confused in
that case, so that's the third problem that this patch fixes.
Run-time partition pruning assumes that indexes into the PartitionDesc
are stable between planning and execution. So, add code so that if
new partitions are added between plan time and execution time, the
indexes stored in the subplan_map[] and subpart_map[] arrays within
the plan's PartitionedRelPruneInfo get adjusted accordingly. There
does not seem to be a simple way to generalize this scheme to cope
with partitions that are removed, mostly because they could then get
added back again with different bounds, but it works OK for added
partitions.
This code does not try to ensure that every backend participating in
a parallel query sees the same view of the PartitionDesc. That
currently doesn't matter, because we never pass PartitionDesc
indexes between backends. Each backend will ignore the concurrently
added partitions which it notices, and it doesn't matter if different
backends are ignoring different sets of concurrently added partitions.
If in the future that matters, for example because we allow writes in
parallel query and want all participants to do tuple routing to the same
set of partitions, the PartitionDirectory concept could be improved to
share PartitionDescs across backends. There is a draft patch to
serialize and restore PartitionDescs on the thread where this patch
was discussed, which may be a useful place to start.
Patch by me. Thanks to Alvaro Herrera, David Rowley, Simon Riggs,
Amit Langote, and Michael Paquier for discussion, and to Alvaro
Herrera for some review.
Discussion: http://postgr.es/m/CA+Tgmobt2upbSocvvDej3yzokd7AkiT+PvgFH+a9-5VV1oJNSQ@mail.gmail.com
Discussion: http://postgr.es/m/CA+TgmoZE0r9-cyA-aY6f8WFEROaDLLL7Vf81kZ8MtFCkxpeQSw@mail.gmail.com
Discussion: http://postgr.es/m/CA+TgmoY13KQZF-=HNTrt9UYWYx3_oYOQpu9ioNT49jGgiDpUEA@mail.gmail.com
Until now the the slot to store the conflicting tuple, and the result
of the ON CONFLICT SET, where reused between partitions. That
necessitated changing slots descriptor when switching partitions.
Besides the overhead of switching descriptors on a slot (which
requires memory allocations and prevents JITing), that's importantly
also problematic for tableam. There individual partitions might belong
to different tableams, needing different kinds of slots.
In passing also fix ExecOnConflictUpdate to clear the existing slot at
exit. Otherwise that slot could continue to hold a pin till the query
ends, which could be far too long if the input data set is large, and
there's no further conflicts. While previously also problematic, it's
now more important as there will be more such slots when partitioned.
Author: Andres Freund
Reviewed-By: Robert Haas, David Rowley
Discussion: https://postgr.es/m/20180703070645.wchpu5muyto5n647@alap3.anarazel.de
I broke/typoed this in 4da597edf1. Astonishingly this mostly
doesn't cause breakage, except when trying to change the tuple
descriptor of a slot (because TTS_FLAG_FIXED is assumed to be set).
Author: Andres Freund
StoreIndexTuple was a loop over index_getattr, which is O(N^2)
if the index columns are variable-width, and the performance
impact is already quite visible at ten columns. The obvious
move is to replace that with a call to index_deform_tuple ...
but that's *also* a loop over index_getattr. Improve it to
be essentially a clone of heap_deform_tuple.
(There are a few other places that loop over all index columns
with index_getattr, and perhaps should be changed likewise,
but most of them don't seem performance-critical. Anyway, the
rest would mostly only be interested in the index key columns,
which there aren't likely to be so many of. Wide index tuples
are a new thing with INCLUDE.)
Konstantin Knizhnik
Discussion: https://postgr.es/m/e06b2d27-04fc-5c0e-bb8c-ecd72aa24959@postgrespro.ru
For the upcoming pluggable table access methods it's quite
inconvenient to store tuples as HeapTuples, as that'd require
converting tuples from a their native format into HeapTuples. Instead
use slots to manage epq tuples.
To fit into that scheme, change the foreign data wrapper callback
RefetchForeignRow, to store the tuple in a slot. Insist on using the
caller provided slot, so it conveniently can be stored in the
corresponding EPQ slot. As there is no in core user of
RefetchForeignRow, that change was done blindly, but we plan to test
that soon.
To avoid duplicating that work for row locks, move row locks to just
directly use the EPQ slots - it previously temporarily stored tuples
in LockRowsState.lr_curtuples, but that doesn't seem beneficial, given
we'd possibly end up with a significant number of additional slots.
The behaviour of es_epqTupleSet[rti -1] is now checked by
es_epqTupleSlot[rti -1] != NULL, as that is distinguishable from a
slot containing an empty tuple.
Author: Andres Freund, Haribabu Kommi, Ashutosh Bapat
Discussion: https://postgr.es/m/20180703070645.wchpu5muyto5n647@alap3.anarazel.de
After 5408e233f0 it's not necessary to force materializing the
target slot when copying from one buffer slot to another. Previously
that was required because the HeapTupleData portion of the source slot
wasn't guaranteed to stay valid long enough, but now we can simply
copy that part into the destination slot's tupdata.
Author: Andres Freund
We have forboth() and forthree() macros that simplify iterating
through several parallel lists, but not everyplace that could
reasonably use those was doing so. Also invent forfour() and
forfive() macros to do the same for four or five parallel lists,
and use those where applicable.
The immediate motivation for doing this is to reduce the number
of ad-hoc lnext() calls, to reduce the footprint of a WIP patch.
However, it seems like good cleanup and error-proofing anyway;
the places that were combining forthree() with a manually iterated
loop seem particularly illegible and bug-prone.
There was some speculation about restructuring related parsetree
representations to reduce the need for parallel list chasing of
this sort. Perhaps that's a win, or perhaps not, but in any case
it would be considerably more invasive than this patch; and it's
not particularly related to my immediate goal of improving the
List infrastructure. So I'll leave that question for another day.
Patch by me; thanks to David Rowley for review.
Discussion: https://postgr.es/m/11587.1550975080@sss.pgh.pa.us
In preparation for abstracting table storage, convert trigger.c to
track tuples in slots. Which also happens to make code calling
triggers simpler.
As the calling interface for triggers themselves is not changed in
this patch, HeapTuples still are extracted from the slot at that
time. But that's handled solely inside trigger.c, not visible to
callers. It's quite likely that we'll want to revise the external
trigger interface, but that's a separate large project.
As part of this work the slots used for old/new/return tuples are
moved from EState into ResultRelInfo, as different updated tables
might need different slots. The slots are now also now created
on-demand, which is good both from an efficiency POV, but also makes
the modifying code simpler.
Author: Andres Freund, Amit Khandekar and Ashutosh Bapat
Discussion: https://postgr.es/m/20180703070645.wchpu5muyto5n647@alap3.anarazel.de
After the introduction of tuple table slots all table AMs need to
support returning the table oid of the tuple stored in a slot created
by said AM. It does not make sense to re-implement that in every AM,
therefore move handling of table OIDs into the TupleTableSlot
structure itself. It's possible that we, at a later date, might want
to get rid of HeapTupleData.t_tableOid entirely, but doing so before
the abstractions for table AMs are integrated turns out to be too
hard, so delay that for now.
Similarly, every AM needs to support the concept of a tuple
identifier (tid / item pointer) for its tuples. It's quite possible
that we'll generalize the exact form of a tid at a future point (to
allow for things like index organized tables), but for now many parts
of the code know about tids, so there's not much point in abstracting
tids away. Therefore also move into slot (rather than providing API to
set/get the tid associated with the tuple in a slot).
Once table AM includes insert/updating/deleting tuples, the
responsibility to set the correct tid after such an action will move
into that. After that change, code doing such modifications, should
not have to deal with HeapTuples directly anymore.
Author: Andres Freund, Haribabu Kommi and Ashutosh Bapat
Discussion: https://postgr.es/m/20180703070645.wchpu5muyto5n647@alap3.anarazel.de
This allows to avoid an unnecessary pin/unpin cycle when storing a
tuple in an already pinned buffer into a slot, when the pin isn't
further needed at the call site.
Only a single caller for now (to ensure coverage), but upcoming
patches will increase use of the new function.
Author: Andres Freund
Discussion: https://postgr.es/m/20180703070645.wchpu5muyto5n647@alap3.anarazel.de
This is similar in spirit to the existing partbounds.c file in the
same directory, except that there's a lot less code in the new file
created by this commit. Pending work in this area proposes to add a
bunch more code related to PartitionDescs, though, and this will give
us a good place to put it.
Discussion: http://postgr.es/m/CA+TgmoZUwPf_uanjF==gTGBMJrn8uCq52XYvAEorNkLrUdoawg@mail.gmail.com
Instead of locking all partitions to which we might route a tuple at
executor startup, just lock them as we use them. In some cases such a
partition might get locked at executor startup anyway because it
appears in the query's range table for some other reason, but in other
cases this is a bit savings.
This changes the order in which partitions are locked in some cases,
which might conceivably create deadlock hazards that don't exist
today, but per discussion, it seems like such cases should be rare
enough that we can neglect them in favor of improving performance.
David Rowley, reviewed and tested by Tomas Vondra, Sho Kato, John
Naylor, Tom Lane, and me.
Discussion: http://postgr.es/m/CAKJS1f-=FnMqmQP6qitkD+xEddxw22ySLP-0xFk3JAqUX2yfMw@mail.gmail.com
This has the advantage that the comparator expression, the table's
slot, etc do not have to be rebuilt. Additionally the simplehash.h
hashtable within the tuple hashtable now keeps its previous size and
doesn't need to be reallocated. That both reduces allocator overhead,
and improves performance in cases where the input estimation was off
by a significant factor.
To avoid an API/ABI break, the new parameter is exposed via the new
BuildTupleHashTableExt(), and BuildTupleHashTable() now is a wrapper
around the former, that continues to allocate the table itself in the
tablecxt.
Using this fixes performance issues discovered in the two bugs
referenced. This commit however has not converted the callers, that's
done in a separate commit.
Bug: #15592#15486
Reported-By: Jakub Janeček, Dmitry Marakasov
Author: Andres Freund
Discussion:
https://postgr.es/m/15486-05850f065da42931@postgresql.orghttps://postgr.es/m/20190114180423.ywhdg2iagzvh43we@alap3.anarazel.de
Backpatch: 11, this is a prerequisite for other fixes
In bf6c614a2f I added a expr context to evaluate the grouping
expression. Unfortunately the code I added initialized them while in
the calling context, rather the table context. Additionally, I used
CreateExprContext() rather than CreateStandaloneExprContext(), which
creates the econtext in the estate's query context.
Fix that by using CreateStandaloneExprContext when in the table's
tablecxt. As we rely on the memory being freed by a memory context
reset that means that the econtext's shutdown callbacks aren't being
called, but that seems ok as the expressions are tightly controlled
due to ExecBuildGroupingEqual().
Bug: #15592
Reported-By: Dmitry Marakasov
Author: Andres Freund
Discussion: https://postgr.es/m/20190114222838.h6r3fuyxjxkykf6t@alap3.anarazel.de
Backpatch: 11, where I broke this in bf6c614a2f
The old name of this file was never a very good indication of what it
was for. Now that there's also access/relation.h, we have a potential
confusion hazard as well, so let's rename it to something more apropos.
Per discussion, "pathnodes.h" is reasonable, since a good fraction of
the file is Path node definitions.
While at it, tweak a couple of other headers that were gratuitously
importing relation.h into modules that don't need it.
Discussion: https://postgr.es/m/7719.1548688728@sss.pgh.pa.us
Create a new header optimizer/optimizer.h, which exposes just the
planner functions that can be used "at arm's length", without need
to access Paths or the other planner-internal data structures defined
in nodes/relation.h. This is intended to provide the whole planner
API seen by most of the rest of the system; although FDWs still need
to use additional stuff, and more thought is also needed about just
what selfuncs.c should rely on.
The main point of doing this now is to limit the amount of new
#include baggage that will be needed by "planner support functions",
which I expect to introduce later, and which will be in relevant
datatype modules rather than anywhere near the planner.
This commit just moves relevant declarations into optimizer.h from
other header files (a couple of which go away because everything
got moved), and adjusts #include lists to match. There's further
cleanup that could be done if we want to decide that some stuff
being exposed by optimizer.h doesn't belong in the planner at all,
but I'll leave that for another day.
Discussion: https://postgr.es/m/11460.1548706639@sss.pgh.pa.us
Move a few very simple node-creation and node-type-testing functions
from the planner's clauses.c to nodes/makefuncs and nodes/nodeFuncs.
There's nothing planner-specific about them, as evidenced by the
number of other places that were using them.
While at it, rename and_clause() etc to is_andclause() etc, to clarify
that they are node-type-testing functions not node-creation functions.
And use "static inline" implementations for the shortest ones.
Also, modify flatten_join_alias_vars() and some subsidiary functions
to take a Query not a PlannerInfo to define the join structure that
Vars should be translated according to. They were only using the
"parse" field of the PlannerInfo anyway, so this just requires removing
one level of indirection. The advantage is that now parse_agg.c can
use flatten_join_alias_vars() without the horrid kluge of creating an
incomplete PlannerInfo, which will allow that file to be decoupled from
relation.h in a subsequent patch.
Discussion: https://postgr.es/m/11460.1548706639@sss.pgh.pa.us
The fact that "SELECT expression" has no base relations has long been a
thorn in the side of the planner. It makes it hard to flatten a sub-query
that looks like that, or is a trivial VALUES() item, because the planner
generally uses relid sets to identify sub-relations, and such a sub-query
would have an empty relid set if we flattened it. prepjointree.c contains
some baroque logic that works around this in certain special cases --- but
there is a much better answer. We can replace an empty FROM clause with a
dummy RTE that acts like a table of one row and no columns, and then there
are no such corner cases to worry about. Instead we need some logic to
get rid of useless dummy RTEs, but that's simpler and covers more cases
than what was there before.
For really trivial cases, where the query is just "SELECT expression" and
nothing else, there's a hazard that adding the extra RTE makes for a
noticeable slowdown; even though it's not much processing, there's not
that much for the planner to do overall. However testing says that the
penalty is very small, close to the noise level. In more complex queries,
this is able to find optimizations that we could not find before.
The new RTE type is called RTE_RESULT, since the "scan" plan type it
gives rise to is a Result node (the same plan we produced for a "SELECT
expression" query before). To avoid confusion, rename the old ResultPath
path type to GroupResultPath, reflecting that it's only used in degenerate
grouping cases where we know the query produces just one grouped row.
(It wouldn't work to unify the two cases, because there are different
rules about where the associated quals live during query_planner.)
Note: although this touches readfuncs.c, I don't think a catversion
bump is required, because the added case can't occur in stored rules,
only plans.
Patch by me, reviewed by David Rowley and Mark Dilger
Discussion: https://postgr.es/m/15944.1521127664@sss.pgh.pa.us
Before this change FunctionCallInfoData, the struct arguments etc for
V1 function calls are stored in, always had space for
FUNC_MAX_ARGS/100 arguments, storing datums and their nullness in two
arrays. For nearly every function call 100 arguments is far more than
needed, therefore wasting memory. Arg and argnull being two separate
arrays also guarantees that to access a single argument, two
cachelines have to be touched.
Change the layout so there's a single variable-length array with pairs
of value / isnull. That drastically reduces memory consumption for
most function calls (on x86-64 a two argument function now uses
64bytes, previously 936 bytes), and makes it very likely that argument
value and its nullness are on the same cacheline.
Arguments are stored in a new NullableDatum struct, which, due to
padding, needs more memory per argument than before. But as usually
far fewer arguments are stored, and individual arguments are cheaper
to access, that's still a clear win. It's likely that there's other
places where conversion to NullableDatum arrays would make sense,
e.g. TupleTableSlots, but that's for another commit.
Because the function call information is now variable-length
allocations have to take the number of arguments into account. For
heap allocations that can be done with SizeForFunctionCallInfoData(),
for on-stack allocations there's a new LOCAL_FCINFO(name, nargs) macro
that helps to allocate an appropriately sized and aligned variable.
Some places with stack allocation function call information don't know
the number of arguments at compile time, and currently variably sized
stack allocations aren't allowed in postgres. Therefore allow for
FUNC_MAX_ARGS space in these cases. They're not that common, so for
now that seems acceptable.
Because of the need to allocate FunctionCallInfo of the appropriate
size, older extensions may need to update their code. To avoid subtle
breakages, the FunctionCallInfoData struct has been renamed to
FunctionCallInfoBaseData. Most code only references FunctionCallInfo,
so that shouldn't cause much collateral damage.
This change is also a prerequisite for more efficient expression JIT
compilation (by allocating the function call information on the stack,
allowing LLVM to optimize it away); previously the size of the call
information caused problems inside LLVM's optimizer.
Author: Andres Freund
Reviewed-By: Tom Lane
Discussion: https://postgr.es/m/20180605172952.x34m5uz6ju6enaem@alap3.anarazel.de
Given these routines are heap specific, and that there will be more
generic visibility support in via table AM, it makes sense to move the
prototypes to heapam.h (routines like HeapTupleSatisfiesVacuum will
not be exposed in a generic fashion, because they are too storage
specific).
Similarly, the code in tqual.c is specific to heap, so moving it into
access/heap/ makes sense.
Author: Andres Freund
Discussion: https://postgr.es/m/20180703070645.wchpu5muyto5n647@alap3.anarazel.de
The code in tqual.c is largely heap specific. Due to the upcoming
pluggable storage work, it therefore makes sense to move it into
access/heap/ (as the file's header notes, the tqual name isn't very
good).
But the various statically allocated snapshot and snapshot
initialization functions are now (see previous commit) generic and do
not depend on functions declared in tqual.h anymore. Therefore move.
Also move XidInMVCCSnapshot as that's useful for future AMs, and
already used outside of tqual.c.
Author: Andres Freund
Discussion: https://postgr.es/m/20180703070645.wchpu5muyto5n647@alap3.anarazel.de
Most of these had been obsoleted by 568d4138c / the SnapshotNow
removal.
This is is preparation for moving most of tqual.[ch] into either
snapmgr.h or heapam.h, which in turn is in preparation for pluggable
table AMs.
Author: Andres Freund
Discussion: https://postgr.es/m/20180703070645.wchpu5muyto5n647@alap3.anarazel.de
The parent of some WCO expressions was, apparently by accident, set to
the the source of DML queries, rather than the target table. This
causes problems for the upcoming pluggable storage work, because the
target and source table might be of different storage types.
It's possible that this is already problematic, but neither
experimenting nor inquiries on -hackers have found them. So don't
backpatch for now.
Author: Andres Freund
Discussion: https://postgr.es/m/20181205225213.hiwa3kgoxeybqcqv@alap3.anarazel.de
This is the genam.h equivalent of 4c850ecec6 (which removed
heapam.h from a lot of other headers). There's still a few header
includes of genam.h, but not from central headers anymore.
As a few headers are not indirectly included anymore, execnodes.h and
relscan.h need a few additional includes. Some of the depended on
types were replacable by using the underlying structs, but e.g. for
Snapshot in execnodes.h that'd have gotten more invasive than
reasonable in this commit.
Like the aforementioned commit 4c850ecec6, this requires adding new
genam.h includes to a number of backend files, which likely is also
required in a few external projects.
Author: Andres Freund
Discussion: https://postgr.es/m/20190114000701.y4ttcb74jpskkcfb@alap3.anarazel.de
heapam.h previously was included in a number of widely used
headers (e.g. execnodes.h, indirectly in executor.h, ...). That's
problematic on its own, as heapam.h contains a lot of low-level
details that don't need to be exposed that widely, but becomes more
problematic with the upcoming introduction of pluggable table storage
- it seems inappropriate for heapam.h to be included that widely
afterwards.
heapam.h was largely only included in other headers to get the
HeapScanDesc typedef (which was defined in heapam.h, even though
HeapScanDescData is defined in relscan.h). The better solution here
seems to be to just use the underlying struct (forward declared where
necessary). Similar for BulkInsertState.
Another problem was that LockTupleMode was used in executor.h - parts
of the file tried to cope without heapam.h, but due to the fact that
it indirectly included it, several subsequent violations of that goal
were not not noticed. We could just reuse the approach of declaring
parameters as int, but it seems nicer to move LockTupleMode to
lockoptions.h - that's not a perfect location, but also doesn't seem
bad.
As a number of files relied on implicitly included heapam.h, a
significant number of files grew an explicit include. It's quite
probably that a few external projects will need to do the same.
Author: Andres Freund
Reviewed-By: Alvaro Herrera
Discussion: https://postgr.es/m/20190114000701.y4ttcb74jpskkcfb@alap3.anarazel.de
If trying to use something else than a plain table as logical
replication target, a rather-generic error message gets used to report
the problem. This can be confusing when it comes to foreign tables and
partitioned tables, so use more dedicated messages in these cases.
Author: Amit Langote
Reviewed-by: Peter Eisentraut, Magnus Hagander, Michael Paquier
Discussion: https://postgr.es/m/41799bee-40eb-7bb5-80b1-325ce17518bc@lab.ntt.co.jp
Users of the WaitEventSet and WaitLatch() APIs can now choose between
asking for WL_POSTMASTER_DEATH and then handling it explicitly, or asking
for WL_EXIT_ON_PM_DEATH to trigger immediate exit on postmaster death.
This reduces code duplication, since almost all callers want the latter.
Repair all code that was previously ignoring postmaster death completely,
or requesting the event but ignoring it, or requesting the event but then
doing an unconditional PostmasterIsAlive() call every time through its
event loop (which is an expensive syscall on platforms for which we don't
have USE_POSTMASTER_DEATH_SIGNAL support).
Assert that callers of WaitLatchXXX() under the postmaster remember to
ask for either WL_POSTMASTER_DEATH or WL_EXIT_ON_PM_DEATH, to prevent
future bugs.
The only process that doesn't handle postmaster death is syslogger. It
waits until all backends holding the write end of the syslog pipe
(including the postmaster) have closed it by exiting, to be sure to
capture any parting messages. By using the WaitEventSet API directly
it avoids the new assertion, and as a by-product it may be slightly
more efficient on platforms that have epoll().
Author: Thomas Munro
Reviewed-by: Kyotaro Horiguchi, Heikki Linnakangas, Tom Lane
Discussion: https://postgr.es/m/CAEepm%3D1TCviRykkUb69ppWLr_V697rzd1j3eZsRMmbXvETfqbQ%40mail.gmail.com,
https://postgr.es/m/CAEepm=2LqHzizbe7muD7-2yHUbTOoF7Q+qkSD5Q41kuhttRTwA@mail.gmail.com
Previously tables declared WITH OIDS, including a significant fraction
of the catalog tables, stored the oid column not as a normal column,
but as part of the tuple header.
This special column was not shown by default, which was somewhat odd,
as it's often (consider e.g. pg_class.oid) one of the more important
parts of a row. Neither pg_dump nor COPY included the contents of the
oid column by default.
The fact that the oid column was not an ordinary column necessitated a
significant amount of special case code to support oid columns. That
already was painful for the existing, but upcoming work aiming to make
table storage pluggable, would have required expanding and duplicating
that "specialness" significantly.
WITH OIDS has been deprecated since 2005 (commit ff02d0a05280e0).
Remove it.
Removing includes:
- CREATE TABLE and ALTER TABLE syntax for declaring the table to be
WITH OIDS has been removed (WITH (oids[ = true]) will error out)
- pg_dump does not support dumping tables declared WITH OIDS and will
issue a warning when dumping one (and ignore the oid column).
- restoring an pg_dump archive with pg_restore will warn when
restoring a table with oid contents (and ignore the oid column)
- COPY will refuse to load binary dump that includes oids.
- pg_upgrade will error out when encountering tables declared WITH
OIDS, they have to be altered to remove the oid column first.
- Functionality to access the oid of the last inserted row (like
plpgsql's RESULT_OID, spi's SPI_lastoid, ...) has been removed.
The syntax for declaring a table WITHOUT OIDS (or WITH (oids = false)
for CREATE TABLE) is still supported. While that requires a bit of
support code, it seems unnecessary to break applications / dumps that
do not use oids, and are explicit about not using them.
The biggest user of WITH OID columns was postgres' catalog. This
commit changes all 'magic' oid columns to be columns that are normally
declared and stored. To reduce unnecessary query breakage all the
newly added columns are still named 'oid', even if a table's column
naming scheme would indicate 'reloid' or such. This obviously
requires adapting a lot code, mostly replacing oid access via
HeapTupleGetOid() with access to the underlying Form_pg_*->oid column.
The bootstrap process now assigns oids for all oid columns in
genbki.pl that do not have an explicit value (starting at the largest
oid previously used), only oids assigned later by oids will be above
FirstBootstrapObjectId. As the oid column now is a normal column the
special bootstrap syntax for oids has been removed.
Oids are not automatically assigned during insertion anymore, all
backend code explicitly assigns oids with GetNewOidWithIndex(). For
the rare case that insertions into the catalog via SQL are called for
the new pg_nextoid() function can be used (which only works on catalog
tables).
The fact that oid columns on system tables are now normal columns
means that they will be included in the set of columns expanded
by * (i.e. SELECT * FROM pg_class will now include the table's oid,
previously it did not). It'd not technically be hard to hide oid
column by default, but that'd mean confusing behavior would either
have to be carried forward forever, or it'd cause breakage down the
line.
While it's not unlikely that further adjustments are needed, the
scope/invasiveness of the patch makes it worthwhile to get merge this
now. It's painful to maintain externally, too complicated to commit
after the code code freeze, and a dependency of a number of other
patches.
Catversion bump, for obvious reasons.
Author: Andres Freund, with contributions by John Naylor
Discussion: https://postgr.es/m/20180930034810.ywp2c7awz7opzcfr@alap3.anarazel.de
This commit completes the work prepared in 1a0586de36, splitting the
old TupleTableSlot implementation (which could store buffer, heap,
minimal and virtual slots) into four different slot types. As
described in the aforementioned commit, this is done with the goal of
making tuple table slots extensible, to allow for pluggable table
access methods.
To achieve runtime extensibility for TupleTableSlots, operations on
slots that can differ between types of slots are performed using the
TupleTableSlotOps struct provided at slot creation time. That
includes information from the size of TupleTableSlot struct to be
allocated, initialization, deforming etc. See the struct's definition
for more detailed information about callbacks TupleTableSlotOps.
I decided to rename TTSOpsBufferTuple to TTSOpsBufferHeapTuple and
ExecCopySlotTuple to ExecCopySlotHeapTuple, as that seems more
consistent with other naming introduced in recent patches.
There's plenty optimization potential in the slot implementation, but
according to benchmarking the state after this commit has similar
performance characteristics to before this set of changes, which seems
sufficient.
There's a few changes in execReplication.c that currently need to poke
through the slot abstraction, that'll be repaired once the pluggable
storage patchset provides the necessary infrastructure.
Author: Andres Freund and Ashutosh Bapat, with changes by Amit Khandekar
Discussion: https://postgr.es/m/20181105210039.hh4vvi4vwoq5ba2q@alap3.anarazel.de
This speeds up write operations (INSERT, UPDATE, DELETE, COPY, as well
as the future MERGE) on partitioned tables.
This changes the setup for tuple routing so that it does far less work
during the initial setup and pushes more work out to when partitions
receive tuples. PartitionDispatchData structs for sub-partitioned
tables are only created when a tuple gets routed through it. The
possibly large arrays in the PartitionTupleRouting struct have largely
been removed. The partitions[] array remains but now never contains any
NULL gaps. Previously the NULLs had to be skipped during
ExecCleanupTupleRouting(), which could add a large overhead to the
cleanup when the number of partitions was large. The partitions[] array
is allocated small to start with and only enlarged when we route tuples
to enough partitions that it runs out of space. This allows us to keep
simple single-row partition INSERTs running quickly. Redesign
The arrays in PartitionTupleRouting which stored the tuple translation maps
have now been removed. These have been moved out into a
PartitionRoutingInfo struct which is an additional field in ResultRelInfo.
The find_all_inheritors() call still remains by far the slowest part of
ExecSetupPartitionTupleRouting(). This commit just removes the other slow
parts.
In passing also rename the tuple translation maps from being ParentToChild
and ChildToParent to being RootToPartition and PartitionToRoot. The old
names mislead you into thinking that a partition of some sub-partitioned
table would translate to the rowtype of the sub-partitioned table rather
than the root partitioned table.
Authors: David Rowley and Amit Langote, heavily revised by Álvaro Herrera
Testing help from Jesper Pedersen and Kato Sho.
Discussion: https://postgr.es/m/CAKJS1f_1RJyFquuCKRFHTdcXqoPX-PYqAd7nz=GVBwvGh4a6xA@mail.gmail.com
The assumption made in 1a0586de36 was wrong, as evidenced by
buildfarm failure on locust, which runs with
force_parallel_mode=regress. The tuples accessed in either nodes are
in the outer slot, and we can't trivially rely on the slot type being
known because the leader might execute the subsidiary node directly,
or via the tuple queue on a worker. In the latter case the tuple will
always be a heaptuple slot, but in the former, it'll be whatever the
subsidiary node returns.
This is important so JIT compilation knows what kind of tuple slot the
deforming routine can expect. There's also optimization potential for
expression initialization without JIT compilation. It e.g. seems
plausible to elide EEOP_*_FETCHSOME ops entirely when dealing with
virtual slots.
Author: Andres Freund
Discussion: https://postgr.es/m/20181105210039.hh4vvi4vwoq5ba2q@alap3.anarazel.de
Previously this information was computed when JIT compiling an
expression. But the information is useful for assertions in the
non-JIT case too (for assertions), therefore it makes sense to move
it.
This will, in a followup commit, allow to treat different slot types
differently. E.g. for virtual slots there's no need to generate a JIT
function to deform the slot.
Author: Andres Freund
Discussion: https://postgr.es/m/20181105210039.hh4vvi4vwoq5ba2q@alap3.anarazel.de
Upcoming work intends to allow pluggable ways to introduce new ways of
storing table data. Accessing those table access methods from the
executor requires TupleTableSlots to be carry tuples in the native
format of such storage methods; otherwise there'll be a significant
conversion overhead.
Different access methods will require different data to store tuples
efficiently (just like virtual, minimal, heap already require fields
in TupleTableSlot). To allow that without requiring additional pointer
indirections, we want to have different structs (embedding
TupleTableSlot) for different types of slots. Thus different types of
slots are needed, which requires adapting creators of slots.
The slot that most efficiently can represent a type of tuple in an
executor node will often depend on the type of slot a child node
uses. Therefore we need to track the type of slot is returned by
nodes, so parent slots can create slots based on that.
Relatedly, JIT compilation of tuple deforming needs to know which type
of slot a certain expression refers to, so it can create an
appropriate deforming function for the type of tuple in the slot.
But not all nodes will only return one type of slot, e.g. an append
node will potentially return different types of slots for each of its
subplans.
Therefore add function that allows to query the type of a node's
result slot, and whether it'll always be the same type (whether it's
fixed). This can be queried using ExecGetResultSlotOps().
The scan, result, inner, outer type of slots are automatically
inferred from ExecInitScanTupleSlot(), ExecInitResultSlot(),
left/right subtrees respectively. If that's not correct for a node,
that can be overwritten using new fields in PlanState.
This commit does not introduce the actually abstracted implementation
of different kind of TupleTableSlots, that will be left for a followup
commit. The different types of slots introduced will, for now, still
use the same backing implementation.
While this already partially invalidates the big comment in
tuptable.h, it seems to make more sense to update it later, when the
different TupleTableSlot implementations actually exist.
Author: Ashutosh Bapat and Andres Freund, with changes by Amit Khandekar
Discussion: https://postgr.es/m/20181105210039.hh4vvi4vwoq5ba2q@alap3.anarazel.de
Previously materializing a slot always returned a HeapTuple. As
current work aims to reduce the reliance on HeapTuples (so other
storage systems can work efficiently), that needs to change. Thus
split the tasks of materializing a slot (i.e. making it independent
from the underlying storage / other memory contexts) from fetching a
HeapTuple from the slot. For brevity, allow to fetch a HeapTuple from
a slot and materializing the slot at the same time, controlled by a
parameter.
For now some callers of ExecFetchSlotHeapTuple, with materialize =
true, expect that changes to the heap tuple will be reflected in the
underlying slot. Those places will be adapted in due course, so while
not pretty, that's OK for now.
Also rename ExecFetchSlotTuple to ExecFetchSlotHeapTupleDatum and
ExecFetchSlotTupleDatum to ExecFetchSlotHeapTupleDatum, as it's likely
that future storage methods will need similar methods. There already
is ExecFetchSlotMinimalTuple, so the new names make the naming scheme
more coherent.
Author: Ashutosh Bapat and Andres Freund, with changes by Amit Khandekar
Discussion: https://postgr.es/m/20181105210039.hh4vvi4vwoq5ba2q@alap3.anarazel.de
With run-time partition pruning, there is no longer necessarily an
executor node for each corresponding plan node.
Author: David Rowley <david.rowley@2ndquadrant.com>
The current pattern of reseting expressions both in
ExecProcessReturning() and ExecOnConflictUpdate() makes it harder than
necessary to reason about memory lifetimes. It also requires
materializing slots unnecessarily, although this patch doesn't take
advantage of the fact that that's not necessary anymore.
Instead reset the expression context once for each input tuple.
Author: Ashutosh Bapat
Discussion: https://postgr.es/m/20181105210039.hh4vvi4vwoq5ba2q@alap3.anarazel.de
ExecFindInitialMatchingSubPlans has to update the PartitionPruneState's
subplan mapping data to account for the removal of subplans it prunes.
However, that's only necessary if run-time pruning will also occur,
so we can skip it when that won't happen, which should result in not
needing to do the remapping in many cases. (We now need it only when
some partitions are potentially startup-time prunable and others are
potentially run-time prunable, which seems like an unusual case.)
Also make some marginal performance improvements in the remapping
itself. These will mainly win if most partitions got pruned by
the startup-time pruning, which is perhaps a debatable assumption
in this context.
Also fix some bogus comments, and rearrange code to marginally
reduce space consumption in the executor's query-lifespan context.
David Rowley, reviewed by Yoshikazu Imai
Discussion: https://postgr.es/m/CAKJS1f9+m6-di-zyy4B4AGn0y1B9F8UKDRigtBbNviXOkuyOpw@mail.gmail.com
Before this commit slot_getattr() checked for dropped
columns (returning NULL in that case), but only after checking for
previously deformed columns. As slot_deform_tuple() does not contain
such a check, the check in slot_getattr() would often not have been
reached, depending on previous use of the slot.
These days locking and plan invalidation ought to ensure that dropped
columns are not accessed in query plans. Therefore this commit just
drops the insufficient check in slot_getattr(). It's possible that
we'll find some holes againt use of dropped columns, but if so, those
need to be addressed independent of slot_getattr(), as most accesses
don't go through that function anyway.
Author: Andres Freund
Discussion: https://postgr.es/m/20181107174403.zai7fedgcjoqx44p@alap3.anarazel.de
In a lot of nodes the return slot is not required. That can either be
because the node doesn't do any projection (say an Append node), or
because the node does perform projections but the projection is
optimized away because the projection would yield an identical row.
Slots aren't that small, especially for wide rows, so it's worthwhile
to avoid creating them. It's not possible to just skip creating the
slot - it's currently used to determine the tuple descriptor returned
by ExecGetResultType(). So separate the determination of the result
type from the slot creation. The work previously done internally
ExecInitResultTupleSlotTL() can now also be done separately with
ExecInitResultTypeTL() and ExecInitResultSlot(). That way nodes that
aren't guaranteed to need a result slot, can use
ExecInitResultTypeTL() to determine the result type of the node, and
ExecAssignScanProjectionInfo() (via
ExecConditionalAssignProjectionInfo()) determines that a result slot
is needed, it is created with ExecInitResultSlot().
Besides the advantage of avoiding to create slots that then are
unused, this is necessary preparation for later patches around tuple
table slot abstraction. In particular separating the return descriptor
and slot is a prerequisite to allow JITing of tuple deforming with
knowledge of the underlying tuple format, and to avoid unnecessarily
creating JITed tuple deforming for virtual slots.
This commit removes a redundant argument from
ExecInitResultTupleSlotTL(). While this commit touches a lot of the
relevant lines anyway, it'd normally still not worthwhile to cause
breakage, except that aforementioned later commits will touch *all*
ExecInitResultTupleSlotTL() callers anyway (but fits worse
thematically).
Author: Andres Freund
Discussion: https://postgr.es/m/20181105210039.hh4vvi4vwoq5ba2q@alap3.anarazel.de
This mainly de-duplicates code. As evaluating a system variable isn't
the hottest path and the current inline implementation ends up calling
out to an external function anyway, this is OK from a performance POV.
The main motivation for de-duplicating is the upcoming slot
abstraction work, after which there's not guaranteed to be a HeapTuple
backing the slot.
Author: Andres Freund, Amit Khandekar
Discussion: https://postgr.es/m/20181105210039.hh4vvi4vwoq5ba2q@alap3.anarazel.de
Previously the expressions were built with the HashJoinState as a
parent. That's incorrect.
Currently this does not appear to be harmful, but for the upcoming
'slot abstraction' work this proves to be problematic, as the
underlying slot types can differ between Hash and HashJoin. It's
possible that this already causes a problem, but I've not been able to
come up with a scenario. Therefore don't backpatch at this point.
Author: Andres Freund
Discussion: https://postgr.es/m/20180220224318.gw4oe5jadhpmcdnm@alap3.anarazel.de
This only became a problem with 4c640f4f38, which didn't synchronize
the value agg_strict_input_check.nargs is set to, with the guard
condition for emitting the operation.
Besides such instructions being unnecessary overhead, currently the
LLVM JIT provider doesn't support them. It seems more sensible to
avoid generating such instruction than supporting them. Add assertions
to make it easier to debug a potential further occurance.
Discussion: https://postgr.es/m/2a505161-2727-2473-7c46-591ed108ac52@email.cz
Backpatch: 11-, like 4c640f4f38.
I (Andres) broke this unintentionally in 69c3936a14, by checking
strictness for all input expressions computed for an aggregate, rather
than just the input for the aggregate transition function.
Reported-By: Ondřej Bouda
Bisected-By: Tom Lane
Diagnosed-By: Andrew Gierth
Discussion: https://postgr.es/m/2a505161-2727-2473-7c46-591ed108ac52@email.cz
Backpatch: 11-, like 69c3936a14
NULL keys in left joins were skipped when building batch files.
Repair, by making the keep_nulls argument to ExecHashGetHashValue()
depend on whether this is a left outer join, as we do in other
paths.
Bug #15475. Thinko in 1804284042. Back-patch to 11.
Reported-by: Paul Schaap
Diagnosed-by: Andrew Gierth
Dicussion: https://postgr.es/m/15475-11a7a783fed72a36%40postgresql.org
An array-type coercion appearing within a CASE that has a constant
(after const-folding) test expression was mangled by the planner, causing
all the elements of the resulting array to be equal to the coerced value
of the CASE's test expression. This is my oversight in commit c12d570fa:
that changed ArrayCoerceExpr to use a subexpression involving a
CaseTestExpr, and I didn't notice that eval_const_expressions needed an
adjustment to keep from folding such a CaseTestExpr to a constant when
it's inside a suitable CASE.
This is another in what's getting to be a depressingly long line of bugs
associated with misidentification of the referent of a CaseTestExpr.
We're overdue to redesign that mechanism; but any such fix is unlikely
to be back-patchable into v11. As a stopgap, fix eval_const_expressions
to do what it must here. Also add a bunch of comments pointing out the
restrictions and assumptions that are needed to make this work at all.
Also fix a related oversight: contain_context_dependent_node() was not
aware of the relationship of ArrayCoerceExpr to CaseTestExpr. That was
somewhat fail-soft, in that the outcome of a wrong answer would be to
prevent optimizations that could have been made, but let's fix it while
we're at it.
Per bug #15471 from Matt Williams. Back-patch to v11 where the faulty
logic came in.
Discussion: https://postgr.es/m/15471-1117f49271989bad@postgresql.org
es_leaf_result_relations doesn't exist; perhaps this was an old name
for es_tuple_routing_result_relations, or maybe this comment has gone
unmaintained through multiple rounds of whacking the code around.
Related comment in execnodes.h was both obsolete and ungrammatical.
This allows the compiler / linker to mark affected pages as read-only.
There's a fair number of pre-requisite changes, to allow the const
properly be propagated. Most of consts were already required for
correctness anyway, just not represented on the type-level. Arguably
we could be more aggressive in using consts in related code, but..
This requires using a few of the types underlying typedefs that
removes pointers (e.g. const NameData *) as declaring the typedefed
type constant doesn't have the same meaning (it makes the variable
const, not what it points to).
Discussion: https://postgr.es/m/20181015200754.7y7zfuzsoux2c4ya@alap3.anarazel.de
There's several reasons for this change:
1) It reduces the total size of TupleTableSlot / reduces alignment
padding, making the commonly accessed members fit into a single
cacheline (but we currently do not force proper alignment, so
that's not yet guaranteed to be helpful)
2) Combining the booleans into a flag allows to combine read/writes
from memory.
3) With the upcoming slot abstraction changes, it allows to have core
and extended flags, in a memory efficient way.
Author: Ashutosh Bapat and Andres Freund
Discussion: https://postgr.es/m/20180220224318.gw4oe5jadhpmcdnm@alap3.anarazel.de
heaptuple.c was never a particular good fit for slot_getattr(),
slot_getsomeattrs() and slot_getmissingattrs(), but in upcoming
changes slots will be made more abstract (allowing slots that contain
different types of tuples), making it clearly the wrong place.
Note that slot_deform_tuple() remains in it's current place, as it
clearly deals with a HeapTuple. getmissingattrs() also remains, but
it's less clear that that's correct - but execTuples.c wouldn't be the
right place.
Author: Ashutosh Bapat.
Discussion: https://postgr.es/m/20180220224318.gw4oe5jadhpmcdnm@alap3.anarazel.de
If there are many ExecRowMark structs, we spent O(N^2) time in
ExecFindRowMark during executor startup. Once upon a time this was
not of great concern, but the addition of native partitioning has
squeezed out enough other costs that this can become the dominant
overhead in some use-cases for tables with many partitions.
To fix, simply replace that List data structure with an array.
This adds a little bit of cost to execCurrentOf(), but not much,
and anyway that code path is neither of large importance nor very
efficient now. If we ever decide it is a bottleneck, constructing a
hash table for lookup-by-tableoid would likely be the thing to do.
Per complaint from Amit Langote, though this is different from
his fix proposal.
Discussion: https://postgr.es/m/468c85d9-540e-66a2-1dde-fec2b741e688@lab.ntt.co.jp
In the wake of commit f2343653f, we no longer need some fields that
were used before to control executor lock acquisitions:
* PlannedStmt.nonleafResultRelations can go away entirely.
* partitioned_rels can go away from Append, MergeAppend, and ModifyTable.
However, ModifyTable still needs to know the RT index of the partition
root table if any, which was formerly kept in the first entry of that
list. Add a new field "rootRelation" to remember that. rootRelation is
partly redundant with nominalRelation, in that if it's set it will have
the same value as nominalRelation. However, the latter field has a
different purpose so it seems best to keep them distinct.
Amit Langote, reviewed by David Rowley and Jesper Pedersen,
and whacked around a bit more by me
Discussion: https://postgr.es/m/468c85d9-540e-66a2-1dde-fec2b741e688@lab.ntt.co.jp
Commit 9a3cebeaa changed things so that parallel workers didn't obtain
any lock of their own on tables they access. That was clearly a bad
idea, but I'd mistakenly supposed that it was the intended end result
of the series of patches for simplifying the executor's lock management.
Undo that change in relation_open(), and adjust ExecOpenScanRelation()
so that it gets the correct lock if inside a parallel worker.
In passing, clean up some more obsolete comments about when locks
are acquired.
Discussion: https://postgr.es/m/468c85d9-540e-66a2-1dde-fec2b741e688@lab.ntt.co.jp
We already have appropriate locks on every relation listed in the
query's rangetable before we reach the executor. Take the next step
in exploiting that knowledge by removing code that worries about
taking locks on non-leaf result relations in a partitioned table.
In particular, get rid of ExecLockNonLeafAppendTables and a stanza in
InitPlan that asserts we already have locks on certain such tables.
In passing, clean up some now-obsolete comments in InitPlan.
Amit Langote, reviewed by David Rowley and Jesper Pedersen,
and whacked around a bit more by me
Discussion: https://postgr.es/m/468c85d9-540e-66a2-1dde-fec2b741e688@lab.ntt.co.jp
Historically we forbade datatype-specific comparison functions from
returning INT_MIN, so that it would be safe to invert the sort order
just by negating the comparison result. However, this was never
really safe for comparison functions that directly return the result
of memcmp(), strcmp(), etc, as POSIX doesn't place any such restriction
on those library functions. Buildfarm results show that at least on
recent Linux on s390x, memcmp() actually does return INT_MIN sometimes,
causing sort failures.
The agreed-on answer is to remove this restriction and fix relevant
call sites to not make such an assumption; code such as "res = -res"
should be replaced by "INVERT_COMPARE_RESULT(res)". The same is needed
in a few places that just directly negated the result of memcmp or
strcmp.
To help find places having this problem, I've also added a compile option
to nbtcompare.c that causes some of the commonly used comparators to
return INT_MIN/INT_MAX instead of their usual -1/+1. It'd likely be
a good idea to have at least one buildfarm member running with
"-DSTRESS_SORT_INT_MIN". That's far from a complete test of course,
but it should help to prevent fresh introductions of such bugs.
This is a longstanding portability hazard, so back-patch to all supported
branches.
Discussion: https://postgr.es/m/20180928185215.ffoq2xrq5d3pafna@alap3.anarazel.de
Instead of doing a lot of list_nth() accesses to es_range_table,
create a flattened pointer array during executor startup and index
into that to get at individual RangeTblEntrys.
This eliminates one source of O(N^2) behavior with lots of partitions.
(I'm not exactly convinced that it's the most important source, but
it's an easy one to fix.)
Amit Langote and David Rowley
Discussion: https://postgr.es/m/468c85d9-540e-66a2-1dde-fec2b741e688@lab.ntt.co.jp
Create an array estate->es_relations[] paralleling the es_range_table,
and store references to Relations (relcache entries) there, so that any
given RT entry is opened and closed just once per executor run. Scan
nodes typically still call ExecOpenScanRelation, but ExecCloseScanRelation
is no more; relation closing is now done centrally in ExecEndPlan.
This is slightly more complex than one would expect because of the
interactions with relcache references held in ResultRelInfo nodes.
The general convention is now that ResultRelInfo->ri_RelationDesc does
not represent a separate relcache reference and so does not need to be
explicitly closed; but there is an exception for ResultRelInfos in the
es_trig_target_relations list, which are manufactured by
ExecGetTriggerResultRel and have to be cleaned up by
ExecCleanUpTriggerState. (That much was true all along, but these
ResultRelInfos are now more different from others than they used to be.)
To allow the partition pruning logic to make use of es_relations[] rather
than having its own relcache references, adjust PartitionedRelPruneInfo
to store an RT index rather than a relation OID.
Amit Langote, reviewed by David Rowley and Jesper Pedersen,
some mods by me
Discussion: https://postgr.es/m/468c85d9-540e-66a2-1dde-fec2b741e688@lab.ntt.co.jp
Instead of locking tables during executor startup, just Assert that
suitable locks were obtained already during the parse/plan pipeline
(or re-obtained by the plan cache). This must be so, else we have a
hazard that concurrent DDL has invalidated the plan.
This is pretty inefficient as well as undercommented, but it's all going
to go away shortly, so I didn't try hard. This commit is just another
attempt to use the buildfarm to see if we've missed anything in the plan
to simplify the executor's table management.
Note that the change needed here in relation_open() exposes that
parallel workers now really are accessing tables without holding any
lock of their own, whereas they were not doing that before this commit.
This does not give me a warm fuzzy feeling about that aspect of parallel
query; it does not seem like a good design, and we now know that it's
had exactly no actual testing. I think that we should modify parallel
query so that that change can be reverted.
Discussion: https://postgr.es/m/468c85d9-540e-66a2-1dde-fec2b741e688@lab.ntt.co.jp
I (Andres) was more than a bit hasty in committing 33001fd7a7
after last minute changes, leading to a number of problems (jit output
was only shown for JIT in parallel workers, and just EXPLAIN without
ANALYZE didn't work). Lukas luckily found these issues quickly.
Instead of combining instrumentation in in standard_ExecutorEnd(), do
so on demand in the new ExplainPrintJITSummary().
Also update a documentation example of the JIT output, changed in
52050ad8eb.
Author: Lukas Fittl, with minor changes by me
Discussion: https://postgr.es/m/CAP53PkxmgJht69pabxBXJBM+0oc6kf3KHMborLP7H2ouJ0CCtQ@mail.gmail.com
Backpatch: 11, where JIT compilation was introduced
Instead of recomputing the required lock levels in all these places,
just use what commit fdba460a2 made the parser store in the RTE fields.
This already simplifies the code measurably in these places, and
follow-on changes will remove a bunch of no-longer-needed infrastructure.
In a few cases, this change causes us to acquire a higher lock level
than we did before. This is OK primarily because said higher lock level
should've been acquired already at query parse time; thus, we're saving
a useless extra trip through the shared lock manager to acquire a lesser
lock alongside the original lock. The only known exception to this is
that re-execution of a previously planned SELECT FOR UPDATE/SHARE query,
for a table that uses ROW_MARK_REFERENCE or ROW_MARK_COPY methods, might
have gotten only AccessShareLock before. Now it will get RowShareLock
like the first execution did, which seems fine.
While there's more to do, push it in this state anyway, to let the
buildfarm help verify that nothing bad happened.
Amit Langote, reviewed by David Rowley and Jesper Pedersen,
and whacked around a bit more by me
Discussion: https://postgr.es/m/468c85d9-540e-66a2-1dde-fec2b741e688@lab.ntt.co.jp
It's inefficient to use a single slot for mapping between tuple
descriptors for multiple tuples, as previously done when using
ConvertPartitionTupleSlot(), as that means the slot's tuple descriptors
change for every tuple.
Previously we also, via ConvertPartitionTupleSlot(), built new tuples
after the mapping even in cases where we, immediately afterwards,
access individual columns again.
Refactor the code so one slot, on demand, is used for each
partition. That avoids having to change the descriptor (and allows to
use the more efficient "fixed" tuple slots). Then use slot->slot
mapping, to avoid unnecessarily forming a tuple.
As the naming between the tuple and slot mapping functions wasn't
consistent, rename them to execute_attr_map_{tuple,slot}. It's likely
that we'll also rename convert_tuples_by_* to denote that these
functions "only" build a map, but that's left for later.
Author: Amit Khandekar and Amit Langote, editorialized by me
Reviewed-By: Amit Langote, Amit Khandekar, Andres Freund
Discussion:
https://postgr.es/m/CAJ3gD9fR0wRNeAE8VqffNTyONS_UfFPRpqxhnD9Q42vZB+Jvpg@mail.gmail.comhttps://postgr.es/m/e4f9d743-cd4b-efb0-7574-da21d86a7f36%40lab.ntt.co.jp
Backpatch: -
Add RangeTblEntry.rellockmode, which records the appropriate lock mode for
each RTE_RELATION rangetable entry (either AccessShareLock, RowShareLock,
or RowExclusiveLock depending on the RTE's role in the query).
This patch creates the field and makes all creators of RTE nodes fill it
in reasonably, but for the moment nothing much is done with it. The plan
is to replace assorted post-parser logic that re-determines the right
lockmode to use with simple uses of rte->rellockmode. For now, just add
Asserts in each of those places that the rellockmode matches what they are
computing today. (In some cases the match isn't perfect, so the Asserts
are weaker than you might expect; but this seems OK, as per discussion.)
This passes check-world for me, but it seems worth pushing in this state
to see if the buildfarm finds any problems in cases I failed to test.
catversion bump due to change of stored rules.
Amit Langote, reviewed by David Rowley and Jesper Pedersen,
and whacked around a bit more by me
Discussion: https://postgr.es/m/468c85d9-540e-66a2-1dde-fec2b741e688@lab.ntt.co.jp
TupleDescGetSlot() was kept around for backward compatibility for
user-written SRFs. With the TupleTableSlot abstraction work, that code
will need to be version specific anyway, so there's no point in
keeping the function around any longer.
Author: Ashutosh Bapat
Reviewed-By: Andres Freund
Discussion: https://postgr.es/m/20180220224318.gw4oe5jadhpmcdnm@alap3.anarazel.de
Upcoming changes introduce further types of tuple table slots, in
preparation of making table storage pluggable. New storage methods
will have different representation of tuples, therefore the slot
accessor should refer explicitly to heap tuples.
Instead of just renaming the functions, split it into one function
that accepts heap tuples not residing in buffers, and one accepting
ones in buffers. Previously one function was used for both, but that
was a bit awkward already, and splitting will allow us to represent
slot types for tuples in buffers and normal memory separately.
This is split out from the patch introducing abstract slots, as this
largely consists out of mechanical changes.
Author: Ashutosh Bapat
Reviewed-By: Andres Freund
Discussion: https://postgr.es/m/20180220224318.gw4oe5jadhpmcdnm@alap3.anarazel.de
Previously, when using parallel query, EXPLAIN (ANALYZE)'s JIT
compilation timings did not include the overhead from doing so on the
workers. Fix that.
We do so by simply aggregating the cost of doing JIT compilation on
workers and the leader together. Arguably that's not quite accurate,
because the total time spend doing so is spent in parallel - but it's
hard to do much better. For additional detail, when VERBOSE is
specified, the stats for workers are displayed separately.
Author: Amit Khandekar and Andres Freund
Discussion: https://postgr.es/m/CAJ3gD9eLrz51RK_gTkod+71iDcjpB_N8eC6vU2AW-VicsAERpQ@mail.gmail.com
Backpatch: 11-
In a case where we have multiple relation-scan nodes in a cursor plan,
such as a scan of an inheritance tree, it's possible to fetch from a
given scan node, then rewind the cursor and fetch some row from an
earlier scan node. In such a case, execCurrent.c mistakenly thought
that the later scan node was still active, because ExecReScan hadn't
done anything to make it look not-active. We'd get some sort of
failure in the case of a SeqScan node, because the node's scan tuple
slot would be pointing at a HeapTuple whose t_self gets reset to
invalid by heapam.c. But it seems possible that for other relation
scan node types we'd actually return a valid tuple TID to the caller,
resulting in updating or deleting a tuple that shouldn't have been
considered current. To fix, forcibly clear the ScanTupleSlot in
ExecScanReScan.
Another issue here, which seems only latent at the moment but could
easily become a live bug in future, is that rewinding a cursor does
not necessarily lead to *immediately* applying ExecReScan to every
scan-level node in the plan tree. Upper-level nodes will think that
they can postpone that call if their child node is already marked
with chgParam flags. I don't see a way for that to happen today in
a plan tree that's simple enough for execCurrent.c's search_plan_tree
to understand, but that's one heck of a fragile assumption. So, add
some logic in search_plan_tree to detect chgParam flags being set on
nodes that it descended to/through, and assume that that means we
should consider lower scan nodes to be logically reset even if their
ReScan call hasn't actually happened yet.
Per bug #15395 from Matvey Arye. This has been broken for a long time,
so back-patch to all supported branches.
Discussion: https://postgr.es/m/153764171023.14986.280404050547008575@wrigleys.postgresql.org
The original coding for XMLTABLE thought it could represent a default
namespace by a T_String Value node with a null string pointer. That's
not okay, though; in particular outfuncs.c/readfuncs.c are not on board
with such a representation, meaning you'll get a null pointer crash
if you try to store a view or rule containing this construct.
To fix, change the parsetree representation so that we have a NULL
list element, instead of a bogus Value node.
This isn't really a functional limitation since default XML namespaces
aren't yet implemented in the executor; you'd just get "DEFAULT
namespace is not supported" anyway. But crashes are not nice, so
back-patch to v10 where this syntax was added. Ordinarily we'd consider
a parsetree representation change to be un-backpatchable; but since
existing releases would crash on the way to storing such constructs,
there can't be any existing views/rules to be incompatible with.
Per report from Andrey Lepikhov.
Discussion: https://postgr.es/m/3690074f-abd2-56a9-144a-aa5545d7a291@postgrespro.ru
The EvalPlanQual machinery assumes that any initplans (that is,
uncorrelated sub-selects) used during an EPQ recheck would have already
been evaluated during the main query; this is implicit in the fact that
execPlan pointers are not copied into the EPQ estate's es_param_exec_vals.
But it's possible for that assumption to fail, if the initplan is only
reached conditionally. For example, a sub-select inside a CASE expression
could be reached during a recheck when it had not been previously, if the
CASE test depends on a column that was just updated.
This bug is old, appearing to date back to my rewrite of EvalPlanQual in
commit 9f2ee8f28, but was not detected until Kyle Samson reported a case.
To fix, force all not-yet-evaluated initplans used within the EPQ plan
subtree to be evaluated at the start of the recheck, before entering the
EPQ environment. This could be inefficient, if such an initplan is
expensive and goes unused again during the recheck --- but that's piling
one layer of improbability atop another. It doesn't seem worth adding
more complexity to prevent that, at least not in the back branches.
It was convenient to use the new-in-v11 ExecEvalParamExecParams function
to implement this, but I didn't like either its name or the specifics of
its API, so revise that.
Back-patch all the way. Rather than rewrite the patch to avoid depending
on bms_next_member() in the oldest branches, I chose to back-patch that
function into 9.4 and 9.3. (This isn't the first time back-patches have
needed that, and it exhausted my patience.) I also chose to back-patch
some test cases added by commits 71404af2a and 342a1ffa2 into 9.4 and 9.3,
so that the 9.x versions of eval-plan-qual.spec are all the same.
Andrew Gierth diagnosed the problem and contributed the added test cases,
though the actual code changes are by me.
Discussion: https://postgr.es/m/A033A40A-B234-4324-BE37-272279F7B627@tripadvisor.com
This patch removes two sources of interference between nominally
independent functions when one SPI-using function calls another,
perhaps without knowing that it does so.
Chapman Flack pointed out that xml.c's query_to_xml_internal() expects
SPI_tuptable and SPI_processed to stay valid across datatype output
function calls; but it's possible that such a call could involve
re-entrant use of SPI. It seems likely that there are similar hazards
elsewhere, if not in the core code then in third-party SPI users.
Previously SPI_finish() reset SPI's API globals to zeroes/nulls, which
would typically make for a crash in such a situation. Restoring them
to the values they had at SPI_connect() seems like a considerably more
useful behavior, and it still meets the design goal of not leaving any
dangling pointers to tuple tables of the function being exited.
Also, cause SPI_connect() to reset these variables to zeroes/nulls after
saving them. This prevents interference in the opposite direction: it's
possible that a SPI-using function that's only ever been tested standalone
contains assumptions that these variables start out as zeroes. That was
the case as long as you were the outermost SPI user, but not so much for
an inner user. Now it's consistent.
Report and fix suggestion by Chapman Flack, actual patch by me.
Back-patch to all supported branches.
Discussion: https://postgr.es/m/9fa25bef-2e4f-1c32-22a4-3ad0723c4a17@anastigmatix.net
When executing a SubPlan in an expression, the EState's direction
field was left alone, resulting in an attempt to execute the subplan
backwards if it was encountered during a backwards scan of a cursor.
Also, though much less likely, it was possible to reach the execution
of an InitPlan while in backwards-scan state.
Repair by saving/restoring estate->es_direction and forcing forward
scan mode in the relevant places.
Backpatch all the way, since this has been broken since 8.3 (prior to
commit c7ff7663e, SubPlans had their own EStates rather than sharing
the parent plan's, so there was no confusion over scan direction).
Per bug #15336 reported by Vladimir Baranoff; analysis and patch by
me, review by Tom Lane.
Discussion: https://postgr.es/m/153449812167.1304.1741624125628126322@wrigleys.postgresql.org
In a multi-layer partitioning setup, if at plan time all the
sub-partitions are pruned but the intermediate one remains, the executor
later throws a spurious error that there's nothing to prune. That is
correct, but there's no reason to throw an error. Therefore, don't.
Reported-by: Andreas Seltenreich <seltenreich@gmx.de>
Author: David Rowley <david.rowley@2ndquadrant.com>
Discussion: https://postgr.es/m/87in4h98i0.fsf@ansel.ydns.eu
Currently, we release the asynchronous resources as soon as it is evident
that no more rows will be needed e.g. when a Limit is filled. This can be
problematic especially for custom and foreign scans where we can scan
backward. Fix that by disallowing the shutting down of resources in such
cases.
Reported-by: Robert Haas
Analysed-by: Robert Haas and Amit Kapila
Author: Amit Kapila
Reviewed-by: Robert Haas
Backpatch-through: 9.6 where this code was introduced
Discussion: https://postgr.es/m/86137f17-1dfb-42f9-7421-82fd786b04a1@anayrat.info
Multiple calls to XMLTABLE in a query (e.g. laterally applying it to a
table with an xml column, an important use-case) were leaking large
amounts of memory into the per-query context, blowing up memory usage.
Repair by reorganizing memory context usage in nodeTableFuncscan; use
the usual per-tuple context for row-by-row evaluations instead of
perValueCxt, and use the explicitly created context -- renamed from
perValueCxt to perTableCxt -- for arguments and state for each
individual table-generation operation.
Backpatch to PG10 where this code was introduced.
Original report by IRC user begriffs; analysis and patch by me.
Reviewed by Tom Lane and Pavel Stehule.
Discussion: https://postgr.es/m/153394403528.10284.7530399040974170549@wrigleys.postgresql.org
These changes were put in at some stage of the development process, but
are unnecessary and should not have made it into the final patch. Mea
culpa.
Per gripe from Andreas Freund
Backpatch to REL_11_STABLE
The buffer usage stats is accounted only for the execution phase of the
node. For Gather and Gather Merge nodes, such stats are accumulated at
the time of shutdown of workers which is done after execution of node due
to which we missed to account them for such nodes. Fix it by treating
nodes as running while we shut down them.
We can also miss accounting for a Limit node when Gather or Gather Merge
is beneath it, because it can finish the execution before shutting down
such nodes. So we allow a Limit node to shut down the resources before it
completes the execution.
In the passing fix the gather node code to allow workers to shut down as
soon as we find that all the tuples from the workers have been retrieved.
The original code use to do that, but is accidently removed by commit
01edb5c7fc.
Reported-by: Adrien Nayrat
Author: Amit Kapila and Robert Haas
Reviewed-by: Robert Haas and Andres Freund
Backpatch-through: 9.6 where this code was introduced
Discussion: https://postgr.es/m/86137f17-1dfb-42f9-7421-82fd786b04a1@anayrat.info
In the leader backend, we don't track the buffer usage for ExecutorStart
phase whereas in worker backend we track it for ExecutorStart phase as
well. This leads to different value for buffer usage stats for the
parallel and non-parallel query. Change the code so that worker backend
also starts tracking buffer usage after ExecutorStart.
Author: Amit Kapila and Robert Haas
Reviewed-by: Robert Haas and Andres Freund
Backpatch-through: 9.6 where this code was introduced
Discussion: https://postgr.es/m/86137f17-1dfb-42f9-7421-82fd786b04a1@anayrat.info
The previous coding here supposed that if run-time partitioning applied to
a particular Append/MergeAppend plan, then all child plans of that node
must be members of a single partitioning hierarchy. This is totally wrong,
since an Append could be formed from a UNION ALL: we could have multiple
hierarchies sharing the same Append, or child plans that aren't part of any
hierarchy.
To fix, restructure the related plan-time and execution-time data
structures so that we can have a separate list or array for each
partitioning hierarchy. Also track subplans that are not part of any
hierarchy, and make sure they don't get pruned.
Per reports from Phil Florent and others. Back-patch to v11, since
the bug originated there.
David Rowley, with a lot of cosmetic adjustments by me; thanks also
to Amit Langote for review.
Discussion: https://postgr.es/m/HE1PR03MB17068BB27404C90B5B788BCABA7B0@HE1PR03MB1706.eurprd03.prod.outlook.com
Some operations were being done in a longer-lived memory context,
causing intra-query leaks. It's not noticeable unless you're doing a
large COPY, but if you are, it eats enough memory to cause a problem.
Co-authored-by: Kohei KaiGai <kaigai@heterodb.com>
Co-authored-by: Amit Langote <Langote_Amit_f8@lab.ntt.co.jp>
Co-authored-by: Álvaro Herrera <alvherre@alvh.no-ip.org>
Discussion: https://postgr.es/m/CAOP8fzYtVFWZADq4c=KoTAqgDrHWfng+AnEPEZccyxqxPVbbWQ@mail.gmail.com
CopyFrom allows multi-inserts to be used for non-partitioned tables, but
this was disabled for partitioned tables. The reason for this appeared
to be that the tuple may not belong to the same partition as the
previous tuple did. Not allowing multi-inserts here greatly slowed down
imports into partitioned tables. These could take twice as long as a
copy to an equivalent non-partitioned table. It seems wise to do
something about this, so this change allows the multi-inserts by
flushing the so-far inserted tuples to the partition when the next tuple
does not belong to the same partition, or when the buffer fills. This
improves performance when the next tuple in the stream commonly belongs
to the same partition as the previous tuple.
In cases where the target partition changes on every tuple, using
multi-inserts slightly slows the performance. To get around this we
track the average size of the batches that have been inserted and
adaptively enable or disable multi-inserts based on the size of the
batch. Some testing was done and the regression only seems to exist
when the average size of the insert batch is close to 1, so let's just
enable multi-inserts when the average size is at least 1.3. More
performance testing might reveal a better number for, this, but since
the slowdown was only 1-2% it does not seem critical enough to spend too
much time calculating it. In any case it may depend on other factors
rather than just the size of the batch.
Allowing multi-inserts for partitions required a bit of work around the
per-tuple memory contexts as we must flush the tuples when the next
tuple does not belong the same partition. In which case there is no
good time to reset the per-tuple context, as we've already built the new
tuple by this time. In order to work around this we maintain two
per-tuple contexts and just switch between them every time the partition
changes and reset the old one. This does mean that the first of each
batch of tuples is not allocated in the same memory context as the
others, but that does not matter since we only reset the context once
the previous batch has been inserted.
Author: David Rowley <david.rowley@2ndquadrant.com>
Reviewed-by: Melanie Plageman <melanieplageman@gmail.com>
Instead of repeatedly fishing the data out of the relcache entry,
let's use the version that we cached in the PartitionDispatch. We
could alternatively rip out the PartitionDispatch fields altogether,
but it doesn't make much sense to have them and not use them; before
this patch, partdesc was set but altogether unused. Amit Langote and
I both thought using them was a litle better than removing them, so
this patch takes that approach.
Discussion: http://postgr.es/m/CA+TgmobFnxcaW-Co-XO8=yhJ5pJXoNkCj6Z7jm9Mwj9FGv-D7w@mail.gmail.com
Due to inlining it previously was possible that an ExprContext's
shutdown callback pointed to a JITed function. As the JIT context
previously was shut down before the shutdown callbacks were called,
that could lead to segfaults. Fix the ordering.
Reported-By: Dmitry Dolgov
Author: Andres Freund
Discussion: https://postgr.es/m/CA+q6zcWO7CeAJtHBxgcHn_hj+PenM=tvG0RJ93X1uEJ86+76Ug@mail.gmail.com
Backpatch: 11-, where JIT compilation was added
This comment was copy-pasted from nodeAppend.c to nodeMergeAppend.c, but
while committing 5220bb7533, I modified wrong copy of it.
Spotted by David Rowley
A collection of typos I happened to spot while reading code, as well as
grepping for common mistakes.
Backpatch to all supported versions, as applicable, to avoid conflicts
when backporting other commits in the future.
An update that causes the tuple to be moved to a different partition was
missing out on re-constructing the to-be-updated tuple, based on the latest
tuple in the update chain. Instead, it's simply deleting the latest tuple
and inserting a new tuple in the new partition based on the old tuple.
Commit 2f17844104 didn't consider this case, so some of the updates were
getting lost.
In passing, change the argument order for output parameter in ExecDelete
and add some commentary about it.
Reported-by: Pavan Deolasee
Author: Amit Khandekar, with minor changes by me
Reviewed-by: Dilip Kumar, Amit Kapila and Alvaro Herrera
Backpatch-through: 11
Discussion: https://postgr.es/m/CAJ3gD9fRbEzDqdeDq1jxqZUb47kJn+tQ7=Bcgjc8quqKsDViKQ@mail.gmail.com
nodeWindowAgg.c failed to cope with the possibility that no ordering
columns are defined in the window frame for GROUPS mode or RANGE OFFSET
mode, leading to assertion failures or odd errors, as reported by Masahiko
Sawada and Lukas Eder. In RANGE OFFSET mode, an ordering column is really
required, so add an Assert about that. In GROUPS mode, the code would
work, except that the node initialization code wasn't in sync with the
execution code about when to set up tuplestore read pointers and spare
slots. Fix the latter for consistency's sake (even though I think the
changes described below make the out-of-sync cases unreachable for now).
Per SQL spec, a single ordering column is required for RANGE OFFSET mode,
and at least one ordering column is required for GROUPS mode. The parser
enforced the former but not the latter; add a check for that.
We were able to reach the no-ordering-column cases even with fully spec
compliant queries, though, because the planner would drop partitioning
and ordering columns from the generated plan if they were redundant with
earlier columns according to the redundant-pathkey logic, for instance
"PARTITION BY x ORDER BY y" in the presence of a "WHERE x=y" qual.
While in principle that's an optimization that could save some pointless
comparisons at runtime, it seems unlikely to be meaningful in the real
world. I think this behavior was not so much an intentional optimization
as a side-effect of an ancient decision to construct the plan node's
ordering-column info by reverse-engineering the PathKeys of the input
path. If we give up redundant-column removal then it takes very little
code to generate the plan node info directly from the WindowClause,
ensuring that we have the expected number of ordering columns in all
cases. (If anyone does complain about this, the planner could perhaps
be taught to remove redundant columns only when it's safe to do so,
ie *not* in RANGE OFFSET mode. But I doubt anyone ever will.)
With these changes, the WindowAggPath.winpathkeys field is not used for
anything anymore, so remove it.
The test cases added here are not actually very interesting given the
removal of the redundant-column-removal logic, but they would represent
important corner cases if anyone ever tries to put that back.
Tom Lane and Masahiko Sawada. Back-patch to v11 where RANGE OFFSET
and GROUPS modes were added.
Discussion: https://postgr.es/m/CAD21AoDrWqycq-w_+Bx1cjc+YUhZ11XTj9rfxNiNDojjBx8Fjw@mail.gmail.com
Discussion: https://postgr.es/m/153086788677.17476.8002640580496698831@wrigleys.postgresql.org
When executing CALL in PL/pgSQL, we need to set a snapshot before
invoking the to-be-called procedure. Otherwise, the to-be-called
procedure might end up running without a snapshot. For LANGUAGE SQL
procedures, this would result in an assertion failure. (For most other
languages, this is usually not a problem, because those use SPI and SPI
sets snapshots in most cases.) Setting the snapshot restores the
behavior of how CALL worked when it was handled as a generic SQL
statement in PL/pgSQL (exec_stmt_execsql()).
This change revealed another problem: In SPI_commit(), we popped the
active snapshot before committing the transaction, to avoid "snapshot %p
still active" errors. However, there is no particular reason why only
at most one snapshot should be on the stack. So change this to pop all
active snapshots instead of only one.
The previous coding saved pointers into the partitioned table's relcache
entry, but then closed the relcache entry, causing those pointers to
nominally become dangling. Actual trouble would be seen in the field
only if a relcache flush occurred mid-query, but that's hardly out of
the question.
While we could fix this by copying all the data in question at query
start, it seems better to just hold the relcache entry open for the
whole query.
While at it, improve the handling of support-function lookups: do that
once per query not once per pruning test. There's still something to be
desired here, in that we fail to exploit the possibility of caching data
across queries in the fn_extra fields of the relcache's FmgrInfo structs,
which could happen if we just used those structs in-place rather than
copying them. However, combining that with the possibility of per-query
lookups of cross-type comparison functions seems to require changes in the
APIs of a lot of the pruning support functions, so it's too invasive to
consider as part of this patch. A win would ensue only for complex
partition key data types (e.g. arrays), so it may not be worth the
trouble.
David Rowley and Tom Lane
Discussion: https://postgr.es/m/17850.1528755844@sss.pgh.pa.us
We don't need two passes if we scan child partitions before parents,
as that way the children's present_parts are up to date before they're
needed. I (tgl) think there's actually a bug being fixed here, for the
case of an intermediate partitioned table with no direct leaf children,
but haven't attempted to construct a test case to prove it.
David Rowley
Discussion: https://postgr.es/m/CAKJS1f-6GODRNgEtdPxCnAPme2h2hTztB6LmtfdmcYAAOE0kQg@mail.gmail.com
Starting with commit f0e44751d7, ExecConstraints was in charge of
running the partition constraint; commit 19c47e7c82 modified that so
that caller could request to skip that checking depending on some
conditions, but that commit and 15ce775faa together introduced a small
bug there which caused ExecInsert to request skipping the constraint
check but have this not be honored -- in effect doing the check twice.
This could have been fixed in a very small patch, but on further
analysis of the involved function and its callsites, it turns out to be
simpler to give the responsibility of checking the partition constraint
fully to the caller, and return ExecConstraints to its original
(pre-partitioning) shape where it only checked tuple descriptor-related
constraints. Each caller must do partition constraint checking on its
own schedule, which is more convenient after commit 2f17844104 anyway.
Reported-by: David Rowley
Author: David Rowley, Álvaro Herrera
Reviewed-by: Amit Langote, Amit Khandekar, Simon Riggs
Discussion: https://postgr.es/m/CAKJS1f8w8+awsxgea8wt7_UX8qzOQ=Tm1LD+U1fHqBAkXxkW2w@mail.gmail.com
These struct definitions were originally dropped into primnodes.h,
which is a poor choice since that's mainly intended for primitive
expression node types; these are not in that category. What they
are is auxiliary info in Plan trees, so move them to plannodes.h.
For consistency, also relocate some related code that was apparently
placed with the aid of a dartboard.
There's no interesting code changes in this commit, just reshuffling.
David Rowley and Tom Lane
Discussion: https://postgr.es/m/CAFj8pRBjrufA3ocDm8o4LPGNye9Y+pm1b9kCwode4X04CULG3g@mail.gmail.com
The initial coding of the run-time-pruning feature only coped with cases
where the partition key(s) are compared to Params. That is a bit silly;
we can allow it to work with any non-Var-containing stable expression, as
long as we take special care with expressions containing PARAM_EXEC Params.
The code is hardly any longer this way, and it's considerably clearer
(IMO at least). Per gripe from Pavel Stehule.
David Rowley, whacked around a bit by me
Discussion: https://postgr.es/m/CAFj8pRBjrufA3ocDm8o4LPGNye9Y+pm1b9kCwode4X04CULG3g@mail.gmail.com
Make sure that we don't exceed MaxAllocSize when increasing the number of
buckets. Perhaps later we'll remove that limit and use DSA_ALLOC_HUGE, but
for now just prevent further increases like the non-parallel code. This
change avoids the error from bug report #15225.
Author: Thomas Munro
Reviewed-By: Tom Lane
Reported-by: Frits Jalvingh
Discussion: https://postgr.es/m/152802081668.26724.16985037679312485972%40wrigleys.postgresql.org
The impact of VARIADIC on the combine/serialize/deserialize support
functions of an aggregate wasn't thought through carefully. There is
actually no impact, because variadicity isn't passed through to these
functions (and it doesn't seem like it would need to be). However,
lookup_agg_function was mistakenly told to check things as though it were
passed through. The net result was that it was impossible to declare an
aggregate that had both VARIADIC input and parallelism support functions.
In passing, fix a runtime check in nodeAgg.c for the combine function's
strictness to make its error message agree with the creation-time check.
The previous message was actually backwards, and it doesn't seem like
there's a good reason to have two versions of this message text anyway.
Back-patch to 9.6 where parallel aggregation was introduced.
Alexey Bashtanov; message fix by me
Discussion: https://postgr.es/m/f86dde87-fef4-71eb-0480-62754aaca01b@imap.cc
Since the SPI stack has been moved from TopTransactionContext to
TopMemoryContext, setting _SPI_stack to NULL in AtEOXact_SPI() leaks
memory. In fact, we don't need to do that anymore: We just leave the
allocated stack around for the next SPI use.
Also, refactor the SPI cleanup so that it is run both at transaction end
and when returning to the main loop on an exception. The latter is
necessary when a procedure calls a COMMIT or ROLLBACK command that
itself causes an error.
Recent gcc can warn about switch-case fall throughs that are not
explicitly labeled as intentional. This seems like a good thing,
so clean up the warnings exposed thereby by labeling all such
cases with comments that gcc will recognize.
In files that already had one or more suitable comments, I generally
matched the existing style of those. Otherwise I went with
/* FALLTHROUGH */, which is one of the spellings approved at the
more-restrictive-than-default level -Wimplicit-fallthrough=4.
(At the default level you can also spell it /* FALL ?THRU */,
and it's not picky about case. What you can't do is include
additional text in the same comment, so some existing comments
containing versions of this aren't good enough.)
Testing with gcc 8.0.1 (Fedora 28's current version), I found that
I also had to put explicit "break"s after elog(ERROR) or ereport(ERROR);
apparently, for this purpose gcc doesn't recognize that those don't
return. That seems like possibly a gcc bug, but it's fine because
in most places we did that anyway; so this amounts to a visit from the
style police.
Discussion: https://postgr.es/m/15083.1525207729@sss.pgh.pa.us
Without these fixes, changes to the inserted tuple made by remote
triggers are ignored when building local RETURNING tuples.
In the core code, call ExecInitRoutingInfo at a later point from
within ExecInitPartitionInfo so that the FDW callback gets invoked
after the returning list has been built. But move CheckValidResultRel
out of ExecInitRoutingInfo so that it can happen at an earlier stage.
In postgres_fdw, refactor assorted deparsing functions to work with
the RTE rather than the PlannerInfo, which saves us having to
construct a fake PlannerInfo in cases where we don't have a real one.
Then, we can pass down a constructed RTE that yields the correct
deparse result when no real one exists. Unfortunately, this
necessitates a hack that understands how the core code manages RT
indexes for update tuple routing, which is ugly, but we don't have a
better idea right now.
Original report, analysis, and patch by Etsuro Fujita. Heavily
refactored by me. Then worked over some more by Amit Langote.
Discussion: http://postgr.es/m/5AD4882B.10002@lab.ntt.co.jp
Remove the words "if not already done." This obsolete wording
corresponds to an early development version of what became edd44738bc.
Author: Etsuro Fujita
Reviewed-by: Amit Langote
Discussion: https://postgr.es/m/5ADF117B.5030606@lab.ntt.co.jp
Instead of doing ExecInitExpr every time a Param needs to be evaluated
in run-time partition pruning, do it once during run-time pruning
set-up and cache the exprstate in PartitionPruneContext, saving a lot of
work.
Author: David Rowley
Reviewed-by: Amit Langote, Álvaro Herrera
Discussion: https://postgr.es/m/CAKJS1f8-x+q-90QAPDu_okhQBV4DPEtPz8CJ=m0940GyT4DA4w@mail.gmail.com
I added this "optimization" on top of Amit Langote's 158b7bc6d7, but
the quick path is never taken because the partition uses a different
pg_type oid than its parent table (causing equalTupleDescs to return
false). Changing that requires more analysis and is too considered
dangerous at this point in the cycle, so revert it.
We might make it work someday, but not for pg11.
Discussion: https://postgr.es/m/825031be-942c-8c24-6163-13c27f217a3d@lab.ntt.co.jp
It turns out that after runtime partition pruning, Append's
first_partial_plan does not accurately represent partial plans to run,
if any of those got pruned. This could limit participation of workers
in some partial subplans, if other subplans got pruned. Fix it by
keeping an index of the first valid partial subplan in the state node,
determined at execnode Init time.
Author: David Rowley, with cosmetic changes by me.
Discussion: https://postgr.es/m/CAKJS1f8o2Yd=rOP=Et3A0FWgF+gSAOkFSU6eNhnGzTPV7nN8sQ@mail.gmail.com
We had an Assert() preventing whole-row expressions from being used in
the SET clause of INSERT ON CONFLICT, but it seems unnecessary, given
some tests, so remove it. Add a new test to exercise the case.
Still at ExecInitPartitionInfo, we used map_partition_varattnos (which
constructs an attribute map, then calls map_variable_attnos) using
the same two relations many times in different expressions and with
different parameters. Constructing the map over and over is a waste.
To avoid this repeated work, construct the map once, and use
map_variable_attnos() directly instead.
Author: Amit Langote, per comments by me (Álvaro)
Discussion: https://postgr.es/m/20180326142016.m4st5e34chrzrknk@alvherre.pgsql
This reverts commits d204ef6377,
83454e3c2b and a few more commits thereafter
(complete list at the end) related to MERGE feature.
While the feature was fully functional, with sufficient test coverage and
necessary documentation, it was felt that some parts of the executor and
parse-analyzer can use a different design and it wasn't possible to do that in
the available time. So it was decided to revert the patch for PG11 and retry
again in the future.
Thanks again to all reviewers and bug reporters.
List of commits reverted, in reverse chronological order:
f1464c5380 Improve parse representation for MERGE
ddb4158579 MERGE syntax diagram correction
530e69e59b Allow cpluspluscheck to pass by renaming variable
01b88b4df5 MERGE minor errata
3af7b2b0d4 MERGE fix variable warning in non-assert builds
a5d86181ec MERGE INSERT allows only one VALUES clause
4b2d44031f MERGE post-commit review
4923550c20 Tab completion for MERGE
aa3faa3c7a WITH support in MERGE
83454e3c2b New files for MERGE
d204ef6377 MERGE SQL Command following SQL:2016
Author: Pavan Deolasee
Reviewed-by: Michael Paquier
The HeapFetches counter was using a simple value in IndexOnlyScanState,
which fails to propagate values from parallel workers; so the counts are
wrong when IndexOnlyScan runs in parallel. Move it to Instrumentation,
like all the other counters.
While at it, change INSERT ON CONFLICT conflicting tuple counter to use
the new ntuples2 instead of nfiltered2, which is a blatant misuse.
Discussion: https://postgr.es/m/20180409215851.idwc75ct2bzi6tea@alvherre.pgsql
In 499be013de support for pruning unneeded Append subnodes was added.
The logic in that commit was not correctly checking if the next subplan
was in fact a valid subplan. This could cause parallel workers processes
to be given a subplan to work on which didn't require any work.
Per code review following an otherwise unexplained regression failure in
buildfarm member Pademelon. (We haven't been able to reproduce the
failure, so this is a bit of a blind fix in terms of whether it'll
actually fix it; but it is a clear bug nonetheless).
In passing, also add a comment to explain what first_partial_plan means.
Author: David Rowley
Discussion: https://postgr.es/m/CAKJS1f_E5r05hHUVG3UmCQJ49DGKKHtN=SHybD44LdzBn+CJng@mail.gmail.com
Traditionally, include/catalog/pg_foo.h contains extern declarations
for functions in backend/catalog/pg_foo.c, in addition to its function
as the authoritative definition of the pg_foo catalog's rowtype.
In some cases, we'd been forced to split out those extern declarations
into separate pg_foo_fn.h headers so that the catalog definitions
could be #include'd by frontend code. That problem is gone as of
commit 9c0a0de4c, so let's undo the splits to make things less
confusing.
Discussion: https://postgr.es/m/23690.1523031777@sss.pgh.pa.us
Existing partition pruning is only able to work at plan time, for query
quals that appear in the parsed query. This is good but limiting, as
there can be parameters that appear later that can be usefully used to
further prune partitions.
This commit adds support for pruning subnodes of Append which cannot
possibly contain any matching tuples, during execution, by evaluating
Params to determine the minimum set of subnodes that can possibly match.
We support more than just simple Params in WHERE clauses. Support
additionally includes:
1. Parameterized Nested Loop Joins: The parameter from the outer side of the
join can be used to determine the minimum set of inner side partitions to
scan.
2. Initplans: Once an initplan has been executed we can then determine which
partitions match the value from the initplan.
Partition pruning is performed in two ways. When Params external to the plan
are found to match the partition key we attempt to prune away unneeded Append
subplans during the initialization of the executor. This allows us to bypass
the initialization of non-matching subplans meaning they won't appear in the
EXPLAIN or EXPLAIN ANALYZE output.
For parameters whose value is only known during the actual execution
then the pruning of these subplans must wait. Subplans which are
eliminated during this stage of pruning are still visible in the EXPLAIN
output. In order to determine if pruning has actually taken place, the
EXPLAIN ANALYZE must be viewed. If a certain Append subplan was never
executed due to the elimination of the partition then the execution
timing area will state "(never executed)". Whereas, if, for example in
the case of parameterized nested loops, the number of loops stated in
the EXPLAIN ANALYZE output for certain subplans may appear lower than
others due to the subplan having been scanned fewer times. This is due
to the list of matching subnodes having to be evaluated whenever a
parameter which was found to match the partition key changes.
This commit required some additional infrastructure that permits the
building of a data structure which is able to perform the translation of
the matching partition IDs, as returned by get_matching_partitions, into
the list index of a subpaths list, as exist in node types such as
Append, MergeAppend and ModifyTable. This allows us to translate a list
of clauses into a Bitmapset of all the subpath indexes which must be
included to satisfy the clause list.
Author: David Rowley, based on an earlier effort by Beena Emerson
Reviewers: Amit Langote, Robert Haas, Amul Sul, Rajkumar Raghuwanshi,
Jesper Pedersen
Discussion: https://postgr.es/m/CAOG9ApE16ac-_VVZVvv0gePSgkg_BwYEV1NBqZFqDR2bBE0X0A@mail.gmail.com
When an update moves a row between partitions (supported since
2f17844104), our normal logic for following update chains in READ
COMMITTED mode doesn't work anymore. Cross partition updates are
modeled as an delete from the old and insert into the new
partition. No ctid chain exists across partitions, and there's no
convenient space to introduce that link.
Not throwing an error in a partitioned context when one would have
been thrown without partitioning is obviously problematic. This commit
introduces infrastructure to detect when a tuple has been moved, not
just plainly deleted. That allows to throw an error when encountering
a deletion that's actually a move, while attempting to following a
ctid chain.
The row deleted as part of a cross partition update is marked by
pointing it's t_ctid to an invalid block, instead of self as a normal
update would. That was deemed to be the least invasive and most
future proof way to represent the knowledge, given how few infomask
bits are there to be recycled (there's also some locking issues with
using infomask bits).
External code following ctid chains should be updated to check for
moved tuples. The most likely consequence of not doing so is a missed
error.
Author: Amul Sul, editorialized by me
Reviewed-By: Amit Kapila, Pavan Deolasee, Andres Freund, Robert Haas
Discussion: http://postgr.es/m/CAAJ_b95PkwojoYfz0bzXU8OokcTVGzN6vYGCNVUukeUDrnF3dw@mail.gmail.com
This patch introduces INCLUDE clause to index definition. This clause
specifies a list of columns which will be included as a non-key part in
the index. The INCLUDE columns exist solely to allow more queries to
benefit from index-only scans. Also, such columns don't need to have
appropriate operator classes. Expressions are not supported as INCLUDE
columns since they cannot be used in index-only scans.
Index access methods supporting INCLUDE are indicated by amcaninclude flag
in IndexAmRoutine. For now, only B-tree indexes support INCLUDE clause.
In B-tree indexes INCLUDE columns are truncated from pivot index tuples
(tuples located in non-leaf pages and high keys). Therefore, B-tree indexes
now might have variable number of attributes. This patch also provides
generic facility to support that: pivot tuples contain number of their
attributes in t_tid.ip_posid. Free 13th bit of t_info is used for indicating
that. This facility will simplify further support of index suffix truncation.
The changes of above are backward-compatible, pg_upgrade doesn't need special
handling of B-tree indexes for that.
Bump catalog version
Author: Anastasia Lubennikova with contribition by Alexander Korotkov and me
Reviewed by: Peter Geoghegan, Tomas Vondra, Antonin Houska, Jeff Janes,
David Rowley, Alexander Korotkov
Discussion: https://www.postgresql.org/message-id/flat/56168952.4010101@postgrespro.ru
Also enable this for postgres_fdw.
Etsuro Fujita, based on an earlier patch by Amit Langote. The larger
patch series of which this is a part has been reviewed by Amit
Langote, David Fetter, Maksim Milyutin, Álvaro Herrera, Stephen Frost,
and me. Minor documentation changes to the final version by me.
Discussion: http://postgr.es/m/29906a26-da12-8c86-4fb9-d8f88442f2b9@lab.ntt.co.jp
Review comments from Andres Freund
* Consolidate code into AfterTriggerGetTransitionTable()
* Rename nodeMerge.c to execMerge.c
* Rename nodeMerge.h to execMerge.h
* Move MERGE handling in ExecInitModifyTable()
into a execMerge.c ExecInitMerge()
* Move mt_merge_subcommands flags into execMerge.h
* Rename opt_and_condition to opt_merge_when_and_condition
* Wordsmith various comments
Author: Pavan Deolasee
Reviewer: Simon Riggs
MERGE performs actions that modify rows in the target table
using a source table or query. MERGE provides a single SQL
statement that can conditionally INSERT/UPDATE/DELETE rows
a task that would other require multiple PL statements.
e.g.
MERGE INTO target AS t
USING source AS s
ON t.tid = s.sid
WHEN MATCHED AND t.balance > s.delta THEN
UPDATE SET balance = t.balance - s.delta
WHEN MATCHED THEN
DELETE
WHEN NOT MATCHED AND s.delta > 0 THEN
INSERT VALUES (s.sid, s.delta)
WHEN NOT MATCHED THEN
DO NOTHING;
MERGE works with regular and partitioned tables, including
column and row security enforcement, as well as support for
row, statement and transition triggers.
MERGE is optimized for OLTP and is parameterizable, though
also useful for large scale ETL/ELT. MERGE is not intended
to be used in preference to existing single SQL commands
for INSERT, UPDATE or DELETE since there is some overhead.
MERGE can be used statically from PL/pgSQL.
MERGE does not yet support inheritance, write rules,
RETURNING clauses, updatable views or foreign tables.
MERGE follows SQL Standard per the most recent SQL:2016.
Includes full tests and documentation, including full
isolation tests to demonstrate the concurrent behavior.
This version written from scratch in 2017 by Simon Riggs,
using docs and tests originally written in 2009. Later work
from Pavan Deolasee has been both complex and deep, leaving
the lead author credit now in his hands.
Extensive discussion of concurrency from Peter Geoghegan,
with thanks for the time and effort contributed.
Various issues reported via sqlsmith by Andreas Seltenreich
Authors: Pavan Deolasee, Simon Riggs
Reviewer: Peter Geoghegan, Amit Langote, Tomas Vondra, Simon Riggs
Discussion:
https://postgr.es/m/CANP8+jKitBSrB7oTgT9CY2i1ObfOt36z0XMraQc+Xrz8QB0nXA@mail.gmail.comhttps://postgr.es/m/CAH2-WzkJdBuxj9PO=2QaO9-3h3xGbQPZ34kJH=HukRekwM-GZg@mail.gmail.com
MERGE performs actions that modify rows in the target table
using a source table or query. MERGE provides a single SQL
statement that can conditionally INSERT/UPDATE/DELETE rows
a task that would other require multiple PL statements.
e.g.
MERGE INTO target AS t
USING source AS s
ON t.tid = s.sid
WHEN MATCHED AND t.balance > s.delta THEN
UPDATE SET balance = t.balance - s.delta
WHEN MATCHED THEN
DELETE
WHEN NOT MATCHED AND s.delta > 0 THEN
INSERT VALUES (s.sid, s.delta)
WHEN NOT MATCHED THEN
DO NOTHING;
MERGE works with regular and partitioned tables, including
column and row security enforcement, as well as support for
row, statement and transition triggers.
MERGE is optimized for OLTP and is parameterizable, though
also useful for large scale ETL/ELT. MERGE is not intended
to be used in preference to existing single SQL commands
for INSERT, UPDATE or DELETE since there is some overhead.
MERGE can be used statically from PL/pgSQL.
MERGE does not yet support inheritance, write rules,
RETURNING clauses, updatable views or foreign tables.
MERGE follows SQL Standard per the most recent SQL:2016.
Includes full tests and documentation, including full
isolation tests to demonstrate the concurrent behavior.
This version written from scratch in 2017 by Simon Riggs,
using docs and tests originally written in 2009. Later work
from Pavan Deolasee has been both complex and deep, leaving
the lead author credit now in his hands.
Extensive discussion of concurrency from Peter Geoghegan,
with thanks for the time and effort contributed.
Various issues reported via sqlsmith by Andreas Seltenreich
Authors: Pavan Deolasee, Simon Riggs
Reviewers: Peter Geoghegan, Amit Langote, Tomas Vondra, Simon Riggs
Discussion:
https://postgr.es/m/CANP8+jKitBSrB7oTgT9CY2i1ObfOt36z0XMraQc+Xrz8QB0nXA@mail.gmail.comhttps://postgr.es/m/CAH2-WzkJdBuxj9PO=2QaO9-3h3xGbQPZ34kJH=HukRekwM-GZg@mail.gmail.com
So far, a nested CALL or DO in PL/pgSQL would not establish a context
where transaction control statements were allowed. This fixes that by
handling CALL and DO specially in PL/pgSQL, passing the atomic/nonatomic
execution context through and doing the required management around
transaction boundaries.
Reviewed-by: Tomas Vondra <tomas.vondra@2ndquadrant.com>
Currently adding a column to a table with a non-NULL default results in
a rewrite of the table. For large tables this can be both expensive and
disruptive. This patch removes the need for the rewrite as long as the
default value is not volatile. The default expression is evaluated at
the time of the ALTER TABLE and the result stored in a new column
(attmissingval) in pg_attribute, and a new column (atthasmissing) is set
to true. Any existing row when fetched will be supplied with the
attmissingval. New rows will have the supplied value or the default and
so will never need the attmissingval.
Any time the table is rewritten all the atthasmissing and attmissingval
settings for the attributes are cleared, as they are no longer needed.
The most visible code change from this is in heap_attisnull, which
acquires a third TupleDesc argument, allowing it to detect a missing
value if there is one. In many cases where it is known that there will
not be any (e.g. catalog relations) NULL can be passed for this
argument.
Andrew Dunstan, heavily modified from an original patch from Serge
Rielau.
Reviewed by Tom Lane, Andres Freund, Tomas Vondra and David Rowley.
Discussion: https://postgr.es/m/31e2e921-7002-4c27-59f5-51f08404c858@2ndQuadrant.com
Originally, we treated memory context names as potentially variable in
all cases, and therefore always copied them into the context header.
Commit 9fa6f00b1 rethought this a little bit and invented a distinction
between fixed and variable names, skipping the copy step for the former.
But we can make things both simpler and more useful by instead allowing
there to be two parts to a context's identification, a fixed "name" and
an optional, variable "ident". The name supplied in the context create
call is now required to be a compile-time-constant string in all cases,
as it is never copied but just pointed to. The "ident" string, if
wanted, is supplied later. This is needed because typically we want
the ident to be stored inside the context so that it's cleaned up
automatically on context deletion; that means it has to be copied into
the context before we can set the pointer.
The cost of this approach is basically just an additional pointer field
in struct MemoryContextData, which isn't much overhead, and is bought
back entirely in the AllocSet case by not needing a headerSize field
anymore, since we no longer have to cope with variable header length.
In addition, we can simplify the internal interfaces for memory context
creation still further, saving a few cycles there. And it's no longer
true that a custom identifier disqualifies a context from participating
in aset.c's freelist scheme, so possibly there's some win on that end.
All the places that were using non-compile-time-constant context names
are adjusted to put the variable info into the "ident" instead. This
allows more effective identification of those contexts in many cases;
for example, subsidary contexts of relcache entries are now identified
by both type (e.g. "index info") and relname, where before you got only
one or the other. Contexts associated with PL function cache entries
are now identified more fully and uniformly, too.
I also arranged for plancache contexts to use the query source string
as their identifier. This is basically free for CachedPlanSources, as
they contained a copy of that string already. We pay an extra pstrdup
to do it for CachedPlans. That could perhaps be avoided, but it would
make things more fragile (since the CachedPlanSource is sometimes
destroyed first). I suspect future improvements in error reporting will
require CachedPlans to have a copy of that string anyway, so it's not
clear that it's worth moving mountains to avoid it now.
This also changes the APIs for context statistics routines so that the
context-specific routines no longer assume that output goes straight
to stderr, nor do they know all details of the output format. This
is useful immediately to reduce code duplication, and it also allows
for external code to do something with stats output that's different
from printing to stderr.
The reason for pushing this now rather than waiting for v12 is that
it rethinks some of the API changes made by commit 9fa6f00b1. Seems
better for extension authors to endure just one round of API changes
not two.
Discussion: https://postgr.es/m/CAB=Je-FdtmFZ9y9REHD7VsSrnCkiBhsA4mdsLKSPauwXtQBeNA@mail.gmail.com
Performing JIT compilation for deforming gains performance benefits
over unJITed deforming from compile-time knowledge of the tuple
descriptor. Fixed column widths, NOT NULLness, etc can be taken
advantage of.
Right now the JITed deforming is only used when deforming tuples as
part of expression evaluation (and obviously only if the descriptor is
known). It's likely to be beneficial in other cases, too.
By default tuple deforming is JITed whenever an expression is JIT
compiled. There's a separate boolean GUC controlling it, but that's
expected to be primarily useful for development and benchmarking.
Docs will follow in a later commit containing docs for the whole JIT
feature.
Author: Andres Freund
Discussion: https://postgr.es/m/20170901064131.tazjxwus3k2w3ybh@alap3.anarazel.de
Commit eb7ed3f306 enabled unique constraints on partitioned tables,
but one thing that was not working properly is INSERT/ON CONFLICT.
This commit introduces a new node keeps state related to the ON CONFLICT
clause per partition, and fills it when that partition is about to be
used for tuple routing.
Author: Amit Langote, Álvaro Herrera
Reviewed-by: Etsuro Fujita, Pavan Deolasee
Discussion: https://postgr.es/m/20180228004602.cwdyralmg5ejdqkq@alvherre.pgsql
Coverity complained that this check is pointless, and it's right.
There is no case where we'd call ExecutorStart with a null plannedstmt,
and if we did, it'd have crashed before here. Thinko in commit cc415a56d.
In addition to the interpretation of expressions (which back
evaluation of WHERE clauses, target list projection, aggregates
transition values etc) support compiling expressions to native code,
using the infrastructure added in earlier commits.
To avoid duplicating a lot of code, only support emitting code for
cases that are likely to be performance critical. For expression steps
that aren't deemed that, use the existing interpreter.
The generated code isn't great - some architectural changes are
required to address that. But this already yields a significant
speedup for some analytics queries, particularly with WHERE clauses
filtering a lot, or computing multiple aggregates.
Author: Andres Freund
Tested-By: Thomas Munro
Discussion: https://postgr.es/m/20170901064131.tazjxwus3k2w3ybh@alap3.anarazel.de
Disable JITing for VALUES() nodes.
VALUES() nodes are only ever executed once. This is primarily helpful
for debugging, when forcing JITing even for cheap queries.
Author: Andres Freund
Discussion: https://postgr.es/m/20170901064131.tazjxwus3k2w3ybh@alap3.anarazel.de
This adds simple cost based plan time decision about whether JIT
should be performed. jit_above_cost, jit_optimize_above_cost are
compared with the total cost of a plan, and if the cost is above them
JIT is performed / optimization is performed respectively.
For that PlannedStmt and EState have a jitFlags (es_jit_flags) field
that stores information about what JIT operations should be performed.
EState now also has a new es_jit field, which can store a
JitContext. When there are no errors the context is released in
standard_ExecutorEnd().
It is likely that the default values for jit_[optimize_]above_cost
will need to be adapted further, but in my test these values seem to
work reasonably.
Author: Andres Freund, with feedback by Peter Eisentraut
Discussion: https://postgr.es/m/20170901064131.tazjxwus3k2w3ybh@alap3.anarazel.de
These values can be obtained from the ModifyTable node which is already
a part of both the ModifyTableState and ExecInsert.
Author: Álvaro Herrera, Amit Langote
Reviewed-by: Peter Geoghegan
Discussion: https://postgr.es/m/20180316151303.rml2p5wffn3o6qy6@alvherre.pgsql
We make some changes to ModifyTableState and the EState it uses whenever
we route tuples to partitions; but we weren't restoring properly in all
cases, possibly causing crashes when partitions with different tuple
descriptors are targeted by tuples inserted in the same command.
Refactor some code, creating ExecPrepareTupleRouting, to encapsulate the
needed state changing logic, and have it invoked one level above its
current place (ie. put it in ExecModifyTable instead of ExecInsert);
this makes it all more readable.
Add a test case to exercise this.
We don't support having views as partitions; and since only views can
have INSTEAD OF triggers, there is no point in testing for INSTEAD OF
when processing insertions into a partitioned table. Remove code that
appears to support this (but which is actually never relevant.)
In passing, fix location of some very confusing comments in
ModifyTableState.
Reported-by: Amit Langote
Author: Etsuro Fujita, Amit Langote
Discussion: https://postgr/es/m/0473bf5c-57b1-f1f7-3d58-455c2230bc5f@lab.ntt.co.jp
"UPDATE/DELETE WHERE CURRENT OF cursor_name" failed, with an error message
like "cannot extract system attribute from virtual tuple", if the cursor
was using a index-only scan for the target table. Fix it by digging the
current TID out of the indexscan state.
It seems likely that the same failure could occur for CustomScan plans
and perhaps some FDW plan types, so that leaving this to be treated as an
internal error with an obscure message isn't as good an idea as it first
seemed. Hence, add a bit of heaptuple.c infrastructure to let us deliver
a more on-topic message. I chose to make the message match what you get
for the case where execCurrentOf can't identify the target scan node at
all, "cursor "foo" is not a simply updatable scan of table "bar"".
Perhaps it should be different, but we can always adjust that later.
In the future, it might be nice to provide hooks that would let custom
scan providers and/or FDWs deal with this in other ways; but that's
not a suitable topic for a back-patchable bug fix.
It's been like this all along, so back-patch to all supported branches.
Yugo Nagata and Tom Lane
Discussion: https://postgr.es/m/20180201013349.937dfc5f.nagata@sraoss.co.jp
ExecHashTableCreate allocated some memory that wasn't freed by
ExecHashTableDestroy, specifically the per-hash-key function information.
That's not a huge amount of data, but if one runs a query that repeats
a hash join enough times, it builds up. Fix by arranging for the data
in question to be kept in the hashtable's hashCxt instead of leaving it
"loose" in the query-lifespan executor context. (This ensures that we'll
also clean up anything that the hash functions allocate in fn_mcxt.)
Per report from Amit Khandekar. It's been like this forever, so back-patch
to all supported branches.
Discussion: https://postgr.es/m/CAJ3gD9cFofAWGvcxLOxDHC=B0hjtW8yGmUsF2hdGh97CM38=7g@mail.gmail.com
In a top-level CALL, the values of INOUT arguments will be returned as a
result row. In PL/pgSQL, the values are assigned back to the input
arguments. In other languages, the same convention as for return a
record from a function is used. That does not require any code changes
in the PL implementations.
Reviewed-by: Pavel Stehule <pavel.stehule@gmail.com>
Since edd44738bc WCO expressions of partitioned tables are
initialized with the first subplan as parent. That's not correct, as
the correct context is the ModifyTableState node. That's also what is
used for RETURNING processing, initialized nearby.
This appears not to cause any visible problems for in core code, but
is problematic for in development patch.
Discussion: https://postgr.es/m/20180303043818.tnvlo243bgy7una3@alap3.anarazel.de
The new column distinguishes normal functions, procedures, aggregates,
and window functions. This replaces the existing columns proisagg and
proiswindow, and replaces the convention that procedures are indicated
by prorettype == 0. Also change prorettype to be VOIDOID for procedures.
Reviewed-by: Tom Lane <tgl@sss.pgh.pa.us>
Reviewed-by: Michael Paquier <michael@paquier.xyz>
Parallel-aware plan nodes must be prepared to run without parallelism
if it's not possible at execution time for whatever reason. Commit
ab72716778, which introduced Parallel
Append, overlooked this.
Rajkumar Raghuwanshi reported this problem, and I included his test
case in this patch. The code changes are by me.
Discussion: http://postgr.es/m/CAKcux6=WqkUudLg1GLZZ7fc5ScWC1+Y9qD=pAHeqy32WoeJQvw@mail.gmail.com
Tom Kazimiers reported that transition tables don't work correctly when
they are scanned by more than one executor node. That's because commit
18ce3a4ab allocated separate read pointers for each executor node, as it
must, but failed to make them active at the appropriate times. Repair.
Thomas Munro
Discussion: https://postgr.es/m/20180224034748.bixarv6632vbxgeb%40dewberry.localdomain
It's not necessary to fully initialize the executor data structures
for partitions to which no tuples are ever routed. Consider, for
example, an INSERT statement that inserts only one row: it only cares
about the partition to which that one row is routed. The new function
ExecInitPartitionInfo performs the initialization in question only
when a particular partition is about to receive a tuple. This includes
creating, validating, and saving a pointer to the ResultRelInfo,
setting up for speculative insertions, translating WCOs and
initializing the resulting expressions, translating returning lists
and building the appropriate projection information, and setting up a
tuple conversion map.
One thing that's not deferred is locking the child partitions; that
seems desirable but would need more thought. Still, testing shows
that this makes single-row inserts significantly faster on a table
with many partitions without harming the bulk-insert case.
Amit Langote, reviewed by Etsuro Fujita, with a few changes by me
Discussion: http://postgr.es/m/8975331d-d961-cbdd-f862-fdd3d97dc2d0@lab.ntt.co.jp
An updating query that reads a CTE within an InitPlan or SubPlan could get
incorrect results if it updates rows that are concurrently being modified.
This is caused by CteScanNext supposing that nothing inside its recursive
ExecProcNode call could change which read pointer is selected in the CTE's
shared tuplestore. While that's normally true because of scoping
considerations, it can break down if an EPQ plan tree gets built during the
call, because EvalPlanQualStart builds execution trees for all subplans
whether they're going to be used during the recheck or not. And it seems
like a pretty shaky assumption anyway, so let's just reselect our own read
pointer here.
Per bug #14870 from Andrei Gorita. This has been broken since CTEs were
implemented, so back-patch to all supported branches.
Discussion: https://postgr.es/m/20171024155358.1471.82377@wrigleys.postgresql.org
In what was doubtless a typo, commit bf6c614a2 introduced a duplicate
initialization of a local variable. This made Coverity unhappy, as well
as pretty much anybody reading the code. We don't even have a real use
for the local variable, so just remove it.
The reason for doing so is that it will allow expression evaluation to
optimize based on the underlying tupledesc. In particular it will
allow to JIT tuple deforming together with the expression itself.
For that expression initialization needs to be moved after the
relevant slots are initialized - mostly unproblematic, except in the
case of nodeWorktablescan.c.
After doing so there's no need for ExecAssignResultType() and
ExecAssignResultTypeFromTL() anymore, as all former callers have been
converted to create a slot with a fixed descriptor.
When creating a slot with a fixed descriptor, tts_values/isnull can be
allocated together with the main slot, reducing allocation overhead
and increasing cache density a bit.
Author: Andres Freund
Discussion: https://postgr.es/m/20171206093717.vqdxe5icqttpxs3p@alap3.anarazel.de
This has a performance benefit on own, although not hugely so. The
primary benefit is that it will allow for to JIT tuple deforming and
comparator invocations.
Large parts of this were previously committed (773aec7aa), but the
commit contained an omission around cross-type comparisons and was
thus reverted.
Author: Andres Freund
Discussion: https://postgr.es/m/20171129080934.amqqkke2zjtekd4t@alap3.anarazel.de
This reverts commit 773aec7aa9.
There's an unresolved issue in the reverted commit: It only creates
one comparator function, but in for the nodeSubplan.c case we need
more (c.f. FindTupleHashEntry vs LookupTupleHashEntry calls in
nodeSubplan.c).
This isn't too difficult to fix, but it's not entirely trivial
either. The fact that the issue only causes breakage on 32bit systems
shows that the current test coverage isn't that great. To avoid
turning half the buildfarm red till those two issues are addressed,
revert.
Formerly, DTYPE_REC was used only for variables declared as "record";
variables of named composite types used DTYPE_ROW, which is faster for
some purposes but much less flexible. In particular, the ROW code paths
are entirely incapable of dealing with DDL-caused changes to the number
or data types of the columns of a row variable, once a particular plpgsql
function has been parsed for the first time in a session. And, since the
stored representation of a ROW isn't a tuple, there wasn't any easy way
to deal with variables of domain-over-composite types, since the domain
constraint checking code would expect the value to be checked to be a
tuple. A lesser, but still real, annoyance is that ROW format cannot
represent a true NULL composite value, only a row of per-field NULL
values, which is not exactly the same thing.
Hence, switch to using DTYPE_REC for all composite-typed variables,
whether "record", named composite type, or domain over named composite
type. DTYPE_ROW remains but is used only for its native purpose, to
represent a fixed-at-compile-time list of variables, for instance the
targets of an INTO clause.
To accomplish this without taking significant performance losses, introduce
infrastructure that allows storing composite-type variables as "expanded
objects", similar to the "expanded array" infrastructure introduced in
commit 1dc5ebc90. A composite variable's value is thereby kept (most of
the time) in the form of separate Datums, so that field accesses and
updates are not much more expensive than they were in the ROW format.
This holds the line, more or less, on performance of variables of named
composite types in field-access-intensive microbenchmarks, and makes
variables declared "record" perform much better than before in similar
tests. In addition, the logic involved with enforcing composite-domain
constraints against updates of individual fields is in the expanded
record infrastructure not plpgsql proper, so that it might be reusable
for other purposes.
In further support of this, introduce a typcache feature for assigning a
unique-within-process identifier to each distinct tuple descriptor of
interest; in particular, DDL alterations on composite types result in a new
identifier for that type. This allows very cheap detection of the need to
refresh tupdesc-dependent data. This improves on the "tupDescSeqNo" idea
I had in commit 687f096ea: that assigned identifying sequence numbers to
successive versions of individual composite types, but the numbers were not
unique across different types, nor was there support for assigning numbers
to registered record types.
In passing, allow plpgsql functions to accept as well as return type
"record". There was no good reason for the old restriction, and it
was out of step with most of the other PLs.
Tom Lane, reviewed by Pavel Stehule
Discussion: https://postgr.es/m/8962.1514399547@sss.pgh.pa.us
Doing so causes EXPLAIN ANALYZE to show trigger statistics multiple
times. Commit 2f17844104 seems to
be to blame for this.
Amit Langote, revieed by Amit Khandekar, Etsuro Fujita, and me.
When the previously-chosen plan was non-partial, all pa_finished
flags for partial plans are now set, and pa_next_plan has not yet
been set to INVALID_SUBPLAN_INDEX, the previous code could go into
an infinite loop.
Report by Rajkumar Raghuwanshi. Patch by Amit Khandekar and me.
Review by Kyotaro Horiguchi.
Discussion: http://postgr.es/m/CAJ3gD9cf43z78qY=U=H0HvOEN341qfRO-vLpnKPSviHeWgJQ5w@mail.gmail.com
This patch adds the ability to use "RANGE offset PRECEDING/FOLLOWING"
frame boundaries in window functions. We'd punted on that back in the
original patch to add window functions, because it was not clear how to
do it in a reasonably data-type-extensible fashion. That problem is
resolved here by adding the ability for btree operator classes to provide
an "in_range" support function that defines how to add or subtract the
RANGE offset value. Factoring it this way also allows the operator class
to avoid overflow problems near the ends of the datatype's range, if it
wishes to expend effort on that. (In the committed patch, the integer
opclasses handle that issue, but it did not seem worth the trouble to
avoid overflow failures for datetime types.)
The patch includes in_range support for the integer_ops opfamily
(int2/int4/int8) as well as the standard datetime types. Support for
other numeric types has been requested, but that seems like suitable
material for a follow-on patch.
In addition, the patch adds GROUPS mode which counts the offset in
ORDER-BY peer groups rather than rows, and it adds the frame_exclusion
options specified by SQL:2011. As far as I can see, we are now fully
up to spec on window framing options.
Existing behaviors remain unchanged, except that I changed the errcode
for a couple of existing error reports to meet the SQL spec's expectation
that negative "offset" values should be reported as SQLSTATE 22013.
Internally and in relevant parts of the documentation, we now consistently
use the terminology "offset PRECEDING/FOLLOWING" rather than "value
PRECEDING/FOLLOWING", since the term "value" is confusingly vague.
Oliver Ford, reviewed and whacked around some by me
Discussion: https://postgr.es/m/CAGMVOdu9sivPAxbNN0X+q19Sfv9edEPv=HibOJhB14TJv_RCQg@mail.gmail.com
We don't need to set up the shared space for hash join instrumentation data
if instrumentation hasn't been requested. Let's follow the example of the
similar Sort node code and save a few cycles by skipping that when we can.
This reverts commit d59ff4ab3 and instead allows us to use the safer choice
of passing noError = false to shm_toc_lookup in ExecHashInitializeWorker,
since if we reach that call there should be a TOC entry to be found.
Thomas Munro
Discussion: https://postgr.es/m/E1ehkoZ-0005uW-43%40gemulon.postgresql.org
One or another author of commit 5bcf389ec seems to have thought that
computing an offset from a NULL pointer would yield another NULL pointer.
There may possibly be architectures where that works, but common machines
don't work like that. Per a quick code review of places calling
shm_toc_lookup and not using noError = false.
To make this work, tuplesort.c and logtape.c must also support
parallelism, so this patch adds that infrastructure and then applies
it to the particular case of parallel btree index builds. Testing
to date shows that this can often be 2-3x faster than a serial
index build.
The model for deciding how many workers to use is fairly primitive
at present, but it's better than not having the feature. We can
refine it as we get more experience.
Peter Geoghegan with some help from Rushabh Lathia. While Heikki
Linnakangas is not an author of this patch, he wrote other patches
without which this feature would not have been possible, and
therefore the release notes should possibly credit him as an author
of this feature. Reviewed by Claudio Freire, Heikki Linnakangas,
Thomas Munro, Tels, Amit Kapila, me.
Discussion: http://postgr.es/m/CAM3SWZQKM=Pzc=CAHzRixKjp2eO5Q0Jg1SoFQqeXFQ647JiwqQ@mail.gmail.com
Discussion: http://postgr.es/m/CAH2-Wz=AxWqDoVvGU7dq856S4r6sJAj6DBn7VMtigkB33N5eyg@mail.gmail.com
Once this function has been called, we know that all workers have
started and attached to their error queues -- so if any of them
subsequently exit uncleanly, we'll be sure to throw an ERROR promptly.
Otherwise, users of the ParallelContext machinery must be careful not
to wait forever for a worker that has failed to start. Parallel query
manages to work without needing this for reasons explained in new
comments added by this patch, but it's a useful primitive for other
parallel operations, such as the pending patch to make creating a
btree index run in parallel.
Amit Kapila, revised by me. Additional review by Peter Geoghegan.
Discussion: http://postgr.es/m/CAA4eK1+e2MzyouF5bg=OtyhDSX+=Ao=3htN=T-r_6s3gCtKFiw@mail.gmail.com
It's a common task to evaluate a qual and reset the corresponding
expression context. Currently that requires storing the result of the
qual eval, resetting the context, and then reacting on the result. As
that's awkward several places only reset the context next time through
a node. That's not great, so introduce a helper that evaluates and
resets.
It's a bit ugly that it currently uses MemoryContextReset() instead of
ResetExprContext(), but that seems easier than reordering all of
executor.h.
Author: Andres Freund
Discussion: https://postgr.es/m/20180109222544.f7loxrunqh3xjl5f@alap3.anarazel.de
ExecPushExprSlots didn't initialize ExprEvalStep's resvalue/resnull
steps as it didn't use them. That caused wrong valgrind warnings for
an upcoming patch, so zero-intialize.
Also zero-initialize all scratch ExprEvalStep's allocated on the
stack, to avoid issues with similar future omissions of non-critial
data.
The changes in b81b5a96f4 did not fully
address the issue, because the bit-mixing of the IV into the final
hash-key didn't prevent clustering in the input-data survive in the
output data.
This didn't cause a lot of problems because of the additional growth
conditions added d4c62a6b62. But as we
want to rein those in due to explosive growth in some edges, this
needs to be fixed.
Author: Andres Freund
Discussion: https://postgr.es/m/20171127185700.1470.20362@wrigleys.postgresql.org
Backpatch: 10, where simplehash was introduced
Commit 09529a70b changed nodeIndexscan.c and nodeIndexonlyscan.c to
postpone initialization of the indexscan proper until the first tuple
fetch. It overlooked the question of mark/restore behavior, which means
that if some caller attempts to mark the scan before the first tuple fetch,
you get a null pointer dereference.
The only existing user of mark/restore is nodeMergejoin.c, which (somewhat
accidentally) will never attempt to set a mark before the first inner tuple
unless the inner child node is a Material node. Hence the case can't arise
normally, so it seems sufficient to document the assumption at both ends.
However, during an EvalPlanQual recheck, ExecScanFetch doesn't call
IndexNext but just returns the jammed-in test tuple. Therefore, if we're
doing a recheck in a plan tree with a mergejoin with inner indexscan,
it's possible to reach ExecIndexMarkPos with iss_ScanDesc still null,
as reported by Guo Xiang Tan in bug #15032.
Really, when there's a test tuple supplied during an EPQ recheck, touching
the index at all is the wrong thing: rather, the behavior of mark/restore
ought to amount to saving and restoring the es_epqScanDone flag. We can
avoid finding a place to actually save the flag, for the moment, because
given the assumption that no caller will set a mark before fetching a
tuple, es_epqScanDone must always be set by the time we try to mark.
So the actual behavior change required is just to not reach the index
access if a test tuple is supplied.
The set of plan node types that need to consider this issue are those
that support EPQ test tuples (i.e., call ExecScan()) and also support
mark/restore; which is to say, IndexScan, IndexOnlyScan, and perhaps
CustomScan. It's tempting to try to fix the problem in one place by
teaching ExecMarkPos() itself about EPQ; but ExecMarkPos supports some
plan types that aren't Scans, and also it seems risky to make assumptions
about what a CustomScan wants to do here. Also, the most likely future
change here is to decide that we do need to support marks placed before
the first tuple, which would require additional work in IndexScan and
IndexOnlyScan in any case. Hence, fix the EPQ issue in nodeIndexscan.c
and nodeIndexonlyscan.c, accepting the small amount of code duplicated
thereby, and leave it to CustomScan providers to fix this bug if they
have it.
Back-patch to v10 where commit 09529a70b came in. In earlier branches,
the index_markpos() call is a waste of cycles when EPQ is active, but
no more than that, so it doesn't seem appropriate to back-patch further.
Discussion: https://postgr.es/m/20180126074932.3098.97815@wrigleys.postgresql.org
Avoid compiler warnings on MSVC (which doesn't want to see both
__forceinline and inline) and ancient GCC (which doesn't have
__attribute__((always_inline))).
Don't force inline-ing when building at -O0, as the programmer is probably
hoping for exact source-to-object-line correspondence in that case.
(For the moment this only works for GCC; maybe we can extend it later.)
Make pg_attribute_always_inline be syntactically a drop-in replacement
for inline, rather than an additional wart.
And improve the comments.
Thomas Munro and Michail Nikolaev, small tweaks by me
Discussion: https://postgr.es/m/32278.1514863068@sss.pgh.pa.us
Discussion: https://postgr.es/m/CANtu0oiYp74brgntKOxgg1FK5+t8uQ05guSiFU6FYz_5KUhr6Q@mail.gmail.com
In each of the supplied procedural languages (PL/pgSQL, PL/Perl,
PL/Python, PL/Tcl), add language-specific commit and rollback
functions/commands to control transactions in procedures in that
language. Add similar underlying functions to SPI. Some additional
cleanup so that transaction commit or abort doesn't blow away data
structures still used by the procedure call. Add execution context
tracking to CALL and DO statements so that transaction control commands
can only be issued in top-level procedure and block calls, not function
calls or other procedure or block calls.
- SPI
Add a new function SPI_connect_ext() that is like SPI_connect() but
allows passing option flags. The only option flag right now is
SPI_OPT_NONATOMIC. A nonatomic SPI connection can execute transaction
control commands, otherwise it's not allowed. This is meant to be
passed down from CALL and DO statements which themselves know in which
context they are called. A nonatomic SPI connection uses different
memory management. A normal SPI connection allocates its memory in
TopTransactionContext. For nonatomic connections we use PortalContext
instead. As the comment in SPI_connect_ext() (previously SPI_connect())
indicates, one could potentially use PortalContext in all cases, but it
seems safest to leave the existing uses alone, because this stuff is
complicated enough already.
SPI also gets new functions SPI_start_transaction(), SPI_commit(), and
SPI_rollback(), which can be used by PLs to implement their transaction
control logic.
- portalmem.c
Some adjustments were made in the code that cleans up portals at
transaction abort. The portal code could already handle a command
*committing* a transaction and continuing (e.g., VACUUM), but it was not
quite prepared for a command *aborting* a transaction and continuing.
In AtAbort_Portals(), remove the code that marks an active portal as
failed. As the comment there already predicted, this doesn't work if
the running command wants to keep running after transaction abort. And
it's actually not necessary, because pquery.c is careful to run all
portal code in a PG_TRY block and explicitly runs MarkPortalFailed() if
there is an exception. So the code in AtAbort_Portals() is never used
anyway.
In AtAbort_Portals() and AtCleanup_Portals(), we need to be careful not
to clean up active portals too much. This mirrors similar code in
PreCommit_Portals().
- PL/Perl
Gets new functions spi_commit() and spi_rollback()
- PL/pgSQL
Gets new commands COMMIT and ROLLBACK.
Update the PL/SQL porting example in the documentation to reflect that
transactions are now possible in procedures.
- PL/Python
Gets new functions plpy.commit and plpy.rollback.
- PL/Tcl
Gets new commands commit and rollback.
Reviewed-by: Andrew Dunstan <andrew.dunstan@2ndquadrant.com>
When an UPDATE causes a row to no longer match the partition
constraint, try to move it to a different partition where it does
match the partition constraint. In essence, the UPDATE is split into
a DELETE from the old partition and an INSERT into the new one. This
can lead to surprising behavior in concurrency scenarios because
EvalPlanQual rechecks won't work as they normally did; the known
problems are documented. (There is a pending patch to improve the
situation further, but it needs more review.)
Amit Khandekar, reviewed and tested by Amit Langote, David Rowley,
Rajkumar Raghuwanshi, Dilip Kumar, Amul Sul, Thomas Munro, Álvaro
Herrera, Amit Kapila, and me. A few final revisions by me.
Discussion: http://postgr.es/m/CAJ3gD9do9o2ccQ7j7+tSgiE1REY65XRiMb=yJO3u3QhyP8EEPQ@mail.gmail.com
AclObjectKind was basically just another enumeration for object types,
and we already have a preferred one for that. It's only used in
aclcheck_error. By using ObjectType instead, we can also give some more
precise error messages, for example "index" instead of "relation".
Reviewed-by: Michael Paquier <michael.paquier@gmail.com>
node->partitioned_rels is only set in UPDATE/DELETE cases, but
ExecInitModifyTable only uses its "rel" variable in INSERT cases,
so the extra logic to find the root rel is just a waste of complexity
and cycles.
Etsuro Fujita, reviewed by Amit Langote
Discussion: https://postgr.es/m/93cf9816-2f7d-0f67-8ed2-4a4e497a6ab8@lab.ntt.co.jp
PL/pgSQL "pins" internally generated (unnamed) portals so that user code
cannot close them by guessing their names. This logic is also useful in
other languages and really for any code. So move that logic into SPI.
An unnamed portal obtained through SPI_cursor_open() and related
functions is now automatically pinned, and SPI_cursor_close()
automatically unpins a portal that is pinned.
In the core distribution, this affects PL/Perl and PL/Python, preventing
users from manually closing cursors created by spi_query and
plpy.cursor, respectively. (PL/Tcl does not currently offer any cursor
functionality.)
Reviewed-by: Andrew Dunstan <andrew.dunstan@2ndquadrant.com>
Previously aggregate transition and combination functions were invoked
by special case code in nodeAgg.c, evaluating input and filters
separately using the expression evaluation machinery. That turns out
to not be great for performance for several reasons:
- repeated expression evaluations have some cost
- the transition functions invocations are poorly predicted, as
commonly there are multiple aggregates in a query, resulting in the
same call-stack invoking different functions.
- filter and input computation had to be done separately
- the special case code made it hard to implement JITing of the whole
transition function invocation
Address this by building one large expression that computes input,
evaluates filters, and invokes transition functions.
This leads to moderate speedups in queries bottlenecked by aggregate
computations, and enables large speedups for similar cases once JITing
is done.
There's potential for further improvement:
- It'd be nice if we could simplify the somewhat expensive
aggstate->all_pergroups lookups.
- right now there's still an advance_transition_function invocation in
nodeAgg.c, leading to some code duplication.
Author: Andres Freund
Discussion: https://postgr.es/m/20170901064131.tazjxwus3k2w3ybh@alap3.anarazel.de
After having gotten rid of PortalGetHeapMemory(), there seems little
reason to keep one Portal access macro around that offers no actual
abstraction and isn't consistently used anyway.
Reviewed-by: Andrew Dunstan <andrew.dunstan@2ndquadrant.com>
Reviewed-by: Alvaro Herrera <alvherre@alvh.no-ip.org>
Rename PortalMemory to TopPortalContext, to avoid confusion with
PortalContext and align naming with similar top-level memory contexts.
Rename PortalData's "heap" field to portalContext. The "heap" naming
seems quite antiquated and confusing. Also get rid of the
PortalGetHeapMemory() macro and access the field directly, which we do
for other portal fields, so this abstraction doesn't buy anything.
Reviewed-by: Andrew Dunstan <andrew.dunstan@2ndquadrant.com>
Reviewed-by: Alvaro Herrera <alvherre@alvh.no-ip.org>
At present, we always raise an ERROR if the partition constraint
is violated, but a pending patch for UPDATE tuple routing will
consider instead moving the tuple to the correct partition.
Refactor to make that simpler.
Amit Khandekar, reviewed by Amit Langote, David Rowley, and me.
Discussion: http://postgr.es/m/CAJ3gD9cue54GbEzfV-61nyGpijvjZgCcghvLsB0_nL8Nm8HzCA@mail.gmail.com
Instead of having ExecSetupPartitionTupleRouting return multiple out
parameters, have it return a pointer to a structure containing all of
those different things. Also, provide and use a cleanup function,
ExecCleanupTupleRouting, instead of cleaning up all of the resources
allocated by ExecSetupPartitionTupleRouting individually.
Amit Khandekar, reviewed by Amit Langote, David Rowley, and me
Discussion: http://postgr.es/m/CAJ3gD9fWfxgKC+PfJZF3hkgAcNOy-LpfPxVYitDEXKHjeieWQQ@mail.gmail.com
- Remove unnecessary #include mistakenly added in execnodes.h.
- Fix mistake in comment in choose_next_subplan_for_leader.
- Adjust row estimates in cost_append for a possibly-different
parallel divisor.
- Clamp row estimates in cost_append after operations that may
not produce integers.
Amit Kapila, with cosmetic adjustments by me.
Discussion: http://postgr.es/m/CAA4eK1+qcbeai3coPpRW=GFCzFeLUsuY4T-AKHqMjxpEGZBPQg@mail.gmail.com
Correct ExecParallelHashTuplePrealloc's estimate of whether the
space_allowed limit is exceeded. Be more consistent about tuples that
are exactly HASH_CHUNK_THRESHOLD in size (they're "small", not "large").
Neither of these things explain the current buildfarm unhappiness, but
they're still bugs.
Thomas Munro, per gripe by me
Discussion: https://postgr.es/m/CAEepm=34PDuR69kfYVhmZPgMdy8pSA-MYbpesEN1SR+2oj3Y+w@mail.gmail.com
Previously aggregate transition values for hash and other forms of
aggregation (i.e. sort and no group by) were represented
differently. Hash based aggregation used a grouping set indexed array
pointing to an array of transition values, whereas other forms of
aggregation used one flattened array with the index being computed out
of grouping set and transition offsets.
That made upcoming changes hard, so represent both as grouping set
indexed array of per-group data.
As a nice side-effect this also makes aggregation slightly faster,
because computing offsets with `transno + (setno * numTrans)` turns
out not to be that cheap (too big for x86 lea for example).
Author: Andres Freund
Discussion: https://postgr.es/m/20171128003121.nmxbm2ounxzb6n2t@alap3.anarazel.de
The previous coding relied (without any documentation) on the data[]
member of HashMemoryChunkData being at a MAXALIGN'ed offset. If it
was not, the tuples would not be maxaligned either, leading to failures
on alignment-picky machines. While there seems to be no live bug on any
platform we support, this is clearly pretty fragile: any addition to or
rearrangement of the fields in HashMemoryChunkData could break it.
Let's remove the hazard by getting rid of the data[] member and instead
using pointer arithmetic with an explicitly maxalign'ed offset.
Discussion: https://postgr.es/m/14483.1514938129@sss.pgh.pa.us
In a race case, EXPLAIN ANALYZE could fail to display correct nbatch
and size information. Refactor so that participants report only on
batches they worked on rather than trying to report on all of them,
and teach explain.c to consider the HashInstrumentation object from
all participants instead of picking the first one it can find. This
should fix an occasional build farm failure in the "join" regression
test.
Author: Thomas Munro
Reviewed-By: Andres Freund
Discussion: https://postgr.es/m/30219.1514428346%40sss.pgh.pa.us
This reduces code duplication a bit, but the primary benefit that it
makes JITing expression evaluation easier. When doing so we can't, as
previously done in the interpreted case, really change opcode without
recompiling. Nor dow we just carry around unnecessary branches to
avoid re-checking over and over.
As a minor side-effect this makes ExecEvalStepOp() O(log(N)) rather
than O(N).
Author: Andres Freund
Discussion: https://postgr.es/m/20170901064131.tazjxwus3k2w3ybh@alap3.anarazel.de
This is useful because it gets rid of the sole direct user of
ExecAssignResultType(). A future commit will likely make use of that
and combine creating the targetlist with the initialization of the
result slot. But it seems like good code hygiene anyway.
Author: Andres Freund
Discussion: https://postgr.es/m/20170901064131.tazjxwus3k2w3ybh@alap3.anarazel.de
When a backend runs out of inner tuples to hash, it should detach from
grow_batch_barrier only after it has flushed all batches to disk and
merged counters, not before. Otherwise a concurrent backend in
ExecParallelHashIncreaseNumBatches() could stop waiting for this
backend and try to read tuples before they have been written. This
commit reorders those operations and should fix the assertion failures
seen occasionally on the build farm since commit
1804284042.
Author: Thomas Munro
Discussion: https://postgr.es/m/E1eRwXy-0004IK-TO%40gemulon.postgresql.org
This patch does three interrelated things:
* Create a new expression execution step type EEOP_PARAM_CALLBACK
and add the infrastructure needed for add-on modules to generate that.
As discussed, the best control mechanism for that seems to be to add
another hook function to ParamListInfo, which will be called by
ExecInitExpr if it's supplied and a PARAM_EXTERN Param is found.
For stand-alone expressions, we add a new entry point to allow the
ParamListInfo to be specified directly, since it can't be retrieved
from the parent plan node's EState.
* Redesign the API for the ParamListInfo paramFetch hook so that the
ParamExternData array can be entirely virtual. This also lets us get rid
of ParamListInfo.paramMask, instead leaving it to the paramFetch hook to
decide which param IDs should be accessible or not. plpgsql_param_fetch
was already doing the identical masking check, so having callers do it too
seemed redundant. While I was at it, I added a "speculative" flag to
paramFetch that the planner can specify as TRUE to avoid unwanted failures.
This solves an ancient problem for plpgsql that it couldn't provide values
of non-DTYPE_VAR variables to the planner for fear of triggering premature
"record not assigned yet" or "field not found" errors during planning.
* Rework plpgsql to get rid of the need for "unshared" parameter lists,
by dint of turning the single ParamListInfo per estate into a nearly
read-only data structure that doesn't instantiate any per-variable data.
Instead, the paramFetch hook controls access to per-variable data and can
make the right decisions on the fly, replacing the cases that we used to
need multiple ParamListInfos for. This might perhaps have been a
performance loss on its own, but by using a paramCompile hook we can
bypass plpgsql_param_fetch entirely during normal query execution.
(It's now only called when, eg, we copy the ParamListInfo into a cursor
portal. copyParamList() or SerializeParamList() effectively instantiate
the virtual parameter array as a simple physical array without a
paramFetch hook, which is what we want in those cases.) This allows
reverting most of commit 6c82d8d1f, though I kept the cosmetic
code-consolidation aspects of that (eg the assign_simple_var function).
Performance testing shows this to be at worst a break-even change,
and it can provide wins ranging up to 20% in test cases involving
accesses to fields of "record" variables. The fact that values of
such variables can now be exposed to the planner might produce wins
in some situations, too, but I've not pursued that angle.
In passing, remove the "parent" pointer from the arguments to
ExecInitExprRec and related functions, instead storing that pointer in a
transient field in ExprState. The ParamListInfo pointer for a stand-alone
expression is handled the same way; we'd otherwise have had to add
yet another recursively-passed-down argument in expression compilation.
Discussion: https://postgr.es/m/32589.1513706441@sss.pgh.pa.us
Introduce parallel-aware hash joins that appear in EXPLAIN plans as Parallel
Hash Join with Parallel Hash. While hash joins could already appear in
parallel queries, they were previously always parallel-oblivious and had a
partial subplan only on the outer side, meaning that the work of the inner
subplan was duplicated in every worker.
After this commit, the planner will consider using a partial subplan on the
inner side too, using the Parallel Hash node to divide the work over the
available CPU cores and combine its results in shared memory. If the join
needs to be split into multiple batches in order to respect work_mem, then
workers process different batches as much as possible and then work together
on the remaining batches.
The advantages of a parallel-aware hash join over a parallel-oblivious hash
join used in a parallel query are that it:
* avoids wasting memory on duplicated hash tables
* avoids wasting disk space on duplicated batch files
* divides the work of building the hash table over the CPUs
One disadvantage is that there is some communication between the participating
CPUs which might outweigh the benefits of parallelism in the case of small
hash tables. This is avoided by the planner's existing reluctance to supply
partial plans for small scans, but it may be necessary to estimate
synchronization costs in future if that situation changes. Another is that
outer batch 0 must be written to disk if multiple batches are required.
A potential future advantage of parallel-aware hash joins is that right and
full outer joins could be supported, since there is a single set of matched
bits for each hashtable, but that is not yet implemented.
A new GUC enable_parallel_hash is defined to control the feature, defaulting
to on.
Author: Thomas Munro
Reviewed-By: Andres Freund, Robert Haas
Tested-By: Rafia Sabih, Prabhat Sahu
Discussion:
https://postgr.es/m/CAEepm=2W=cOkiZxcg6qiFQP-dHUe09aqTrEMM7yJDrHMhDv_RA@mail.gmail.comhttps://postgr.es/m/CAEepm=37HKyJ4U6XOLi=JgfSHM3o6B-GaeO-6hkOmneTDkH+Uw@mail.gmail.com
When a Gather or Gather Merge node is started and stopped multiple
times, accumulate instrumentation data only once, at the end, instead
of after each execution, to avoid recording inflated totals.
Commit 778e78ae9f, the previous attempt
at a fix, instead reset the state after every execution, which worked
for the general instrumentation data but had problems for the additional
instrumentation specific to Sort and Hash nodes.
Report by hubert depesz lubaczewski. Analysis and fix by Amit Kapila,
following a design proposal from Thomas Munro, with a comment tweak
by me.
Discussion: http://postgr.es/m/20171127175631.GA405@depesz.com
es_query_dsa turns out to be broken by design, because it supposes
that there is only one DSA for the whole query, whereas there is
actually one per Gather (Merge) node. For now, work around that
problem by setting and clearing the pointer around the sections of
code that might need it. It's probably a better idea to get rid of
es_query_dsa altogether in favor of having each node keep track
individually of which DSA is relevant, but that seems like more than
we would want to back-patch.
Thomas Munro, reviewed and tested by Andreas Seltenreich, Amit
Kapila, and by me.
Discussion: http://postgr.es/m/CAEepm=1U6as=brnVvMNixEV2tpi8NuyQoTmO8Qef0-VV+=7MDA@mail.gmail.com
In order for executor nodes to be able to change their ExecProcNode function
after ExecInitNode() has finished, provide ExecSetExecProcNode(). This allows
any wrappers functions that only execProcnode.c knows about to be reinstalled.
The motivation for wanting to change ExecProcNode after ExecInitNode() has
finished is that it is not known until later whether parallel query is
available, so if a parallel variant is to be installed then ExecInitNode()
is too soon to decide.
Author: Thomas Munro
Reviewed-By: Andres Freund
Discussion: https://postgr.es/m/CAEepm=09rr65VN+cAV5FgyM_z=D77Xy8Fuc9CDDDYbq3pQUezg@mail.gmail.com
I noticed that _SPI_execute_plan initially sets spierrcontext.arg = NULL,
and only fills it in some time later. If an error were to happen in
between, _SPI_error_callback would try to dereference the null pointer.
This is unlikely --- there's not much between those points except
push-snapshot calls --- but it's clearly not impossible. Tweak the
callback to do nothing if the pointer isn't set yet.
It's been like this for awhile, so back-patch to all supported branches.
When we create an Append node, we can spread out the workers over the
subplans instead of piling on to each subplan one at a time, which
should typically be a bit more efficient, both because the startup
cost of any plan executed entirely by one worker is paid only once and
also because of reduced contention. We can also construct Append
plans using a mix of partial and non-partial subplans, which may allow
for parallelism in places that otherwise couldn't support it.
Unfortunately, this patch doesn't handle the important case of
parallelizing UNION ALL by running each branch in a separate worker;
the executor infrastructure is added here, but more planner work is
needed.
Amit Khandekar, Robert Haas, Amul Sul, reviewed and tested by
Ashutosh Bapat, Amit Langote, Rafia Sabih, Amit Kapila, and
Rajkumar Raghuwanshi.
Discussion: http://postgr.es/m/CAJ3gD9dy0K_E8r727heqXoBmWZ83HwLFwdcaSSmBQ1+S+vRuUQ@mail.gmail.com
When a Gather or Gather Merge node is started and stopped multiple
times, the old code wouldn't reset the shared state between executions,
potentially resulting in dramatically inflated instrumentation data
for nodes beneath it. (The per-worker instrumentation ended up OK,
I think, but the overall totals were inflated.)
Report by hubert depesz lubaczewski. Analysis and fix by Amit Kapila,
reviewed and tweaked a bit by me.
Discussion: http://postgr.es/m/20171127175631.GA405@depesz.com
Tom Lane
Thomas Munro
Michael Paquier
Amit Kapila
Heikki Linnakangas
David Rowley
Andres Freund
Andrew Gierth
Peter Eisentraut
Etsuro Fujita
Alvaro Herrera
Robert Haas
Bruce Momjian
Peter Eisentraut
Magnus Hagander
Andrew Dunstan
Simon Riggs
Teodor Sigaev
Stephen Frost