If the hash functions used for hashing tuples leaked any memory,
we failed to clean that up, resulting in query-lifespan memory
leakage in queries using hashed subplans. One way that could
happen is if the values being hashed require de-toasting, since
most of our hash functions don't trouble to clean up de-toasted
inputs.
Prior to commit bf6c614a2, this leakage was largely masked
because TupleHashTableMatch would reset hashtable->tempcxt
(via execTuplesMatch). But it doesn't do that anymore, and
that's not really the right place for this anyway: doing it
there could reset the tempcxt many times per hash lookup,
or not at all. Instead put reset calls into ExecHashSubPlan
and buildSubPlanHash. Along the way to that, rearrange
ExecHashSubPlan so that there's just one place to call
MemoryContextReset instead of several.
This amounts to accepting the de-facto API spec that the caller
of the TupleHashTable routines is responsible for resetting the
tempcxt adequately often. Although the other callers seem to
get this right, it was not documented anywhere, so add a comment
about it.
Bug: #19040
Reported-by: Haiyang Li <mohen.lhy@alibaba-inc.com>
Author: Haiyang Li <mohen.lhy@alibaba-inc.com>
Reviewed-by: Fei Changhong <feichanghong@qq.com>
Reviewed-by: Tom Lane <tgl@sss.pgh.pa.us>
Discussion: https://postgr.es/m/19040-c9b6073ef814f48c@postgresql.org
Backpatch-through: 13
All the calls replaced by this commit use 4-byte integers for their
variables used in input of my_log2(). Hence, the limit against
too-large inputs does not really apply. Thresholds are also applied, as
of:
- In nodeAgg.c, the number of partitions is limited by
HASHAGG_MAX_PARTITIONS.
- In nodeHash.c, ExecChooseHashTableSize() caps its maximum number of
buckets based on HashJoinTuple and palloc() allocation limit.
- In worker.c, the number of subxacts tracked by ApplySubXactData uses
uint32, making pg_ceil_log2_64() safe to use directly.
Several approaches have been discussed, like an integration with
thresholds in pg_bitutils.h, but it was found confusing. This uses
Dean's idea, which gives a simpler result than what I came up with to be
able to remove dynahash.h. dynahash.h will be removed in a follow-up
commit, removing some duplication with the ceil log2 routines.
Reviewed-by: Peter Eisentraut <peter@eisentraut.org>
Reviewed-by: Dean Rasheed <dean.a.rasheed@gmail.com>
Discussion: https://postgr.es/m/CAEZATCUJPQD_7sC-wErak2CQGNa6bj2hY-mr8wsBki=kX7f2_A@mail.gmail.com
rte->alias should point only to a user-written alias, but in these
cases that principle was violated. Fixing this causes some regression
test output changes: wherever rte->alias previously had a value and
is now NULL, rte->eref is now set to a generated name rather than to
rte->alias; and the scheme used to generate eref names differs from
what we were doing for aliases.
The upshot is that instead of "*SELECT*" or "*SELECT* %d",
EXPLAIN will now emit "unnamed_subquery" or "unnamed_subquery_%d".
But that's a reasonable descriptor, and we were already producing
that in yet other cases, so this seems not too objectionable.
Author: Tom Lane <tgl@sss.pgh.pa.us>
Co-authored-by: Robert Haas <rhaas@postgresql.org>
Discussion: https://postgr.es/m/CA+TgmoYSYmDA2GvanzPMci084n+mVucv0bJ0HPbs6uhmMN6HMg@mail.gmail.com
When executing a MERGE UPDATE action, if there is more than one
concurrent update of the target row, the lock-and-retry code would
sometimes incorrectly identify the latest version of the target tuple,
leading to incorrect results.
This was caused by using the ctid field from the TM_FailureData
returned by table_tuple_lock() in a case where the result was TM_Ok,
which is unsafe because the TM_FailureData struct is not guaranteed to
be fully populated in that case. Instead, it should use the tupleid
passed to (and updated by) table_tuple_lock().
To reduce the chances of similar errors in the future, improve the
commentary for table_tuple_lock() and TM_FailureData to make it
clearer that table_tuple_lock() updates the tid passed to it, and most
fields of TM_FailureData should not be relied on in non-failure cases.
An exception to this is the "traversed" field, which is set in both
success and failure cases.
Reported-by: Dmitry <dsy.075@yandex.ru>
Author: Yugo Nagata <nagata@sraoss.co.jp>
Reviewed-by: Dean Rasheed <dean.a.rasheed@gmail.com>
Reviewed-by: Chao Li <li.evan.chao@gmail.com>
Discussion: https://postgr.es/m/1570d30e-2b95-4239-b9c3-f7bf2f2f8556@yandex.ru
Backpatch-through: 15
When executing a MERGE, check that the target relation supports all
actions mentioned in the MERGE command. Specifically, check that it
has a REPLICA IDENTITY if it publishes updates or deletes and the
MERGE command contains update or delete actions. Failing to do this
can silently break replication.
Author: Zhijie Hou <houzj.fnst@fujitsu.com>
Reviewed-by: Ashutosh Bapat <ashutosh.bapat.oss@gmail.com>
Reviewed-by: Dean Rasheed <dean.a.rasheed@gmail.com>
Tested-by: Chao Li <li.evan.chao@gmail.com>
Discussion: https://postgr.es/m/OS3PR01MB57180C87E43A679A730482DF94B62@OS3PR01MB5718.jpnprd01.prod.outlook.com
Backpatch-through: 15
If an INSERT has an ON CONFLICT DO UPDATE clause, the executor must
check that the target relation supports UPDATE as well as INSERT. In
particular, it must check that the target relation has a REPLICA
IDENTITY if it publishes updates. Formerly, it was not doing this
check, which could lead to silently breaking replication.
Fix by adding such a check to CheckValidResultRel(), which requires
adding a new onConflictAction argument. In back-branches, preserve ABI
compatibility by introducing a wrapper function with the original
signature.
Author: Zhijie Hou <houzj.fnst@fujitsu.com>
Reviewed-by: Ashutosh Bapat <ashutosh.bapat.oss@gmail.com>
Reviewed-by: Dean Rasheed <dean.a.rasheed@gmail.com>
Tested-by: Chao Li <li.evan.chao@gmail.com>
Discussion: https://postgr.es/m/OS3PR01MB57180C87E43A679A730482DF94B62@OS3PR01MB5718.jpnprd01.prod.outlook.com
Backpatch-through: 13
If we error out during execution of a SQL-language function, we will
often leave behind non-null pointers in its SQLFunctionCache's cplan
and eslist fields. This is problematic if the SQLFunctionCache is
re-used, because those pointers will point at resources that were
released during error cleanup. This problem escaped detection so far
because ordinarily we won't re-use an FmgrInfo+SQLFunctionCache struct
after a query error. However, in the rather improbable case that
someone implements an opclass support function in SQL language, there
will be long-lived FmgrInfos for it in the relcache, and then the
problem is reachable after the function throws an error.
To fix, add a flag to SQLFunctionCache that tracks whether execution
escapes out of fmgr_sql, and clear out the relevant fields during
init_sql_fcache if so. (This is going to need more thought if we ever
try to share FMgrInfos across threads; but it's very far from being
the only problem such a project will encounter, since many functions
regard fn_extra as being query-local state.)
This broke at commit 0313c5dc6; before that we did not try to re-use
SQLFunctionCache state across calls. Hence, back-patch to v18.
Bug: #19026
Reported-by: Alexander Lakhin <exclusion@gmail.com>
Author: Tom Lane <tgl@sss.pgh.pa.us>
Discussion: https://postgr.es/m/19026-90aed5e71d0c8af3@postgresql.org
Backpatch-through: 18
fb9f955025 optimized code generation by using specialized variants of
ExecSeqScan* for [not] having a qual, projection etc. This allowed the
compiler to optimize the code out the code for qual / projection. However, as
observed by David Rowley at the time, the compiler couldn't prove the
opposite, i.e. that the qual etc *are* present.
By using pg_assume(), introduced in d65eb5b1b8, we can tell the compiler that
the relevant variables are non-null.
This reduces the code size to a surprising degree and seems to lead to a small
but reproducible performance gain.
Reviewed-by: Amit Langote <amitlangote09@gmail.com> Discussion:
https://postgr.es/m/CA+HiwqFk-MbwhfX_kucxzL8zLmjEt9MMcHi2YF=DyhPrSjsBEA@mail.gmail.com
Commit e2d4ef8de8 (the fix for CVE-2017-7484) added security checks
to the selectivity estimation functions to prevent them from running
user-supplied operators on data obtained from pg_statistic if the user
lacks privileges to select from the underlying table. In cases
involving inheritance/partitioning, those checks were originally
performed against the child RTE (which for plain inheritance might
actually refer to the parent table). Commit 553d2ec271 then extended
that to also check the parent RTE, allowing access if the user had
permissions on either the parent or the child. It turns out, however,
that doing any checks using the child RTE is incorrect, since
securityQuals is set to NULL when creating an RTE for an inheritance
child (whether it refers to the parent table or the child table), and
therefore such checks do not correctly account for any RLS policies or
security barrier views. Therefore, do the security checks using only
the parent RTE. This is consistent with how RLS policies are applied,
and the executor's ACL checks, both of which use only the parent
table's permissions/policies. Similar checks are performed in the
extended stats code, so update that in the same way, centralizing all
the checks in a new function.
In addition, note that these checks by themselves are insufficient to
ensure that the user has access to the table's data because, in a
query that goes via a view, they only check that the view owner has
permissions on the underlying table, not that the current user has
permissions on the view itself. In the selectivity estimation
functions, there is no easy way to navigate from underlying tables to
views, so add permissions checks for all views mentioned in the query
to the planner startup code. If the user lacks permissions on a view,
a permissions error will now be reported at planner-startup, and the
selectivity estimation functions will not be run.
Checking view permissions at planner-startup in this way is a little
ugly, since the same checks will be repeated at executor-startup.
Longer-term, it might be better to move all the permissions checks
from the executor to the planner so that permissions errors can be
reported sooner, instead of creating a plan that won't ever be run.
However, such a change seems too far-reaching to be back-patched.
Back-patch to all supported versions. In v13, there is the added
complication that UPDATEs and DELETEs on inherited target tables are
planned using inheritance_planner(), which plans each inheritance
child table separately, so that the selectivity estimation functions
do not know that they are dealing with a child table accessed via its
parent. Handle that by checking access permissions on the top parent
table at planner-startup, in the same way as we do for views. Any
securityQuals on the top parent table are moved down to the child
tables by inheritance_planner(), so they continue to be checked by the
selectivity estimation functions.
Author: Dean Rasheed <dean.a.rasheed@gmail.com>
Reviewed-by: Tom Lane <tgl@sss.pgh.pa.us>
Reviewed-by: Noah Misch <noah@leadboat.com>
Backpatch-through: 13
Security: CVE-2025-8713
Fix a couple more places where an explicit Datum conversion
is needed (not clear how we missed these in ff89e182d and
previous commits).
Replace the minority usage "(Datum) NULL" with "(Datum) 0".
The former depends on the assumption that Datum is the same
width as Pointer, the latter doesn't. Anyway consistency
is a good thing.
This is, I believe, the last of the notational mop-up needed
before we can consider changing Datum to uint64 everywhere.
It's also important cleanup for more aggressive ideas such
as making Datum a struct.
Discussion: https://postgr.es/m/1749799.1752797397@sss.pgh.pa.us
Discussion: https://postgr.es/m/8246d7ff-f4b7-4363-913e-827dadfeb145@eisentraut.org
Add various missing conversions from and to Datum. The previous code
mostly relied on implicit conversions or its own explicit casts
instead of using the correct DatumGet*() or *GetDatum() functions.
We think these omissions are harmless. Some actual bugs that were
discovered during this process have been committed
separately (80c758a2e1, fd2ab03fea).
Reviewed-by: Tom Lane <tgl@sss.pgh.pa.us>
Discussion: https://www.postgresql.org/message-id/flat/8246d7ff-f4b7-4363-913e-827dadfeb145%40eisentraut.org
This enhancement builds upon the infrastructure introduced in commit
228c370868, which enables the preservation of deleted tuples and their
origin information on the subscriber. This capability is crucial for
handling concurrent transactions replicated from remote nodes.
The update introduces support for detecting update_deleted conflicts
during the application of update operations on the subscriber. When an
update operation fails to locate the target row-typically because it has
been concurrently deleted-we perform an additional table scan. This scan
uses the SnapshotAny mechanism and we do this additional scan only when
the retain_dead_tuples option is enabled for the relevant subscription.
The goal of this scan is to locate the most recently deleted tuple-matching
the old column values from the remote update-that has not yet been removed
by VACUUM and is still visible according to our slot (i.e., its deletion
is not older than conflict-detection-slot's xmin). If such a tuple is
found, the system reports an update_deleted conflict, including the origin
and transaction details responsible for the deletion.
This provides a groundwork for more robust and accurate conflict
resolution process, preventing unexpected behavior by correctly
identifying cases where a remote update clashes with a deletion from
another origin.
Author: Zhijie Hou <houzj.fnst@fujitsu.com>
Reviewed-by: shveta malik <shveta.malik@gmail.com>
Reviewed-by: Nisha Moond <nisha.moond412@gmail.com>
Reviewed-by: Dilip Kumar <dilipbalaut@gmail.com>
Reviewed-by: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Reviewed-by: Amit Kapila <amit.kapila16@gmail.com>
Discussion: https://postgr.es/m/OS0PR01MB5716BE80DAEB0EE2A6A5D1F5949D2@OS0PR01MB5716.jpnprd01.prod.outlook.com
Commit 719dcf3c42 introduced a field called CachedPlanType in
PlannedStmt to allow extensions to determine whether a cached plan is
generic or custom.
After discussion, the concepts that we want to track are a bit wider
than initially anticipated, as it is closer to knowing from which
"source" or "origin" a PlannedStmt has been generated or retrieved.
Custom and generic cached plans are a subset of that.
Based on the state of HEAD, we have been able to define two more
origins:
- "standard", for the case where PlannedStmt is generated in
standard_planner(), the most common case.
- "internal", for the fake PlannedStmt generated internally by some
query patterns.
This could be tuned in the future depending on what is needed. This
looks like a good starting point, at least. The default value is called
"UNKNOWN", provided as fallback value. This value is not used in the
core code, the idea is to let extensions building their own PlannedStmts
know about this new field.
Author: Michael Paquier <michael@paquier.xyz>
Co-authored-by: Sami Imseih <samimseih@gmail.com>
Discussion: https://postgr.es/m/aILaHupXbIGgF2wJ@paquier.xyz
PlannedStmt gains a new field, called CachedPlanType, able to track if a
given plan tree originates from the cache and if we are dealing with a
generic or custom cached plan.
This field can be used for monitoring or statistical purposes, in the
executor hooks, for example, based on the planned statement attached to
a QueryDesc. A patch is under discussion for pg_stat_statements to
provide an equivalent of the counters in pg_prepared_statements for
custom and generic plans, to provide a more global view of such data, as
this data is now restricted to the current session.
The concept introduced in this commit is useful on its own, and has been
extracted from a larger patch by the same author.
Author: Sami Imseih <samimseih@gmail.com>
Reviewed-by: Andrei Lepikhov <lepihov@gmail.com>
Reviewed-by: Michael Paquier <michael@paquier.xyz>
Discussion: https://postgr.es/m/CAA5RZ0uFw8Y9GCFvafhC=OA8NnMqVZyzXPfv_EePOt+iv1T-qQ@mail.gmail.com
If a MERGE inside a CTE attempts an UPDATE or DELETE on a table with
BEFORE ROW triggers, and a concurrent UPDATE or DELETE happens, the
merge code would fail (crashing in the case of an UPDATE action, and
potentially executing the wrong action for a DELETE action).
This is the same issue that 9321c79c86 attempted to fix, except now
for a MERGE inside a CTE. As noted in 9321c79c86, what needs to happen
is for the trigger code to exit early, returning the TM_Result and
TM_FailureData information to the merge code, if a concurrent
modification is detected, rather than attempting to do an EPQ
recheck. The merge code will then do its own rechecking, and rescan
the action list, potentially executing a different action in light of
the concurrent update. In particular, the trigger code must never call
ExecGetUpdateNewTuple() for MERGE, since that is bound to fail because
MERGE has its own per-action projection information.
Commit 9321c79c86 did this using estate->es_plannedstmt->commandType
in the trigger code to detect that a MERGE was being executed, which
is fine for a plain MERGE command, but does not work for a MERGE
inside a CTE. Fix by passing that information to the trigger code as
an additional parameter passed to ExecBRUpdateTriggers() and
ExecBRDeleteTriggers().
Back-patch as far as v17 only, since MERGE cannot appear inside a CTE
prior to that. Additionally, take care to preserve the trigger ABI in
v17 (though not in v18, which is still in beta).
Bug: #18986
Reported-by: Yaroslav Syrytsia <me@ys.lc>
Author: Dean Rasheed <dean.a.rasheed@gmail.com>
Reviewed-by: Michael Paquier <michael@paquier.xyz>
Discussion: https://postgr.es/m/18986-e7a8aac3d339fa47@postgresql.org
Backpatch-through: 17
When a MERGE's target table is the parent of an inheritance tree, any
INSERT actions insert into the parent table using ModifyTableState's
rootResultRelInfo. However, there are two bugs in the way is
initialized:
1. ExecInitMerge() incorrectly uses a different ResultRelInfo entry
from ModifyTableState's resultRelInfo array to build the insert
projection, which may not be compatible with rootResultRelInfo.
2. ExecInitModifyTable() does not fully initialize rootResultRelInfo.
Specifically, ri_WithCheckOptions, ri_WithCheckOptionExprs,
ri_returningList, and ri_projectReturning are not initialized.
This can lead to crashes, or incorrect query results due to failing to
check WCO's or process the RETURNING list for INSERT actions.
Fix both these bugs in ExecInitMerge(), noting that it is only
necessary to fully initialize rootResultRelInfo if the MERGE has
INSERT actions and the target table is a plain inheritance parent.
Backpatch to v15, where MERGE was introduced.
Reported-by: Andres Freund <andres@anarazel.de>
Author: Dean Rasheed <dean.a.rasheed@gmail.com>
Reviewed-by: Jian He <jian.universality@gmail.com>
Reviewed-by: Tender Wang <tndrwang@gmail.com>
Discussion: https://postgr.es/m/4rlmjfniiyffp6b3kv4pfy4jw3pciy6mq72rdgnedsnbsx7qe5@j5hlpiwdguvc
Backpatch-through: 15
The code carelessly modified mtstate->ps.plan->targetlist,
which it's not supposed to do. Fortunately, there's not
really any need to do that because the planner already
set up a perfectly acceptable targetlist for the plan node.
We just need to remove the erroneous assignments and update some
relevant comments.
As it happens, the erroneous assignments caused the targetlist to
point to a different part of the source plan tree, so that there
isn't really a risk of the pointer becoming dangling after executor
termination. The only visible effect of this change we can find is
that EXPLAIN will show upper references to the ModifyTable's output
expressions using different variables. Formerly it showed Vars from
the first target relation that survived executor-startup pruning.
Now it always shows such references using the first relation appearing
in the planner output, independently of what happens during executor
pruning. On the whole that seems like a good thing.
Also make a small tweak in ExplainPreScanNode to ensure that the first
relation will receive a refname assignment in set_rtable_names, even
if it got pruned at startup. Previously the Vars might be shown
without any table qualification, which is confusing in a multi-table
query.
I considered back-patching this, but since the bug doesn't seem to
have any really terrible consequences in existing branches, it
seems better to not change their EXPLAIN output. It's not too late
for v18 though, especially since v18 already made other changes in
the EXPLAIN output for these cases.
Reported-by: Tom Lane <tgl@sss.pgh.pa.us>
Author: Andres Freund <andres@anarazel.de>
Co-authored-by: Tom Lane <tgl@sss.pgh.pa.us>
Discussion: https://postgr.es/m/213261.1747611093@sss.pgh.pa.us
As pointed out by Tom Lane, the patch introduced fragile and invasive
design around plan invalidation handling when locking of prunable
partitions was deferred from plancache.c to the executor. In
particular, it violated assumptions about CachedPlan immutability and
altered executor APIs in ways that are difficult to justify given the
added complexity and overhead.
This also removes the firstResultRels field added to PlannedStmt in
commit 28317de72, which was intended to support deferred locking of
certain ModifyTable result relations.
Reported-by: Tom Lane <tgl@sss.pgh.pa.us>
Discussion: https://postgr.es/m/605328.1747710381@sss.pgh.pa.us
The macros INJECTION_POINT() and INJECTION_POINT_CACHED() are extended
with an optional argument that can be passed down to the callback
attached when an injection point is run, giving to callbacks the
possibility to manipulate a stack state given by the caller. The
existing callbacks in modules injection_points and test_aio have their
declarations adjusted based on that.
da7226993f (core AIO infrastructure) and 93bc3d75d8 (test_aio) and
been relying on a set of workarounds where a static variable called
pgaio_inj_cur_handle is used as runtime argument in the injection point
callbacks used by the AIO tests, in combination with a TRY/CATCH block
to reset the argument value. The infrastructure introduced in this
commit will be reused for the AIO tests, simplifying them.
Reviewed-by: Greg Burd <greg@burd.me>
Discussion: https://postgr.es/m/Z_y9TtnXubvYAApS@paquier.xyz
We only need a tuplestore if we're actually going to accumulate
multiple result tuples. Obviously then we don't need one for non-set-
returning functions; but even a SRF doesn't need one if we decide to
use "lazyEval" (one row at a time) mode. In these cases, it's
sufficient to use the junkfilter's result slot to hold the single row
that's due to be returned. We just need to "materialize" that slot
to ensure it holds onto the data past shutdown of the sub-executor.
The original intent of this patch was partially to save a few cycles
(by not putting tuples into a tuplestore only to pull them back out
immediately), but mostly to ensure that we don't use a tuplestore
in non-set-returning functions. That's because I had concerns
about whether a tuplestore is safe to keep across queries,
which was possible for functions invoked via long-lived FmgrInfos
such as those kept in the typcache. There are no cases where SRFs
are called that way, so getting rid of the tuplestore in non-SRFs
should make things safer.
However, it emerges that running fmgr_sql in a short-lived context
(as 595d1efed made it do) makes the existing coding unsafe anyway:
we can end up with a long-lived TupleTableSlot holding a freeable
reference to a short-lived tuple, resulting in a double-free crash.
Not trying to pull tuples out of the tuplestore using that slot
dodges the problem, so I'm going to commit this now rather than
invent a band-aid solution for v18.
Reported-by: Alexander Lakhin <exclusion@gmail.com>
Author: Tom Lane <tgl@sss.pgh.pa.us>
Discussion: https://postgr.es/m/2443532.1744919968@sss.pgh.pa.us
Discussion: https://postgr.es/m/9f975803-1a1c-4f21-b987-f572e110e860@gmail.com
This gets rid of repetitive get_typlen calls in postquel_sub_params,
which show up as costing a few percent of the runtime in simple test
cases (more with more parameters).
In combination with the preceding patches, this gets us most of the
way back down to the amount of per-call overhead that functions.c
had before commit 0dca5d68d. There are some more things that could
be done, but this seems like an okay place to stop for v18.
At this point, the only data structures we allocate directly in
fcontext are the SQLFunctionCache struct itself, the ParamListInfo
struct, and the execution_state array, all of which are small and
perfectly capable of being re-used across executions of the same
FmgrInfo. Hence, let's give them the same lifespan as the FmgrInfo.
This step gets rid of the separate SQLFunctionLink struct and makes
fn_extra point to SQLFunctionCache again. We also get rid of the
separate fcontext memory context and allocate these items directly
in fn_mcxt.
For notational simplicity, SQLFunctionCache still has an fcontext
field, but it's just a copy of fn_mcxt.
The motivation for this is to allow these structures to live as
long as the FmgrInfo and be re-used across calls, restoring the
original design without its propensity for memory leaks. This
gets rid of some per-call overhead that we added in 0dca5d68d.
We also make an effort to re-use the JunkFilter and result slot.
Those might need to change if the function definition changes,
so we compromise by rebuilding them if the cached plan changes.
This also moves the tuplestore into fn_mcxt so that it can be
re-used across calls, again undoing a change made in 0dca5d68d.
Put the JunkFilter and its result slot (and thence also
some subsidiary data such as the result tupledesc) into a
separate subcontext "jfcontext". This doesn't accomplish
a lot at this point, because we make a new JunkFilter each
time through the SQL function. However, the plan is to make
the fcontext long-lived, and that raises the possibility
that we'll need a new JunkFilter because the plan for the
result-generating query changes. A separate context makes
it easy to free the obsoleted data when that happens.
Also, instead of always running the sub-executor in fcontext,
make a separate context for it if we're doing lazy eval of
a SRF, and otherwise just run it inside CurrentMemoryContext.
Previously, much of this code ran with CurrentMemoryContext set
to be the function's fcontext, so that we tended to leak a lot of
stuff there. Commit 0dca5d68d dealt with that by releasing the
fcontext at the completion of each SQL function call, but we'd
like to go back to the previous approach of allowing the fcontext
to be query-lifespan. To control the leakage problem, rearrange
the code so that we mostly run in the memory context that fmgr_sql
is called in (which we expect to be short-lived). Notably, this
means that parsing/planning is all done in the short-lived context
and doesn't leak cruft into fcontext.
This patch also fixes the allocation of execution_state records
so that we don't leak them across executions. I set that up
with a re-usable array that contains at least as many
execution_state structs as we need for the current querytree.
The chain structure is still there, but it's not really doing
much for us, and maybe somebody will be motivated to get rid
of it. I'm not though.
This incidentally also moves the call of BlessTupleDesc to be
with the code that creates the JunkFilter. That doesn't make
much difference now, but a later patch will reduce the number
of times the JunkFilter gets made, and we needn't bless the
results any more often than that.
We still leak a fair amount in fcontext, particularly when
executing utility statements, but that's material for a
separate patch step; the point here is only to get rid of
unintentional allocations in fcontext.
Late in the development of commit 0dca5d68d, I added a step to copy
the result tlist we extract from the cached final query, because
I was afraid that that might not last as long as the JunkFilter that
we're passing it off to. However, that turns out to cost a noticeable
number of cycles, and it's really quite unnecessary because the
JunkFilter will not examine that tlist after it's been created.
(ExecFindJunkAttribute would use it, but we don't use that function
on this JunkFilter.) Hence, remove the copy step. For safety,
reset the might-become-dangling jf_targetList pointer to NIL.
In passing, remove DR_sqlfunction.cxt, which we don't use anymore;
it's confusing because it's not entirely clear which context it
ought to point at.
If a GENERATED column is declared to have a domain data type where
the domain's constraints disallow null values, INSERT commands failed
because we built a targetlist that included coercing a null constant
to the domain's type. The failure occurred even when the generated
value would have been perfectly OK. This is adjacent to the issues
fixed in 0da39aa76, but we didn't notice for lack of testing a domain
with such a constraint.
We aren't going to use the result of the targetlist entry for the
generated column --- ExecComputeStoredGenerated will overwrite it.
So it's not really necessary that it have the exact datatype of
the generated column. This patch fixes the problem by changing
the targetlist entry to be a null Const of the domain's base type,
which should be sufficiently legal. (We do have to tweak
ExecCheckPlanOutput to accept the situation, though.)
This has been broken since we implemented generated columns.
However, this patch only applies easily as far back as v14, partly
because I (tgl) only carried 0da39aa76 back that far, but mostly
because v14 significantly refactored the handling of INSERT/UPDATE
targetlists. Given the lack of field complaints and the short
remaining support lifetime of v13, I judge the cost-benefit ratio
not good for devising a version that would work in v13.
Reported-by: jian he <jian.universality@gmail.com>
Author: jian he <jian.universality@gmail.com>
Reviewed-by: Tom Lane <tgl@sss.pgh.pa.us>
Discussion: https://postgr.es/m/CACJufxG59tip2+9h=rEv-ykOFjt0cbsPVchhi0RTij8bABBA0Q@mail.gmail.com
Backpatch-through: 14
Make sure that function declarations use names that exactly match the
corresponding names from function definitions in a few places. These
inconsistencies were all introduced during Postgres 18 development.
This commit was written with help from clang-tidy, by mechanically
applying the same rules as similar clean-up commits (the earliest such
commit was commit 035ce1fe).
The issue happens when building conflict information during apply of
INSERT or UPDATE operations that violate unique constraints on leaf
partitions.
The problem was introduced in commit 9ff68679b5, which removed the
redundant calls to ExecOpenIndices/ExecCloseIndices. The previous code was
relying on the redundant ExecOpenIndices call in
apply_handle_tuple_routing() to build the index information required for
unique key conflict detection.
The fix is to delay building the index information until a conflict is
detected instead of relying on ExecOpenIndices to do the same. The
additional benefit of this approach is that it avoids building index
information when there is no conflict.
Author: Hou Zhijie <houzj.fnst@fujitsu.com>
Reviewed-by:Reviewed-by: Amit Kapila <amit.kapila16@gmail.com>
Discussion: https://postgr.es/m/TYAPR01MB57244ADA33DDA57119B9D26494A62@TYAPR01MB5724.jpnprd01.prod.outlook.com
There were several places in ordering-related planning where a
requirement for btree was hardcoded but an amcanorder index could
suffice. This fixes that. We just need to do the necessary mapping
between strategy numbers and compare types and adjust some related
APIs so that this works independent of btree strategy numbers. For
instance, non-btree amcanorder indexes can now be used to support
sorting and merge joins. Also, predtest.c works independent of btree
strategy numbers now.
To avoid performance regressions, some details on btree and other
built-in index types are still hardcoded as shortcuts, but other index
types now have access to the same features by providing the required
flags and callbacks.
Author: Mark Dilger <mark.dilger@enterprisedb.com>
Co-authored-by: Peter Eisentraut <peter@eisentraut.org>
Discussion: https://www.postgresql.org/message-id/flat/E72EAA49-354D-4C2E-8EB9-255197F55330@enterprisedb.com
This is yet another bit of fallout from the fact that backend/parser
(like other code) feels free to scribble on the parse tree it's
handed. In this case that resulted in modifying the
relatively-short-lived copy in the cached function's source_list.
That would be fine since we only need each source_list tree once
... except that if the parser fails after making some changes,
the function cache entry remains as-is and will still be there
if the user tries to execute the function again. Then we have
problems because we're feeding a non-pristine tree to the parser.
The most expedient fix is a quick copyObject(). I considered
other answers like somehow marking the cache entry invalid
temporarily, but that would add complexity and I'm not sure
it's worth it. In typical scenarios we'd only do this once
per function query per session.
Reported-by: Alexander Lakhin <exclusion@gmail.com>
Author: Tom Lane <tgl@sss.pgh.pa.us>
Discussion: https://postgr.es/m/6d442183-102c-498a-81d1-eeeb086cdc5a@gmail.com
As coded, fmgr_sql() would get an assertion failure for a SQL function
that has an empty body and is declared to return some type other than
VOID. Typically you'd never get that far because fmgr_sql_validator()
would reject such a definition (I suspect that's how come I managed to
miss the bug). But if check_function_bodies is off or the function is
polymorphic, the validation check wouldn't get made.
Reported-by: Alexander Lakhin <exclusion@gmail.com>
Author: Tom Lane <tgl@sss.pgh.pa.us>
Discussion: https://postgr.es/m/0fde377a-3870-4d18-946a-ce008ee5bb88@gmail.com
The optimization does not take the removal of TIDs by a concurrent vacuum into
account. The concurrent vacuum can remove dead TIDs and make pages ALL_VISIBLE
while those dead TIDs are referenced in the bitmap. This can lead to a
skip_fetch scan returning too many tuples.
It likely would be possible to implement this optimization safely, but we
don't have the necessary infrastructure in place. Nor is it clear that it's
worth building that infrastructure, given how limited the skip_fetch
optimization is.
In the backbranches we just disable the optimization by always passing
need_tuples=true to table_beginscan_bm(). We can't perform API/ABI changes in
the backbranches and we want to make the change as minimal as possible.
Author: Matthias van de Meent <boekewurm+postgres@gmail.com>
Reported-By: Konstantin Knizhnik <knizhnik@garret.ru>
Discussion: https://postgr.es/m/CAEze2Wg3gXXZTr6_rwC+s4-o2ZVFB5F985uUSgJTsECx6AmGcQ@mail.gmail.com
Backpatch-through: 13
In the historical implementation of SQL functions (if they don't get
inlined), we built plans for all the contained queries at first call
within an outer query, and then re-used those plans for the duration
of the outer query, and then forgot everything. This was not ideal,
not least because the plans could not be customized to specific values
of the function's parameters. Our plancache infrastructure seems
mature enough to be used here. That will solve both the problem with
not being able to build custom plans and the problem with not being
able to share work across successive outer queries.
Aside from those performance concerns, this change fixes a
longstanding bugaboo with SQL functions: you could not write DDL that
would affect later statements in the same function. That's mostly
still true with new-style SQL functions, since the results of parse
analysis are baked into the stored query trees (and protected by
dependency records). But for old-style SQL functions, it will now
work much as it does with PL/pgSQL functions, because we delay parse
analysis and planning of each query until we're ready to run it.
Some edge cases that require replanning are now handled better too;
see for example the new rowsecurity test, where we now detect an RLS
context change that was previously missed.
One other edge-case change that might be worthy of a release note
is that we now insist that a SQL function's result be generated
by the physically-last query within it. Previously, if the last
original query was deleted by a DO INSTEAD NOTHING rule, we'd be
willing to take the result from the preceding query instead.
This behavior was undocumented except in source-code comments,
and it seems hard to believe that anyone's relying on it.
Along the way to this feature, we needed a few infrastructure changes:
* The plancache can now take either a raw parse tree or an
analyzed-but-not-rewritten Query as the starting point for a
CachedPlanSource. If given a Query, it is caller's responsibility
that nothing will happen to invalidate that form of the query.
We use this for new-style SQL functions, where what's in pg_proc is
serialized Query(s) and we trust the dependency mechanism to disallow
DDL that would break those.
* The plancache now offers a way to invoke a post-rewrite callback
to examine/modify the rewritten parse tree when it is rebuilding
the parse trees after a cache invalidation. We need this because
SQL functions sometimes adjust the parse tree to make its output
exactly match the declared result type; if the plan gets rebuilt,
that has to be re-done.
* There is a new backend module utils/cache/funccache.c that
abstracts the idea of caching data about a specific function
usage (a particular function and set of input data types).
The code in it is moved almost verbatim from PL/pgSQL, which
has done that for a long time. We use that logic now for
SQL-language functions too, and maybe other PLs will have use
for it in the future.
Author: Alexander Pyhalov <a.pyhalov@postgrespro.ru>
Co-authored-by: Tom Lane <tgl@sss.pgh.pa.us>
Reviewed-by: Pavel Stehule <pavel.stehule@gmail.com>
Discussion: https://postgr.es/m/8216639.NyiUUSuA9g@aivenlaptop
ExecInitPartitionInfo() duplicates much of the logic in
ExecInitMerge(), except that it failed to handle DO NOTHING
actions. This would cause an "unknown action in MERGE WHEN clause"
error if a MERGE with any DO NOTHING actions attempted to insert into
a partition not already initialised by ExecInitModifyTable().
Bug: #18871
Reported-by: Alexander Lakhin <exclusion@gmail.com>
Author: Tender Wang <tndrwang@gmail.com>
Reviewed-by: Gurjeet Singh <gurjeet@singh.im>
Discussion: https://postgr.es/m/18871-b44e3c96de3bd2e8%40postgresql.org
Backpatch-through: 15
This was left out of the original patch for virtual generated columns
(commit 83ea6c5402).
This just involves a bit of extra work in the executor to expand the
generation expressions and run a "IS NOT NULL" test against them.
There is also a bit of work to make sure that not-null constraints are
checked during a table rewrite.
Author: jian he <jian.universality@gmail.com>
Reviewed-by: Xuneng Zhou <xunengzhou@gmail.com>
Reviewed-by: Navneet Kumar <thanit3111@gmail.com>
Reviewed-by: Álvaro Herrera <alvherre@alvh.no-ip.org>
Discussion: https://postgr.es/m/CACJufxHArQysbDkWFmvK+D1TPHQWWTxWN15cMuUaTYX3xhQXgg@mail.gmail.com
Modernize code in ExecRelCheck() and ExecConstraints() a bit,
preparing the way for some new code.
Co-authored-by: jian he <jian.universality@gmail.com>
Reviewed-by: Xuneng Zhou <xunengzhou@gmail.com>
Reviewed-by: Navneet Kumar <thanit3111@gmail.com>
Reviewed-by: Álvaro Herrera <alvherre@alvh.no-ip.org>
Discussion: https://postgr.es/m/CACJufxHArQysbDkWFmvK+D1TPHQWWTxWN15cMuUaTYX3xhQXgg@mail.gmail.com
Reduces memory required for hash aggregation by avoiding an allocation
and a pointer in the TupleHashEntryData structure. That structure is
used for all buckets, whether occupied or not, so the savings is
substantial.
Discussion: https://postgr.es/m/AApHDvpN4v3t_sdz4dvrv1Fx_ZPw=twSnxuTEytRYP7LFz5K9A@mail.gmail.com
Reviewed-by: David Rowley <dgrowleyml@gmail.com>
Allows an "extra" argument that allocates extra memory at the end of
the MinimalTuple. This is important for callers that need to store
additional data, but do not want to perform an additional allocation.
Suggested-by: David Rowley <dgrowleyml@gmail.com>
Discussion: https://postgr.es/m/CAApHDvppeqw2pNM-+ahBOJwq2QmC0hOAGsmCpC89QVmEoOvsdg@mail.gmail.com
The entries aren't freed until the entire hash table is destroyed, so
use the Bump allocator to improve allocation speed, avoid wasting
space on the chunk header, and avoid wasting space due to the
power-of-two allocations.
Discussion: https://postgr.es/m/CAApHDvqv1aNB4cM36FzRwivXrEvBO_LsG_eQ3nqDXTjECaatOQ@mail.gmail.com
Reviewed-by: David Rowley
Introduce a new conflict type, multiple_unique_conflicts, to handle cases
where an incoming row during logical replication violates multiple UNIQUE
constraints.
Previously, the apply worker detected and reported only the first
encountered key conflict (insert_exists/update_exists), causing repeated
failures as each constraint violation needs to be handled one by one
making the process slow and error-prone.
With this patch, the apply worker checks all unique constraints upfront
once the first key conflict is detected and reports
multiple_unique_conflicts if multiple violations exist. This allows users
to resolve all conflicts at once by deleting all conflicting tuples rather
than dealing with them individually or skipping the transaction.
In the future, this will also allow us to specify different resolution
handlers for such a conflict type.
Add the stats for this conflict type in pg_stat_subscription_stats.
Author: Nisha Moond <nisha.moond412@gmail.com>
Author: Zhijie Hou <houzj.fnst@fujitsu.com>
Reviewed-by: Amit Kapila <amit.kapila16@gmail.com>
Reviewed-by: Peter Smith <smithpb2250@gmail.com>
Reviewed-by: Dilip Kumar <dilipbalaut@gmail.com>
Discussion: https://postgr.es/m/CABdArM7FW-_dnthGkg2s0fy1HhUB8C3ELA0gZX1kkbs1ZZoV3Q@mail.gmail.com
This field can be optionally set in a PlannedStmt through the planner
hook, giving extensions the possibility to assign an identifier related
to a computed plan. The backend is changed to report it in the backend
entry of a process running (including the extended query protocol), with
semantics and APIs to set or get it similar to what is used for the
existing query ID (introduced in the backend via 4f0b0966c8). The plan
ID is reset at the same timing as the query ID. Currently, this
information is not added to the system view pg_stat_activity; extensions
can access it through PgBackendStatus.
Some patches have been proposed to provide some features in the planning
area, where a plan identifier is used as a key to know the plan involved
(for statistics, plan storage and manipulations, etc.), and the point of
this commit is to provide an anchor in the backend that extensions can
rely on for future work. The reset of the plan identifier is
controlled by core and follows the same pattern as the query identifier
added in 4f0b0966c8.
The contents of this commit are extracted from a larger set proposed
originally by Lukas Fittl, that Sami Imseih has proposed as an
independent change, with a few tweaks sprinkled by me.
Author: Lukas Fittl <lukas@fittl.com>
Author: Sami Imseih <samimseih@gmail.com>
Reviewed-by: Bertrand Drouvot <bertranddrouvot.pg@gmail.com>
Reviewed-by: Michael Paquier <michael@paquier.xyz>
Discussion: https://postgr.es/m/CAP53Pkyow59ajFMHGpmb1BK9WHDypaWtUsS_5DoYUEfsa_Hktg@mail.gmail.com
Discussion: https://postgr.es/m/CAA5RZ0vyWd4r35uUBUmhngv8XqeiJUkJDDKkLf5LCoWxv-t_pw@mail.gmail.com
Commit cbc127917e introduced tracking of unpruned relids to avoid
processing pruned relations, and changed ExecInitModifyTable() to
initialize only unpruned result relations. As a result, MERGE
statements that prune all target partitions can now lead to crashes
or incorrect behavior during execution.
The crash occurs because some executor code paths rely on
ModifyTableState.resultRelInfo[0] being present and initialized,
even when no result relations remain after pruning. For example,
ExecMerge() and ExecMergeNotMatched() use the first resultRelInfo
to determine the appropriate action. Similarly,
ExecInitPartitionInfo() assumes that at least one result relation
exists.
To preserve these assumptions, ExecInitModifyTable() now includes the
first result relation in the initialized result relation list if all
result relations for that ModifyTable were pruned. To enable that,
ExecDoInitialPruning() ensures the first relation is locked if it was
pruned and locking is necessary.
To support this exception to the pruning logic, PlannedStmt now
includes a list of RT indexes identifying the first result relation
of each ModifyTable node in the plan. This allows
ExecDoInitialPruning() to check whether each such relation was
pruned and, if so, lock it if necessary.
Bug: #18830
Reported-by: Robins Tharakan <tharakan@gmail.com>
Diagnozed-by: Tender Wang <tndrwang@gmail.com>
Diagnozed-by: Dean Rasheed <dean.a.rasheed@gmail.com>
Co-authored-by: Dean Rasheed <dean.a.rasheed@gmail.com>
Reviewed-by: Tender Wang <tndrwang@gmail.com>
Reviewed-by: Dean Rasheed <dean.a.rasheed@gmail.com>
Discussion: https://postgr.es/m/18830-1f31ea1dc930d444%40postgresql.org
Modules can use RegisterExtensionExplainOption to register new
EXPLAIN options, and GetExplainExtensionId, GetExplainExtensionState,
and SetExplainExtensionState to store related state inside the
ExplainState object.
Since this substantially increases the amount of code that needs
to handle ExplainState-related tasks, move a few bits of existing
code to a new file explain_state.c and add the rest of this
infrastructure there.
See the comments at the top of explain_state.c for further
explanation of how this mechanism works.
This does not yet provide a way for such such options to do anything
useful. The intention is that we'll add hooks for that purpose in a
separate commit.
Discussion: http://postgr.es/m/CA+TgmoYSzg58hPuBmei46o8D3SKX+SZoO4K_aGQGwiRzvRApLg@mail.gmail.com
Reviewed-by: Srinath Reddy <srinath2133@gmail.com>
Reviewed-by: Andrei Lepikhov <lepihov@gmail.com>
Reviewed-by: Tom Lane <tgl@sss.pgh.pa.us>
Reviewed-by: Sami Imseih <samimseih@gmail.com>
After pushing the bitmap iterator into table-AM specific code (as part
of making bitmap heap scan use the read stream API in 2b73a8cd33),
scan_bitmap_next_block() no longer returns the current block number.
Since scan_bitmap_next_block() isn't returning any relevant information
to bitmap table scan code, it makes more sense to get rid of it.
Now, bitmap table scan code only calls table_scan_bitmap_next_tuple(),
and the heap AM implementation of scan_bitmap_next_block() is a local
helper in heapam_handler.c.
Reviewed-by: Tomas Vondra <tomas@vondra.me>
Discussion: https://postgr.es/m/flat/CAAKRu_ZwCwWFeL_H3ia26bP2e7HiKLWt0ZmGXPVwPO6uXq0vaA%40mail.gmail.com
Make Bitmap Heap Scan use the read stream API instead of invoking
ReadBuffer() for each block indicated by the bitmap.
The read stream API handles prefetching, so remove all of the explicit
prefetching from bitmap heap scan code.
Now, heap table AM implements a read stream callback which uses the
bitmap iterator to return the next required block to the read stream
code.
Tomas Vondra conducted extensive regression testing of this feature.
Andres Freund, Thomas Munro, and I analyzed regressions and Thomas Munro
patched the read stream API.
Author: Melanie Plageman <melanieplageman@gmail.com>
Reviewed-by: Tomas Vondra <tomas@vondra.me>
Tested-by: Tomas Vondra <tomas@vondra.me>
Tested-by: Andres Freund <andres@anarazel.de>
Tested-by: Thomas Munro <thomas.munro@gmail.com>
Tested-by: Nazir Bilal Yavuz <byavuz81@gmail.com>
Discussion: https://postgr.es/m/flat/CAAKRu_ZwCwWFeL_H3ia26bP2e7HiKLWt0ZmGXPVwPO6uXq0vaA%40mail.gmail.com
Remove the TBMIterateResult member from the TBMPrivateIterator and
TBMSharedIterator and make tbm_[shared|private_]iterate() take a
TBMIterateResult as a parameter.
This allows tidbitmap API users to manage multiple TBMIterateResults per
scan. This is required for bitmap heap scan to use the read stream API,
with which there may be multiple I/Os in flight at once, each one with a
TBMIterateResult.
Reviewed-by: Tomas Vondra <tomas@vondra.me>
Discussion: https://postgr.es/m/d4bb26c9-fe07-439e-ac53-c0e244387e01%40vondra.me
We did not wake up on interrupts while waiting on async events on an
async-capable append node. For example, if you tried to cancel the
query, nothing would happen until one of the async subplans becomes
readable. To fix, add WL_LATCH_SET to the WaitEventSet.
Backpatch down to v14 where async Append execution was introduced.
Discussion: https://www.postgresql.org/message-id/37a40570-f558-40d3-b5ea-5c2079b3b30b@iki.fi
Expose the count of index searches/index descents in EXPLAIN ANALYZE's
output for index scan/index-only scan/bitmap index scan nodes. This
information is particularly useful with scans that use ScalarArrayOp
quals, where the number of index searches can be unpredictable due to
implementation details that interact with physical index characteristics
(at least with nbtree SAOP scans, since Postgres 17 commit 5bf748b8).
The information shown also provides useful context when EXPLAIN ANALYZE
runs a plan with an index scan node that successfully applied the skip
scan optimization (set to be added to nbtree by an upcoming patch).
The instrumentation works by teaching all index AMs to increment a new
nsearches counter whenever a new index search begins. The counter is
incremented at exactly the same point that index AMs already increment
the pg_stat_*_indexes.idx_scan counter (we're counting the same event,
but at the scan level rather than the relation level). Parallel queries
have workers copy their local counter struct into shared memory when an
index scan node ends -- even when it isn't a parallel aware scan node.
An earlier version of this patch that only worked with parallel aware
scans became commit 5ead85fb (though that was quickly reverted by commit
d00107cd following "debug_parallel_query=regress" buildfarm failures).
Our approach doesn't match the approach used when tracking other index
scan related costs (e.g., "Rows Removed by Filter:"). It is comparable
to the approach used in similar cases involving costs that are only
readily accessible inside an access method, not from the executor proper
(e.g., "Heap Blocks:" output for a Bitmap Heap Scan, which was recently
enhanced to show per-worker costs by commit 5a1e6df3, using essentially
the same scheme as the one used here). It is necessary for index AMs to
have direct responsibility for maintaining the new counter, since the
counter might need to be incremented multiple times per amgettuple call
(or per amgetbitmap call). But it is also necessary for the executor
proper to manage the shared memory now used to transfer each worker's
counter struct to the leader.
Author: Peter Geoghegan <pg@bowt.ie>
Reviewed-By: Robert Haas <robertmhaas@gmail.com>
Reviewed-By: Tomas Vondra <tomas@vondra.me>
Reviewed-By: Masahiro Ikeda <ikedamsh@oss.nttdata.com>
Reviewed-By: Matthias van de Meent <boekewurm+postgres@gmail.com>
Discussion: https://postgr.es/m/CAH2-WzkRqvaqR2CTNqTZP0z6FuL4-3ED6eQB0yx38XBNj1v-4Q@mail.gmail.com
Discussion: https://postgr.es/m/CAH2-Wz=PKR6rB7qbx+Vnd7eqeB5VTcrW=iJvAsTsKbdG+kW_UA@mail.gmail.com
Many STRICT function calls will have one or two arguments, in which
case we can speed up checking for NULL input by avoiding setting up
a loop over the arguments. This adds EEOP_FUNCEXPR_STRICT_1 and the
corresponding EEOP_FUNCEXPR_STRICT_2 for functions with one and two
arguments respectively.
Author: Andres Freund <andres@anarazel.de>
Co-authored-by: Daniel Gustafsson <daniel@yesql.se>
Reviewed-by: Andreas Karlsson <andreas@proxel.se>
Discussion: https://postgr.es/m/415721CE-7D2E-4B74-B5D9-1950083BA03E@yesql.se
Discussion: https://postgr.es/m/20191023163849.sosqbfs5yenocez3@alap3.anarazel.de
Knowing when the side-effects of an expression is the intended result
of the execution, rather than the returnvalue, is important for being
able generate more efficient JITed code. This replaces EEOP_DONE with
two new steps: EEOP_DONE_RETURN and EEOP_DONE_NO_RETURN. Expressions
which return a value should use the former step; expressions used for
their side-effects which don't return value should use the latter.
Author: Andres Freund <andres@anarazel.de>
Co-authored-by: Daniel Gustafsson <daniel@yesql.se>
Reviewed-by: Andreas Karlsson <andreas@proxel.se>
Discussion: https://postgr.es/m/415721CE-7D2E-4B74-B5D9-1950083BA03E@yesql.se
Discussion: https://postgr.es/m/20191023163849.sosqbfs5yenocez3@alap3.anarazel.de
The function calls GetLatestSnapshot() to acquire a fresh snapshot,
makes it active, and was meant to pass it to table_tuple_lock(), but
instead called GetLatestSnapshot() again to acquire yet another
snapshot. It was harmless because the heap AM and all other known
table AMs ignore the 'snapshot' argument anyway, but let's be tidy.
In the long run, this perhaps should be redesigned so that snapshot
was not needed in the first place. The table AM API uses TID +
snapshot as the unique identifier for the row version, which is
questionable when the row came from an index scan with a Dirty
snapshot. You might lock a different row version when you use a
different snapshot in the table_tuple_lock() call (a fresh MVCC
snapshot) than in the index scan (DirtySnapshot). However, in the heap
AM and other AMs where the TID alone identifies the row version, it
doesn't matter. So for now, just fix the obvious albeit harmless bug.
This has been wrong ever since the table AM API was introduced in
commit 5db6df0c01, so backpatch to all supported versions.
Discussion: https://www.postgresql.org/message-id/83d243d6-ad8d-4307-8b51-2ee5844f6230@iki.fi
Backpatch-through: 13
Stop comparing access method OID values against HASH_AM_OID and
BTREE_AM_OID, and instead check the IndexAmRoutine for an index to see
if it advertises its ability to perform the necessary ordering,
hashing, or cross-type comparing functionality. A field amcanorder
already existed, this uses it more widely. Fields amcanhash and
amcancrosscompare are added for the other purposes.
Author: Mark Dilger <mark.dilger@enterprisedb.com>
Discussion: https://www.postgresql.org/message-id/flat/E72EAA49-354D-4C2E-8EB9-255197F55330@enterprisedb.com
A non-leaf partition with a subplan that is an Append node was
omitted from PlannedStmt.unprunableRelids because it was mistakenly
included in PlannerGlobal.prunableRelids due to the way
PartitionedRelPruneInfo.leafpart_rti_map[] is constructed. This
happened when a non-leaf partition used an unflattened Append or
MergeAppend. As a result, ExecGetRangeTableRelation() reported an
error when called from CreatePartitionPruneState() to process the
partition's own PartitionPruneInfo, since it was treated as prunable
when it should not have been.
Reported-by: Alexander Lakhin <exclusion@gmail.com> (via sqlsmith)
Diagnosed-by: Tender Wang <tndrwang@gmail.com>
Reviewed-by: Tender Wang <tndrwang@gmail.com>
Discussion: https://postgr.es/m/74839af6-aadc-4f60-ae77-ae65f94bf607@gmail.com
The type argument wasn't actually really necessary. It was a remnant
of converting the API of the gist strategy translation from using
opclass to using opfamily+opcintype (commits c09e5a6a01,
622f678c10). For looking up the gist translation function, we used
the convention "amproclefttype = amprocrighttype = opclass's
opcintype" (see pg_amproc.h). But each operator family should only
have one translation function, and getting the right type for the
lookup is sometimes cumbersome and fragile, so this is all
unnecessarily complicated.
To simplify this, change the gist stategy support procedure to take
"any", "any" as argument. (This is arbitrary but seems intuitive.
The alternative of using InvalidOid as argument(s) upsets various DDL
commands, so it's not practical.) Then we don't need opcintype for
the lookup, and we can remove it from all the API layers introduced by
commit c09e5a6a01.
This also adds some more documentation about the correct signature of
the gist support function and adds more checks in gistvalidate().
This was previously underspecified. (It relied implicitly on
convention mentioned above.)
Discussion: https://www.postgresql.org/message-id/flat/E72EAA49-354D-4C2E-8EB9-255197F55330@enterprisedb.com
Before executing a cached generic plan, AcquireExecutorLocks() in
plancache.c locks all relations in a plan's range table to ensure the
plan is safe for execution. However, this locks runtime-prunable
relations that will later be pruned during "initial" runtime pruning,
introducing unnecessary overhead.
This commit defers locking for such relations to executor startup and
ensures that if the CachedPlan is invalidated due to concurrent DDL
during this window, replanning is triggered. Deferring these locks
avoids unnecessary locking overhead for pruned partitions, resulting
in significant speedup, particularly when many partitions are pruned
during initial runtime pruning.
* Changes to locking when executing generic plans:
AcquireExecutorLocks() now locks only unprunable relations, that is,
those found in PlannedStmt.unprunableRelids (introduced in commit
cbc127917e), to avoid locking runtime-prunable partitions
unnecessarily. The remaining locks are taken by
ExecDoInitialPruning(), which acquires them only for partitions that
survive pruning.
This deferral does not affect the locks required for permission
checking in InitPlan(), which takes place before initial pruning.
ExecCheckPermissions() now includes an Assert to verify that all
relations undergoing permission checks, none of which can be in the
set of runtime-prunable relations, are properly locked.
* Plan invalidation handling:
Deferring locks introduces a window where prunable relations may be
altered by concurrent DDL, invalidating the plan. A new function,
ExecutorStartCachedPlan(), wraps ExecutorStart() to detect and handle
invalidation caused by deferred locking. If invalidation occurs,
ExecutorStartCachedPlan() updates CachedPlan using the new
UpdateCachedPlan() function and retries execution with the updated
plan. To ensure all code paths that may be affected by this handle
invalidation properly, all callers of ExecutorStart that may execute a
PlannedStmt from a CachedPlan have been updated to use
ExecutorStartCachedPlan() instead.
UpdateCachedPlan() replaces stale plans in CachedPlan.stmt_list. A new
CachedPlan.stmt_context, created as a child of CachedPlan.context,
allows freeing old PlannedStmts while preserving the CachedPlan
structure and its statement list. This ensures that loops over
statements in upstream callers of ExecutorStartCachedPlan() remain
intact.
ExecutorStart() and ExecutorStart_hook implementations now return a
boolean value indicating whether plan initialization succeeded with a
valid PlanState tree in QueryDesc.planstate, or false otherwise, in
which case QueryDesc.planstate is NULL. Hook implementations are
required to call standard_ExecutorStart() at the beginning, and if it
returns false, they should do the same without proceeding.
* Testing:
To verify these changes, the delay_execution module tests scenarios
where cached plans become invalid due to changes in prunable relations
after deferred locks.
* Note to extension authors:
ExecutorStart_hook implementations must verify plan validity after
calling standard_ExecutorStart(), as explained earlier. For example:
if (prev_ExecutorStart)
plan_valid = prev_ExecutorStart(queryDesc, eflags);
else
plan_valid = standard_ExecutorStart(queryDesc, eflags);
if (!plan_valid)
return false;
<extension-code>
return true;
Extensions accessing child relations, especially prunable partitions,
via ExecGetRangeTableRelation() must now ensure their RT indexes are
present in es_unpruned_relids (introduced in commit cbc127917e), or
they will encounter an error. This is a strict requirement after this
change, as only relations in that set are locked.
The idea of deferring some locks to executor startup, allowing locks
for prunable partitions to be skipped, was first proposed by Tom Lane.
Reviewed-by: Robert Haas <robertmhaas@gmail.com> (earlier versions)
Reviewed-by: David Rowley <dgrowleyml@gmail.com> (earlier versions)
Reviewed-by: Tom Lane <tgl@sss.pgh.pa.us> (earlier versions)
Reviewed-by: Tomas Vondra <tomas@vondra.me>
Reviewed-by: Junwang Zhao <zhjwpku@gmail.com>
Discussion: https://postgr.es/m/CA+HiwqFGkMSge6TgC9KQzde0ohpAycLQuV7ooitEEpbKB0O_mg@mail.gmail.com
These functions should be called at most once per ResultRelInfo;
it's wasteful to do otherwise, and certainly the pattern of
opening twice and then closing twice is a bad idea. Moreover,
aminsertcleanup functions might not be prepared to be called twice,
as the just-hardened code in BRIN demonstrates.
This amounts to an API change, since such coding patterns were
safe even if wasteful before v17. Hence, apply to HEAD only.
(Extension code violating this new rule faces some risk in v17,
but we just fixed brininsertcleanup and there are probably few
other aminsertcleanup functions as yet. So the odds of breaking
usable code seem higher than the odds of doing something useful
with a back-patch.)
Bug: #18815
Reported-by: Sergey Belyashov <sergey.belyashov@gmail.com>
Discussion: https://postgr.es/m/18815-2a0407cc7f40b327@postgresql.org
Until now ExecChooseHashTableSize() considered only the size of the
in-memory hash table, and ignored the memory needed for the batch files.
Which can be a significant amount, because each batch needs two BufFiles
(each with a BLCKSZ buffer). The same issue applies to increasing the
number of batches during execution.
It's also possible to trigger a "batch explosion", e.g. due to duplicate
values or skew. We've seen reports of joins with hundreds of thousands
(or even millions) of batches, consuming gigabytes of memory, triggering
OOM errors. These cases may be fairly rare, but it's clearly possible to
hit them.
These issues can't be prevented during planning. Even if we improve
that, it does not help with execution-time batch explosion. We can
however reduce the impact and use as little memory as possible.
This patch improves the behavior by adjusting how the memory is divided
between the hash table and batch files. It may be better to use fewer
batch files, even if it means the hash table will exceed the limit.
The capacity of the hash node may be increased either by doubling he
number of batches, or doubling the size of the in-memory hash table. The
outcome is the same, but the memory usage may be very different. For low
nbatch values it's better to add batches, for high nbatch values it's
better to allow a larger hash table.
The patch considers both options, both during the initial sizing and
then during execution, to minimize how much the limit gets exceeded.
It might seem this patch is relaxing the memory limit - allowing it to
be exceeded. But that's not really the case. It has always been like
that, except the memory used by batches was ignored.
Allowing the hash table to grow may also prevent the batch explosion.
If there's a large batch that can't be split (due to hash collisions or
duplicate values), at some point the memory limit will increase enough
for the batch to fit into the hash table.
This patch was in the works for a long time. The early versions were
posted in 2019, and revived every year or two when we happened to get
the next report of OOM due to a hashjoin batch explosion. Each of those
patch versions were reviewed by a couple people. I'm mentioning only
Melanie Plageman and Robert Haas, because they reviewed the last
version, and the older patches are very different.
Reviewed-by: Melanie Plageman, Robert Haas
Discussion: https://postgr.es/m/7bed6c08-72a0-4ab9-a79c-e01fcdd0940f@vondra.me
Discussion: https://postgr.es/m/20190504003414.bulcbnge3rhwhcsh%40development
Discussion: https://postgr.es/m/20190428141901.5dsbge2ka3rxmpk6%40development
ExecInitModifyTable() forgot to trim MERGE-related lists to exclude
entries for result relations pruned during initial pruning, so fix
that.
While at it, make the function's use of the pruned resultRelations
list, rather than ModifyTable.resultRelations, more consistent.
Reported-by: Alexander Lakhin <exclusion@gmail.com> (via sqlsmith)
Reviewed-by: Junwang Zhao <zhjwpku@gmail.com>
Discussion: https://postgr.es/m/e72c94d9-e5f9-4753-9bc1-69d72bd54b8a@gmail.com
wal_buffers_full has been introduced in pg_stat_wal in 8d9a935965, as
some information providing metrics for the tuning of the GUC
wal_buffers. WalUsage has been introduced before that in df3b181499.
Moving this field is proving to be beneficial for several reasons:
- This information can now be made available in more layers, providing
more granularity than just pg_stat_wal, on a per-query basis: EXPLAIN,
pgss and VACUUM/ANALYZE logs.
- A patch is under discussion to provide statistics for WAL at backend
level, and this move simplifies a bit the handling of pending
statistics. The remaining data in PgStat_PendingWalStats now relates to
write/sync counters and times, with equivalents present in pg_stat_io,
that backend statistics are able to already track. So this should cut
all the dependencies between PgStat_PendingWalStats and WAL stats at
backend level.
As of this change, wal_buffers_full only shows in pg_stat_wal.
Author: Bertrand Drouvot
Reviewed-by: Ilia Evdokimov
Discussion: https://postgr.es/m/Z6SOha5YFFgvpwQY@ip-10-97-1-34.eu-west-3.compute.internal
Commit 27cc7cd2bc accidentally placed the assertion ensuring
that the pointer isn't NULL after it had already been accessed.
Fix by moving the pointer dereferencing to after the assertion.
Backpatch to all supported branches.
Author: Dmitry Koval <d.koval@postgrespro.ru>
Reviewed-by: Daniel Gustafsson <daniel@yesql.se>
Reviewed-by: Michael Paquier <michael@paquier.xyz>
Discussion: https://postgr.es/m/1618848d-cdc7-414b-9c03-08cf4bef4408@postgrespro.ru
Backpatch-through: 13
This adds a new variant of generated columns that are computed on read
(like a view, unlike the existing stored generated columns, which are
computed on write, like a materialized view).
The syntax for the column definition is
... GENERATED ALWAYS AS (...) VIRTUAL
and VIRTUAL is also optional. VIRTUAL is the default rather than
STORED to match various other SQL products. (The SQL standard makes
no specification about this, but it also doesn't know about VIRTUAL or
STORED.) (Also, virtual views are the default, rather than
materialized views.)
Virtual generated columns are stored in tuples as null values. (A
very early version of this patch had the ambition to not store them at
all. But so much stuff breaks or gets confused if you have tuples
where a column in the middle is completely missing. This is a
compromise, and it still saves space over being forced to use stored
generated columns. If we ever find a way to improve this, a bit of
pg_upgrade cleverness could allow for upgrades to a newer scheme.)
The capabilities and restrictions of virtual generated columns are
mostly the same as for stored generated columns. In some cases, this
patch keeps virtual generated columns more restricted than they might
technically need to be, to keep the two kinds consistent. Some of
that could maybe be relaxed later after separate careful
considerations.
Some functionality that is currently not supported, but could possibly
be added as incremental features, some easier than others:
- index on or using a virtual column
- hence also no unique constraints on virtual columns
- extended statistics on virtual columns
- foreign-key constraints on virtual columns
- not-null constraints on virtual columns (check constraints are supported)
- ALTER TABLE / DROP EXPRESSION
- virtual column cannot have domain type
- virtual columns are not supported in logical replication
The tests in generated_virtual.sql have been copied over from
generated_stored.sql with the keyword replaced. This way we can make
sure the behavior is mostly aligned, and the differences can be
visible. Some tests for currently not supported features are
currently commented out.
Reviewed-by: Jian He <jian.universality@gmail.com>
Reviewed-by: Dean Rasheed <dean.a.rasheed@gmail.com>
Tested-by: Shlok Kyal <shlok.kyal.oss@gmail.com>
Discussion: https://www.postgresql.org/message-id/flat/a368248e-69e4-40be-9c07-6c3b5880b0a6@eisentraut.org
This commit introduces changes to track unpruned relations explicitly,
making it possible for top-level plan nodes, such as ModifyTable and
LockRows, to avoid processing partitions pruned during initial
pruning. Scan-level nodes, such as Append and MergeAppend, already
avoid the unnecessary processing by accessing partition pruning
results directly via part_prune_index. In contrast, top-level nodes
cannot access pruning results directly and need to determine which
partitions remain unpruned.
To address this, this commit introduces a new bitmapset field,
es_unpruned_relids, which the executor uses to track the set of
unpruned relations. This field is referenced during plan
initialization to skip initializing certain nodes for pruned
partitions. It is initialized with PlannedStmt.unprunableRelids,
a new field that the planner populates with RT indexes of relations
that cannot be pruned during runtime pruning. These include relations
not subject to partition pruning and those required for execution
regardless of pruning.
PlannedStmt.unprunableRelids is computed during set_plan_refs() by
removing the RT indexes of runtime-prunable relations, identified
from PartitionPruneInfos, from the full set of relation RT indexes.
ExecDoInitialPruning() then updates es_unpruned_relids by adding
partitions that survive initial pruning.
To support this, PartitionedRelPruneInfo and PartitionedRelPruningData
now include a leafpart_rti_map[] array that maps partition indexes to
their corresponding RT indexes. The former is used in set_plan_refs()
when constructing unprunableRelids, while the latter is used in
ExecDoInitialPruning() to convert partition indexes returned by
get_matching_partitions() into RT indexes, which are then added to
es_unpruned_relids.
These changes make it possible for ModifyTable and LockRows nodes to
process only relations that remain unpruned after initial pruning.
ExecInitModifyTable() trims lists, such as resultRelations,
withCheckOptionLists, returningLists, and updateColnosLists, to
consider only unpruned partitions. It also creates ResultRelInfo
structs only for these partitions. Similarly, child RowMarks for
pruned relations are skipped.
By avoiding unnecessary initialization of structures for pruned
partitions, these changes improve the performance of updates and
deletes on partitioned tables during initial runtime pruning.
Due to ExecInitModifyTable() changes as described above, EXPLAIN on a
plan for UPDATE and DELETE that uses runtime initial pruning no longer
lists partitions pruned during initial pruning.
Reviewed-by: Robert Haas <robertmhaas@gmail.com> (earlier versions)
Reviewed-by: Tomas Vondra <tomas@vondra.me>
Discussion: https://postgr.es/m/CA+HiwqFGkMSge6TgC9KQzde0ohpAycLQuV7ooitEEpbKB0O_mg@mail.gmail.com
This turns GistTranslateCompareType() into a callback function of the
gist index AM instead of a standalone function. The existing callers
are changed to use IndexAmTranslateCompareType(). This then makes
that code not hardcoded toward gist.
This means in particular that the temporal keys code is now
independent of gist. Also, this generalizes commit 74edabce7a, so
other index access methods other than the previously hardcoded ones
could now work as REPLICA IDENTITY in a logical replication
subscriber.
Author: Mark Dilger <mark.dilger@enterprisedb.com>
Co-authored-by: Peter Eisentraut <peter@eisentraut.org>
Discussion: https://www.postgresql.org/message-id/flat/E72EAA49-354D-4C2E-8EB9-255197F55330@enterprisedb.com
Follow up to commit 630f9a43ce. The previous name had become
confusing, because it doesn't actually translate a strategy number but
a CompareType into a strategy number. We might add the inverse at
some point, which would then probably be called something like
GistTranslateStratnum.
Reviewed-by: Mark Dilger <mark.dilger@enterprisedb.com>
Discussion: https://www.postgresql.org/message-id/flat/E72EAA49-354D-4C2E-8EB9-255197F55330@enterprisedb.com
Consistently use "Size" (or size_t, or in some places int64 or double)
as the type for variables holding memory allocation sizes. In most
places variables' data types were fine already, but we had an ancient
habit of computing bytes from kilobytes-units GUCs with code like
"work_mem * 1024L". That risks overflow on Win64 where they did not
make "long" as wide as "size_t". We worked around that by restricting
such GUCs' ranges, so you couldn't set work_mem et al higher than 2GB
on Win64. This patch removes that restriction, after replacing such
calculations with "work_mem * (Size) 1024" or variants of that.
It should be noted that this patch was constructed by searching
outwards from the GUCs that have MAX_KILOBYTES as upper limit.
So I can't positively guarantee there are no other places doing
memory-size arithmetic in int or long variables. I do however feel
pretty confident that increasing MAX_KILOBYTES on Win64 is safe now.
Also, nothing in our code should be dealing in multiple-gigabyte
allocations without authorization from a relevant GUC, so it seems
pretty likely that this search caught everything that could be at
risk of overflow.
Author: Vladlen Popolitov <v.popolitov@postgrespro.ru>
Co-authored-by: Tom Lane <tgl@sss.pgh.pa.us>
Discussion: https://postgr.es/m/1a01f0-66ec2d80-3b-68487680@27595217
This commit builds on the prior change that moved PartitionPruneInfos
out of individual plan nodes into a list in PlannedStmt, making it
possible to initialize PartitionPruneStates without traversing the
plan tree and perform runtime initial pruning before ExecInitNode()
initializes the plan trees. These tasks are now handled in a new
routine, ExecDoInitialPruning(), which is called by InitPlan()
before calling ExecInitNode() on various plan trees.
ExecDoInitialPruning() performs the initial pruning and saves the
result -- a Bitmapset of indexes for surviving child subnodes -- in
es_part_prune_results, a list in EState.
PartitionPruneStates created for initial pruning are stored in
es_part_prune_states, another list in EState, for later use during
exec pruning. Both lists are parallel to es_part_prune_infos, which
holds the PartitionPruneInfos from PlannedStmt, enabling shared
indexing.
PartitionPruneStates initialized in ExecDoInitialPruning() now
include only the PartitionPruneContexts for initial pruning steps.
Exec pruning contexts are initialized later in
ExecInitPartitionExecPruning() when the parent plan node is
initialized, as the exec pruning step expressions depend on the parent
node's PlanState.
The existing function PartitionPruneFixSubPlanMap() has been
repurposed for this initialization to avoid duplicating a similar
loop structure for finding PartitionedRelPruningData to initialize
exec pruning contexts for. It has been renamed to
InitExecPruningContexts() to reflect its new primary responsibility.
The original logic to "fix subplan maps" remains intact but is now
encapsulated within the renamed function.
This commit removes two obsolete Asserts in partkey_datum_from_expr().
The ExprContext used for pruning expression evaluation is now
independent of the parent PlanState, making these Asserts unnecessary.
By centralizing pruning logic and decoupling it from the plan
initialization step (ExecInitNode()), this change sets the stage for
future patches that will use the result of initial pruning to
save the overhead of redundant processing for pruned partitions.
Reviewed-by: Robert Haas <robertmhaas@gmail.com>
Reviewed-by: Tomas Vondra <tomas@vondra.me>
Discussion: https://postgr.es/m/CA+HiwqFGkMSge6TgC9KQzde0ohpAycLQuV7ooitEEpbKB0O_mg@mail.gmail.com
Add BitmapTableScanSetup(), a helper which contains all of the code that
must be done on every scan of the table in a bitmap table scan. This
includes scanning the index, building the bitmap, and setting up the
scan descriptors.
Pushing this setup into a helper function makes BitmapHeapNext() more
readable.
Reviewed-by: Nazir Bilal Yavuz <byavuz81@gmail.com>
Discussion: https://postgr.es/m/CAN55FZ1vXu%2BZdT0_MM-i1vbTdfHHf0KR3cK6R5gs6dNNNpyrJw%40mail.gmail.com
Instead of deciding at runtime whether to read from casetest.value
or caseValue_datum, split EEOP_CASE_TESTVAL into two opcodes and
make the decision during expression compilation. Similarly for
EEOP_DOMAIN_TESTVAL. This actually results in net less code,
mainly because llvmjit_expr.c's code for handling these opcodes
gets shorter. The performance gain is doubtless negligible, but
this seems worth changing anyway on grounds of simplicity and
understandability.
Author: Andreas Karlsson <andreas@proxel.se>
Co-authored-by: Xing Guo <higuoxing@gmail.com>
Reviewed-by: Tom Lane <tgl@sss.pgh.pa.us>
Discussion: https://postgr.es/m/CACpMh+AiBYAWn+D1aU7Rsy-V1tox06Cbc0H3qA7rwL5zdJ=anQ@mail.gmail.com
This moves PartitionPruneInfo from plan nodes to PlannedStmt,
simplifying traversal by centralizing all PartitionPruneInfo
structures in a single list in it, which holds all instances for the
main query and its subqueries. Instead of plan nodes (Append or
MergeAppend) storing PartitionPruneInfo pointers, they now reference
an index in this list.
A bitmapset field is added to PartitionPruneInfo to store the RT
indexes corresponding to the apprelids field in Append or MergeAppend.
This allows execution pruning logic to verify that it operates on the
correct plan node, mainly to facilitate debugging.
Duplicated code in set_append_references() and
set_mergeappend_references() is refactored into a new function,
register_pruneinfo(). This updates RT indexes by applying rtoffet
and adds PartitionPruneInfo to the global list in PlannerGlobal.
By allowing pruning to be performed without traversing the plan tree,
this change lays the groundwork for runtime initial pruning to occur
independently of plan tree initialization.
Reviewed-by: Alvaro Herrera <alvherre@alvh.no-ip.org> (earlier version)
Reviewed-by: Robert Haas <robertmhaas@gmail.com>
Reviewed-by: Tomas Vondra <tomas@vondra.me>
Discussion: https://postgr.es/m/CA+HiwqFGkMSge6TgC9KQzde0ohpAycLQuV7ooitEEpbKB0O_mg@mail.gmail.com
This commit refactors ExecScan() by moving its tuple-fetching,
filtering, and projection logic into an inline-able function,
ExecScanExtended(), defined in src/include/executor/execScan.h.
ExecScanExtended() accepts parameters for EvalPlanQual state,
qualifiers (ExprState), and projection (ProjectionInfo).
Specialized variants of the execution function of a given Scan node
(for example, ExecSeqScan() for SeqScan) can then pass const-NULL for
unused parameters. This allows the compiler to inline the logic and
eliminate unnecessary branches or checks. Each variant function thus
contains only the necessary code, optimizing execution for scans
where these features are not needed.
The variant function to be used is determined in the ExecInit*()
function of the node and assigned to the ExecProcNode function pointer
in the node's PlanState, effectively turning runtime checks and
conditional branches on the NULLness of epqstate, qual, and projInfo
into static ones, provided the compiler successfully eliminates
unnecessary checks from the inlined code of ExecScanExtended().
Currently, only ExecSeqScan() is modified to take advantage of this
inline-ability. Other Scan nodes might benefit from such specialized
variant functions but that is left as future work.
Benchmarks performed by Junwang Zhao, David Rowley and myself show up
to a 5% reduction in execution time for queries that rely heavily on
Seq Scans. The most significant improvements were observed in
scenarios where EvalPlanQual, qualifiers, and projection were not
required, but other cases also benefit from reduced runtime overhead
due to the inlining and removal of unnecessary code paths.
The idea for this patch first came from Andres Freund in an off-list
discussion. The refactoring approach implemented here is based on a
proposal by David Rowley, significantly improving upon the patch I
(amitlan) initially proposed.
Suggested-by: Andres Freund <andres@anarazel.de>
Co-authored-by: David Rowley <dgrowleyml@gmail.com>
Reviewed-by: David Rowley <dgrowleyml@gmail.com>
Reviewed-by: Junwang Zhao <zhjwpku@gmail.com>
Tested-by: Junwang Zhao <zhjwpku@gmail.com>
Tested-by: David Rowley <dgrowleyml@gmail.com>
Discussion: https://postgr.es/m/CA+HiwqGaH-otvqW_ce-paL=96JvU4j+Xbuk+14esJNDwefdkOg@mail.gmail.com
This allows the RETURNING list of INSERT/UPDATE/DELETE/MERGE queries
to explicitly return old and new values by using the special aliases
"old" and "new", which are automatically added to the query (if not
already defined) while parsing its RETURNING list, allowing things
like:
RETURNING old.colname, new.colname, ...
RETURNING old.*, new.*
Additionally, a new syntax is supported, allowing the names "old" and
"new" to be changed to user-supplied alias names, e.g.:
RETURNING WITH (OLD AS o, NEW AS n) o.colname, n.colname, ...
This is useful when the names "old" and "new" are already defined,
such as inside trigger functions, allowing backwards compatibility to
be maintained -- the interpretation of any existing queries that
happen to already refer to relations called "old" or "new", or use
those as aliases for other relations, is not changed.
For an INSERT, old values will generally be NULL, and for a DELETE,
new values will generally be NULL, but that may change for an INSERT
with an ON CONFLICT ... DO UPDATE clause, or if a query rewrite rule
changes the command type. Therefore, we put no restrictions on the use
of old and new in any DML queries.
Dean Rasheed, reviewed by Jian He and Jeff Davis.
Discussion: https://postgr.es/m/CAEZATCWx0J0-v=Qjc6gXzR=KtsdvAE7Ow=D=mu50AgOe+pvisQ@mail.gmail.com
This changes commit 7406ab623f in that the gist strategy number
mapping support function is changed to use the CompareType enum as
input, instead of the "well-known" RT*StrategyNumber strategy numbers.
This is a bit cleaner, since you are not dealing with two sets of
strategy numbers. Also, this will enable us to subsume this system
into a more general system of using CompareType to define operator
semantics across index methods.
Discussion: https://www.postgresql.org/message-id/flat/E72EAA49-354D-4C2E-8EB9-255197F55330@enterprisedb.com
RowCompareType served as a way to describe the fundamental meaning of
an operator, notionally independent of an operator class (although so
far this was only really supported for btrees). Its original purpose
was for use inside RowCompareExpr, and it has also found some small
use outside, such as for get_op_btree_interpretation().
We want to expand this now, as a more general way to describe operator
semantics for other index access methods, including gist (to improve
GistTranslateStratnum()) and others not written yet. To avoid future
confusion, we rename the type to CompareType and the symbols from
ROWCOMPARE_XXX to COMPARE_XXX to reflect their more general purpose.
Reviewed-by: Mark Dilger <mark.dilger@enterprisedb.com>
Discussion: https://www.postgresql.org/message-id/flat/E72EAA49-354D-4C2E-8EB9-255197F55330@enterprisedb.com
This adds support for the NOT ENFORCED/ENFORCED flag for constraints,
with support for check constraints.
The plan is to eventually support this for foreign key constraints,
where it is typically more useful.
Note that CHECK constraints do not currently support ALTER operations,
so changing the enforceability of an existing constraint isn't
possible without dropping and recreating it. This could be added
later.
Author: Amul Sul <amul.sul@enterprisedb.com>
Reviewed-by: Peter Eisentraut <peter@eisentraut.org>
Reviewed-by: jian he <jian.universality@gmail.com>
Tested-by: Triveni N <triveni.n@enterprisedb.com>
Discussion: https://www.postgresql.org/message-id/flat/CAAJ_b962c5AcYW9KUt_R_ER5qs3fUGbe4az-SP-vuwPS-w-AGA@mail.gmail.com
Previously, the caller needed to allocate the memory and the
TupleHashTable would store a pointer to it. That wastes space for the
palloc overhead as well as the size of the pointer itself.
Now, the TupleHashTable relies on the caller to correctly specify the
additionalsize, and allocates that amount of space. The caller can
then request a pointer into that space.
Discussion: https://postgr.es/m/b9cbf0219a9859dc8d240311643ff4362fd9602c.camel@j-davis.com
Reviewed-by: Heikki Linnakangas
CHUNKHDRSZ was defined as 16 bytes, which was true when that code went in,
but since c6e0fe1f2, 8 is a more accurate value. Here we adjust it to use
sizeof(MemoryChunk), which is normally 8, or 16 for cassert builds.
c6e0fe1f2 first appeared in v16, so this is technically wrong in v16 up
to master, but let's apply this only to master as adjusting this does
influence the estimated number of batches in the aggregate costing code
and we don't want to cause plan instability in released versions.
Reviewed-by: Tom Lane
Discussion: https://postgr.es/m/CAApHDvpMpRQvsTqZo3FinXkgytwxwF8sCyZm83xDj-1s_hLe+w@mail.gmail.com
This adjusts slot_deform_heap_tuple() to add special-case loops to
eliminate much of the branching that was done within the body of the
main deform loop.
Previously, while looping over each attribute to deform,
slot_deform_heap_tuple() would always recheck if the given attribute was
NULL by looking at HeapTupleHasNulls() and if so, went on to check the
tuple's NULL bitmap. Since many tuples won't contain any NULLs, we can
just check HeapTupleHasNulls() once and when there are no NULLs, use a
more compact version of the deforming loop which contains no NULL checking
code at all.
The same is possible for the "slow" mode checking part of the loop. That
variable was checked several times for each attribute, once to determine
if the offset to the attribute value could be taken from the attcacheoff,
and again to check if the offset could be cached for next time.
These "slow" checks can mostly be eliminated by instead having multiple
loops. Initially, we can start in the non-slow loop and break out of
that loop if and only if we must stop caching the offset. This
eliminates branching for both slow and non-slow deforming methods. The
amount of code required for the no nulls / non-slow version is very
small. It's possible to have separate loops like this due to the fact
that once we move into slow mode, we never need to switch back into
non-slow mode for a given tuple.
We have the compiler take care of writing out the multiple required
loops by having a pg_attribute_always_inline function which gets called
various times passing in constant values for the "slow" and "hasnulls"
parameters. This allows the compiler to eliminate const-false branches
and remove comparisons for const-true ones.
This commit has shown overall query performance increases of around 5-20%
in deform-heavy OLAP-type workloads.
Author: David Rowley
Reviewed-by: Victor Yegorov
Discussion: https://postgr.es/m/CAGnEbog92Og2CpC2S8=g_HozGsWtt_3kRS1sXjLz0jKSoCNfLw@mail.gmail.com
Discussion: https://postgr.es/m/CAApHDvo9e0XG71WrefYaRv5n4xNPLK4k8LjD0mSR3c9KR2vi2Q@mail.gmail.com
Here we convert CompactAttribute.attalign from a char, which is directly
derived from pg_attribute.attalign into a uint8, which stores the number
of bytes to align the column's value by in the tuple.
This allows tuple deformation and tuple size calculations to move away
from using the inefficient att_align_nominal() macro, which manually
checks each TYPALIGN_* char to translate that into the alignment bytes
for the given type. Effectively, this commit changes those to TYPEALIGN
calls, which are branchless and only perform some simple arithmetic with
some bit-twiddling.
The removed branches were often mispredicted by CPUs, especially so in
real-world tables which often contain a mishmash of different types
with different alignment requirements.
Author: David Rowley
Reviewed-by: Andres Freund, Victor Yegorov
Discussion: https://postgr.es/m/CAApHDvrBztXP3yx=NKNmo3xwFAFhEdyPnvrDg3=M0RhDs+4vYw@mail.gmail.com
The new compact_attrs array stores a few select fields from
FormData_pg_attribute in a more compact way, using only 16 bytes per
column instead of the 104 bytes that FormData_pg_attribute uses. Using
CompactAttribute allows performance-critical operations such as tuple
deformation to be performed without looking at the FormData_pg_attribute
element in TupleDesc which means fewer cacheline accesses.
For some workloads, tuple deformation can be the most CPU intensive part
of processing the query. Some testing with 16 columns on a table
where the first column is variable length showed around a 10% increase in
transactions per second for an OLAP type query performing aggregation on
the 16th column. However, in certain cases, the increases were much
higher, up to ~25% on one AMD Zen4 machine.
This also makes pg_attribute.attcacheoff redundant. A follow-on commit
will remove it, thus shrinking the FormData_pg_attribute struct by 4
bytes.
Author: David Rowley
Reviewed-by: Andres Freund, Victor Yegorov
Discussion: https://postgr.es/m/CAApHDvrBztXP3yx=NKNmo3xwFAFhEdyPnvrDg3=M0RhDs+4vYw@mail.gmail.com
It was reasonable to preserve the old API of BuildTupleHashTable()
in the back branches, but in HEAD we should actively discourage use
of that version. There are no remaining callers in core, so just
get rid of it. Then rename BuildTupleHashTableExt() back to
BuildTupleHashTable().
While at it, fix up the miserably-poorly-maintained header comment
for BuildTupleHashTable[Ext]. It looks like more than one patch in
this area has had the opinion that updating comments is beneath them.
Discussion: https://postgr.es/m/538343.1734646986@sss.pgh.pa.us
Append, MergeAppend, and RecursiveUnion can all use the support
functions added in commit 276279295. The first two can report a
fixed result slot type if all their children return the same fixed
slot type. That does nothing for the append step itself, but might
allow optimizations in the parent plan node. RecursiveUnion can
optimize tuple hash table operations in the same way as SetOp now
does.
Patch by me; thanks to Richard Guo and David Rowley for review.
Discussion: https://postgr.es/m/1850138.1731549611@sss.pgh.pa.us
The original design for set operations involved appending the two
input relations into one and adding a flag column that allows
distinguishing which side each row came from. Then the SetOp node
pries them apart again based on the flag. This is bizarre. The
only apparent reason to do it is that when sorting, we'd only need
one Sort node not two. But since sorting is at least O(N log N),
sorting all the data is actually worse than sorting each side
separately --- plus, we have no chance of taking advantage of
presorted input. On top of that, adding the flag column frequently
requires an additional projection step that adds cycles, and then
the Append node isn't free either. Let's get rid of all of that
and make the SetOp node have two separate children, using the
existing outerPlan/innerPlan infrastructure.
This initial patch re-implements nodeSetop.c and does a bare minimum
of work on the planner side to generate correctly-shaped plans.
In particular, I've tried not to change the cost estimates here,
so that the visible changes in the regression test results will only
involve removal of useless projection steps and not any changes in
whether to use sorted vs hashed mode.
For SORTED mode, we combine successive identical tuples from each
input into groups, and then merge-join the groups. The tuple
comparisons now use SortSupport instead of simple equality, but
the group-formation part should involve roughly the same number of
tuple comparisons as before. The cross-comparisons between left and
right groups probably add to that, but I'm not sure to quantify how
many more comparisons we might need.
For HASHED mode, nodeSetop's logic is almost the same as before,
just refactored into two separate loops instead of one loop that
has an assumption that it will see all the left-hand inputs first.
In both modes, I added early-exit logic to not bother reading the
right-hand relation if the left-hand input is empty, since neither
INTERSECT nor EXCEPT modes can produce any output if the left input
is empty. This could have been done before in the hashed mode, but
not in sorted mode. Sorted mode can also stop as soon as it exhausts
the left input; any remaining right-hand tuples cannot have matches.
Also, this patch adds some infrastructure for detecting whether
child plan nodes all output the same type of tuple table slot.
If they do, the hash table logic can use slightly more efficient
code based on assuming that that's the input slot type it will see.
We'll make use of that infrastructure in other plan node types later.
Patch by me; thanks to Richard Guo and David Rowley for review.
Discussion: https://postgr.es/m/1850138.1731549611@sss.pgh.pa.us
1a0da347a7 replaced Bitmap Table Scan's separate private and
shared bitmap iterators with a unified iterator. It accidentally set up
the prefetch iterator twice for non-parallel bitmap table scans. Remove
the extra set up call to tbm_begin_iterate().
1a0da347a7 replaced Bitmap Table Scan's individual private and
shared iterators with a unified iterator. It neglected, however, to
check if the iterator had already been cleaned up before doing so on
rescan. Add this check both on rescan and end scan to be safe.
Reported-by: Richard Guo
Author: Richard Guo
Discussion: https://postgr.es/m/CAMbWs48nrhcLY1kcd-u9oD%2B6yiS631F_8Fx8ZGsO-BYDwH%2Bbyw%40mail.gmail.com
8f4ee9626 fixed an old Assert failure that could happen when the slot
type used to look up the hash table for BuildTupleHashTableExt() users
wasn't a TTSOpsMinimalTuple slot. The fix for that in the back branches
had to be to pass the TupleTableSlotOps as NULL, however in master,
since we have the inputOps parameter as was added by d96d1d515, we can
pass that down instead.
At least one caller uses a fixed slot that's always TTSOpsVirtual, so
passing down inputOps for these cases allows ExecBuildGroupingEqual() to
skip adding the EEOP_INNER_FETCHSOME ExprEvalStep.
This should increase the performance of hashed subplans very slightly.
Author: Tom Lane, David Rowley
Discussion: https://postgr.es/m/2543667.1734483723@sss.pgh.pa.us
If the non-recursive part of a recursive CTE ended up using
TTSOpsBufferHeapTuple as the table slot type, then a duplicate value
could cause an Assert failure in CheckOpSlotCompatibility() when
checking the hash table for the duplicate value. The expected slot type
for the deform step was TTSOpsMinimalTuple so the Assert failed when the
TTSOpsBufferHeapTuple slot was used.
This is a long-standing bug which we likely didn't notice because it
seems much more likely that the non-recursive term would have required
projection and used a TTSOpsVirtual slot, which CheckOpSlotCompatibility
is ok with.
There doesn't seem to be any harm done here other than the Assert
failure. Both TTSOpsMinimalTuple and TTSOpsBufferHeapTuple slot types
require tuple deformation, so the EEOP_*_FETCHSOME ExprState step would
have properly existed in the ExprState.
The solution is to pass NULL for the ExecBuildGroupingEqual's 'lops'
parameter. This means the ExprState's EEOP_*_FETCHSOME step won't
expect a fixed slot type. This makes CheckOpSlotCompatibility() happy as
no checking is performed when the ExprEvalStep is not expecting a fixed
slot type.
Reported-by: Richard Guo
Reviewed-by: Tom Lane
Discussion: https://postgr.es/m/CAMbWs4-8U9q2LAtf8+ghV11zeUReA3AmrYkxzBEv0vKnDxwkKA@mail.gmail.com
Backpatch-through: 13, all supported versions
With the repurposing of TBMIterator as an interface for both parallel
and serial iteration through TIDBitmaps in commit 7f9d4187e7,
bitmap table scans may now use it.
Modify bitmap table scan code to use the TBMIterator. This requires
moving around a bit of code, so a few variables are initialized
elsewhere.
Author: Melanie Plageman
Reviewed-by: Tomas Vondra
Discussion: https://postgr.es/m/c736f6aa-8b35-4e20-9621-62c7c82e2168%40vondra.me
Add and use TBMPrivateIterator, which replaces the current TBMIterator
for serial use cases, and repurpose TBMIterator to be a unified
interface for both the serial ("private") and parallel ("shared") TID
Bitmap iterator interfaces. This encapsulation simplifies call sites for
callers supporting both parallel and serial TID Bitmap access.
TBMIterator is not yet used in this commit.
Author: Melanie Plageman
Reviewed-by: Tomas Vondra, Heikki Linnakangas
Discussion: https://postgr.es/m/063e4eb4-32d9-439e-a0b1-75565a9835a8%40iki.fi
0f5738202 adjusted the execGrouping.c code so it made use of ExprStates to
generate hash values. That commit made a wrong assumption that the slot
type to pass to ExecBuildHash32FromAttrs() is always &TTSOpsMinimalTuple.
That's not the case as the slot type depends on the slot type passed to
LookupTupleHashEntry(), which for nodeRecursiveunion.c, could be any of
the current slot types.
Here we fix this by adding a new parameter to BuildTupleHashTableExt()
to allow the slot type to be passed in. In the case of nodeSubplan.c
and nodeAgg.c the slot type is always &TTSOpsVirtual, so for both of
those cases, it's beneficial to pass the known slot type as that allows
ExecBuildHash32FromAttrs() to skip adding the tuple deform step to the
resulting ExprState. Another possible fix would have been to have
ExecBuildHash32FromAttrs() set "fetch.kind" to NULL so that
ExecComputeSlotInfo() always determines the EEOP_INNER_FETCHSOME is
required, however, that option isn't favorable as slows down aggregation
and hashed subplan evaluation due to the extra (needless) deform step.
Thanks to Nathan Bossart for bisecting to find the offending commit
based on Paul's report.
Reported-by: Paul Ramsey <pramsey@cleverelephant.ca>
Discussion: https://postgr.es/m/99F064C1-B3EB-4BE7-97D2-D2A0AA487A71@cleverelephant.ca
This speeds up obtaining hash values for GROUP BY and hashed SubPlans by
using the ExprState support for hashing, thus allowing JIT compilation for
obtaining hash values for these operations.
This, even without JIT compilation, has been shown to improve Hash
Aggregate performance in some cases by around 15% and hashed NOT IN
queries in one case by over 30%, however, real-world cases are likely to
see smaller gains as the test cases used were purposefully designed to
have high hashing overheads by keeping the hash table small to prevent
additional memory overheads that would be a factor when working with large
hash tables.
In passing, fix a hypothetical bug in ExecBuildHash32Expr() so that the
initial value is stored directly in the ExprState's result field if
there are no expressions to hash. None of the current users of this
function use an initial value, so the bug is only hypothetical.
Reviewed-by: Andrei Lepikhov <lepihov@gmail.com>
Discussion: https://postgr.es/m/CAApHDvpYSO3kc9UryMevWqthTBrxgfd9djiAjKHMPUSQeX9vdQ@mail.gmail.com
Hashing of a single Var is a very common operation for ExprState to
perform. Here we add dedicated ExecJust* functions which helps speed up
Hash Joins by removing the interpretation overhead in ExecInterpExpr().
This change currently only affects Hash Joins on a single column. Hash
Joins with multiple join keys or an expression still run through
ExecInterpExpr().
Some testing has shown up to 10% query performance increases on recent AMD
hardware and nearly 7% increase on an Apple M2 for a query performing a
hash join with a large number of lookups on a small hash table.
This change was extracted from a larger patch which adjusts GROUP BY /
hashed subplans / hashed set operations to use ExprState hashing.
Discussion: https://postgr.es/m/CAApHDvr8Zc0ZgzVoCZLdHGOFNhiJeQ6vrUcS9V7N23zMWQb-eA@mail.gmail.com
Robert Haas expressed concern about whether commit 3eea7a0c9 exposed
the parallel-execution machinery to a case it isn't tested for, namely
a second non-parallel execution of a plan after a parallel execution.
Investigation shows that that can't happen because of pquery.c's
manipulation of the scan direction, but it sure wasn't obvious to
start with. Add some commentary about that.
Discussion: https://postgr.es/m/CA+TgmoagyKQy=HFw+wLo0AKTYWHui+iKswZ8Jnqqd-cFby-WVg@mail.gmail.com
A WITHOUT OVERLAPS primary key or unique constraint is accepted as a
REPLICA IDENTITY, since it guarantees uniqueness. But subscribers
applying logical decoding messages would fail because there was not
support for looking up the equals operator for a gist index. This
fixes that: For GiST indexes we can use the stratnum GiST support
function.
Reviewed-by: Paul Jungwirth <pj@illuminatedcomputing.com>
Reviewed-by: vignesh C <vignesh21@gmail.com>
Discussion: https://www.postgresql.org/message-id/flat/CA+renyUApHgSZF9-nd-a0+OPGharLQLO=mDHcY4_qQ0+noCUVg@mail.gmail.com
get_equal_strategy_number_for_am() gets the equal strategy number for
an AM. This currently only supports btree and hash. In the more
general case, this also depends on the operator class (see for example
GistTranslateStratnum()). To support that, replace this function with
get_equal_strategy_number() that takes an opclass and derives it from
there. (This function already existed before as a static function, so
the signature is kept for simplicity.)
This patch is only a refactoring, it doesn't add support for other
index AMs such as gist. This will be done separately.
Reviewed-by: Paul Jungwirth <pj@illuminatedcomputing.com>
Reviewed-by: vignesh C <vignesh21@gmail.com>
Discussion: https://www.postgresql.org/message-id/flat/CA+renyUApHgSZF9-nd-a0+OPGharLQLO=mDHcY4_qQ0+noCUVg@mail.gmail.com
This "shouldn't happen", except right now it can with a temporal gist
index (to be fixed soon), because of missing gist support in
get_equal_strategy_number(). But right now, the error is not caught
right away, but instead you get the subsequent error about a "missing
operator 0". This makes the error more accurate.
Author: Paul Jungwirth <pj@illuminatedcomputing.com>
Discussion: https://www.postgresql.org/message-id/flat/CA+renyUApHgSZF9-nd-a0+OPGharLQLO=mDHcY4_qQ0+noCUVg@mail.gmail.com
Our parallel-mode code only works when we are executing a query
in full, so ExecutePlan must disable parallel mode when it is
asked to do partial execution. The previous logic for this
involved passing down a flag (variously named execute_once or
run_once) from callers of ExecutorRun or PortalRun. This is
overcomplicated, and unsurprisingly some of the callers didn't
get it right, since it requires keeping state that not all of
them have handy; not to mention that the requirements for it were
undocumented. That led to assertion failures in some corner
cases. The only state we really need for this is the existing
QueryDesc.already_executed flag, so let's just put all the
responsibility in ExecutePlan. (It could have been done in
ExecutorRun too, leading to a slightly shorter patch -- but if
there's ever more than one caller of ExecutePlan, it seems better
to have this logic in the subroutine than the callers.)
This makes those ExecutorRun/PortalRun parameters unnecessary.
In master it seems okay to just remove them, returning the
API for those functions to what it was before parallelism.
Such an API break is clearly not okay in stable branches,
but for them we can just leave the parameters in place after
documenting that they do nothing.
Per report from Yugo Nagata, who also reviewed and tested
this patch. Back-patch to all supported branches.
Discussion: https://postgr.es/m/20241206062549.710dc01cf91224809dd6c0e1@sraoss.co.jp
When resetting a HashJoin node for rescans, if it is a single-batch
join and there are no parameter changes for the inner subnode, we can
just reuse the existing hash table without rebuilding it. However,
for join types that depend on the inner-tuple match flags in the hash
table, we need to reset these match flags to avoid incorrect results.
This applies to right, right-anti, right-semi, and full joins.
When I introduced "Right Semi Join" plan shapes in aa86129e1, I failed
to reset the match flags in the hash table for right-semi joins in
rescans. This oversight has been shown to produce incorrect results.
This patch fixes it.
Author: Richard Guo
Discussion: https://postgr.es/m/CAMbWs4-nQF9io2WL2SkD0eXvfPdyBc9Q=hRwfQHCGV2usa0jyA@mail.gmail.com
The error detail message "Replica identity consists of an unpublished
generated column." implies that the entire replica identity is made up of
an unpublished generated column which may not be the case.
Reported-by: Peter Smith
Author: Shlok Kyal
Reviewed-by: Peter Smith, Amit Kapila
Discussion: https://postgr.es/m/CAHut+PuwMhKx0PhOA4APhJTLoBGNykbeCQpr_CuwGT-SkswG5w@mail.gmail.com
When short-circuiting WindowAgg node evaluation on the top-level
WindowAgg node using quals on monotonic window functions, because the
WindowAgg run condition can mean there's no need to evaluate subsequent
window function results in the same partition once the run condition
becomes false, it was possible that the executor would use stale results
from the previous invocation of the window function in some cases.
A fix for this was partially done by a5832722, but that commit only
fixed the issue for non-top-level WindowAgg nodes. I mistakenly thought
that the top-level WindowAgg didn't have this issue, but Jayesh's example
case clearly shows that's incorrect. At the time, I also thought that
this only affected 32-bit systems as all window functions which then
supported run conditions returned BIGINT, however, that's wrong as
ExecProject is still called and that could cause evaluation of any other
window function belonging to the same WindowAgg node, one of which may
return a byref type.
The only queries affected by this are WindowAggs with a "Run Condition"
which contains at least one window function with a byref result type,
such as lead() or lag() on a byref column. The window clause must also
contain a PARTITION BY clause (without a PARTITION BY, execution of the
WindowAgg stops immediately when the run condition becomes false and
there's no risk of using the stale results).
Reported-by: Jayesh Dehankar
Discussion: https://postgr.es/m/193261e2c4d.3dd3cd7c1842.871636075166132237@zohocorp.com
Backpatch-through: 15, where WindowAgg run conditions were added
Ensure stored generated columns that are part of REPLICA IDENTITY must be
published explicitly for UPDATE and DELETE operations to be published. We
can publish generated columns by listing them in the column list or by
enabling the publish_generated_columns option.
This commit changes the behavior of the test added in commit adedf54e65 by
giving an ERROR for the UPDATE operation in such cases. There is no way to
trigger the bug reported in commit adedf54e65 but we didn't remove the
corresponding code change because it is still relevant when replicating
changes from a publisher with version less than 18.
We decided not to backpatch this behavior change to avoid the risk of
breaking existing output plugins that may be sending generated columns by
default although we are not aware of any such plugin. Also, we didn't see
any reports related to this on STABLE branches which is another reason not
to backpatch this change.
Author: Shlok Kyal, Hou Zhijie
Reviewed-by: Vignesh C, Amit Kapila
Discussion: https://postgr.es/m/CANhcyEVw4V2Awe2AB6i0E5AJLNdASShGfdBLbUd1XtWDboymCA@mail.gmail.com
The new compact_attrs array stores a few select fields from
FormData_pg_attribute in a more compact way, using only 16 bytes per
column instead of the 104 bytes that FormData_pg_attribute uses. Using
CompactAttribute allows performance-critical operations such as tuple
deformation to be performed without looking at the FormData_pg_attribute
element in TupleDesc which means fewer cacheline accesses. With this
change, NAMEDATALEN could be increased with a much smaller negative impact
on performance.
For some workloads, tuple deformation can be the most CPU intensive part
of processing the query. Some testing with 16 columns on a table
where the first column is variable length showed around a 10% increase in
transactions per second for an OLAP type query performing aggregation on
the 16th column. However, in certain cases, the increases were much
higher, up to ~25% on one AMD Zen4 machine.
This also makes pg_attribute.attcacheoff redundant. A follow-on commit
will remove it, thus shrinking the FormData_pg_attribute struct by 4
bytes.
Author: David Rowley
Discussion: https://postgr.es/m/CAApHDvrBztXP3yx=NKNmo3xwFAFhEdyPnvrDg3=M0RhDs+4vYw@mail.gmail.com
Reviewed-by: Andres Freund, Victor Yegorov
If passed a read-write expanded object pointer, the EEOP_NULLIF
code would hand that same pointer to the equality function
and then (unless equality was reported) also return the same
pointer as its value. This is no good, because a function that
receives a read-write expanded object pointer is fully entitled
to scribble on or even delete the object, thus corrupting the
NULLIF output. (This problem is likely unobservable with the
equality functions provided in core Postgres, but it's easy to
demonstrate with one coded in plpgsql.)
To fix, make sure the pointer passed to the equality function
is read-only. We can still return the original read-write
pointer as the NULLIF result, allowing optimization of later
operations.
Per bug #18722 from Alexander Lakhin. This has been wrong
since we invented expanded objects, so back-patch to all
supported branches.
Discussion: https://postgr.es/m/18722-fd9e645448cc78b4@postgresql.org
If a CTE, subquery, sublink, security invoker view, or coercion
projection references a table with row-level security policies, we
neglected to mark the plan as potentially dependent on which role
is executing it. This could lead to later executions in the same
session returning or hiding rows that should have been hidden or
returned instead.
Reported-by: Wolfgang Walther
Reviewed-by: Noah Misch
Security: CVE-2024-10976
Backpatch-through: 12
Two attributes are added to pg_stat_database:
* parallel_workers_to_launch, counting the total number of parallel
workers that were planned to be launched.
* parallel_workers_launched, counting the total number of parallel
workers actually launched.
The ratio of both fields can provide hints that there are not enough
slots available when launching parallel workers, also useful when
pg_stat_statements is not deployed on an instance (i.e. cf54a2c002).
This commit relies on de3a2ea3b2, that has added two fields to EState,
that get incremented when executing Gather or GatherMerge nodes.
A test is added in select_parallel, where parallel workers are spawned.
Bump catalog version.
Author: Benoit Lobréau
Discussion: https://postgr.es/m/783bc7f7-659a-42fa-99dd-ee0565644e25@dalibo.com
Commit ac04aa84a put the shutoff for this into the planner, which is
not ideal because it doesn't prevent us from re-using a previously
made parallel plan. Revert the planner change and instead put the
shutoff into InitializeParallelDSM, modeling it on the existing code
there for recovering from failure to allocate a DSM segment.
However, that code path is mostly untested, and testing a bit harder
showed there's at least one bug: ExecHashJoinReInitializeDSM is not
prepared for us to have skipped doing parallel DSM setup. I also
thought the Assert in ReinitializeParallelWorkers is pretty
ill-advised, and replaced it with a silent Min() operation.
The existing test case added by ac04aa84a serves fine to test this
version of the fix, so no change needed there.
Patch by me, but thanks to Noah Misch for the core idea that we
could shut off worker creation when !INTERRUPTS_CAN_BE_PROCESSED.
Back-patch to v12, as ac04aa84a was.
Discussion: https://postgr.es/m/CAC-SaSzHUKT=vZJ8MPxYdC_URPfax+yoA1hKTcF4ROz_Q6z0_Q@mail.gmail.com
adf97c156 gave ExprStates the ability to hash expressions and return a
single hash value. That commit supports seeding the hash value with an
initial value to have that blended into the final hash value.
Here we fix a hypothetical bug where if there are zero expressions to
hash, the initial value is stored in the wrong location. The existing
code stored the initial value in an intermediate location expecting that
when the expressions were hashed that those steps would store the final
hash value in the ExprState.resvalue field. However, that wouldn't happen
when there are zero expressions to hash. The correct thing to do instead
is to have a special case for zero expressions and when we hit that case,
store the initial value directly in the ExprState.resvalue. The reason
that this is a hypothetical bug is that no code currently calls
ExecBuildHash32Expr passing a non-zero initial value.
Discussion: https://postgr.es/m/CAApHDvpMAL_zxbMRr1LOex3O7Y7R7ZN2i8iUFLQhqQiJMAg3qw@mail.gmail.com
Move all responsibility for indicating a block is exhuasted into
table_scan_bitmap_next_tuple() and advance the main iterator in
heap-specific code. This flow control makes more sense and is a step
toward using the read stream API for bitmap heap scans.
Previously, table_scan_bitmap_next_block() returned false to indicate
table_scan_bitmap_next_tuple() should not be called for the tuples on
the page. This happened both when 1) there were no visible tuples on the
page and 2) when the block returned by the iterator was past the end of
the table. BitmapHeapNext() (generic bitmap table scan code) handled the
case when the bitmap was exhausted.
It makes more sense for table_scan_bitmap_next_tuple() to return false
when there are no visible tuples on the page and
table_scan_bitmap_next_block() to return false when the bitmap is
exhausted or there are no more blocks in the table.
As part of this new design, TBMIterateResults are no longer used as a
flow control mechanism in BitmapHeapNext(), so we removed
table_scan_bitmap_next_tuple's TBMIterateResult parameter.
Note that the prefetch iterator is still saved in the
BitmapHeapScanState node and advanced in generic bitmap table scan code.
This is because 1) it was not necessary to change the prefetch iterator
location to change the flow control in BitmapHeapNext() 2) modifying
prefetch iterator management requires several more steps better split
over multiple commits and 3) the prefetch iterator will be removed once
the read stream API is used.
Author: Melanie Plageman
Reviewed-by: Tomas Vondra, Andres Freund, Heikki Linnakangas, Mark Dilger
Discussion: https://postgr.es/m/063e4eb4-32d9-439e-a0b1-75565a9835a8%40iki.fi
Increment the lossy and exact page counters for EXPLAIN of bitmap heap
scans in heapam_scan_bitmap_next_block(). Note that other table AMs will
need to do this as well
Pushing the counters into heapam_scan_bitmap_next_block() is required to
be able to use the read stream API for bitmap heap scans. The bitmap
iterator must be advanced from inside the read stream callback, so
TBMIterateResults cannot be used as a flow control mechanism in
BitmapHeapNext().
Author: Melanie Plageman
Reviewed-by: Tomas Vondra, Heikki Linnakangas
Discussion: https://postgr.es/m/063e4eb4-32d9-439e-a0b1-75565a9835a8%40iki.fi
The decision in b6e1157e7 to ignore raw_expr when evaluating a
JsonValueExpr was incorrect. While its value is not ultimately
used (since formatted_expr's value is), failing to initialize it
can lead to problems, for instance, when the expression tree in
raw_expr contains Aggref nodes, which must be initialized to
ensure the parent Agg node works correctly.
Also, optimize eval_const_expressions_mutator()'s handling of
JsonValueExpr a bit. Currently, when formatted_expr cannot be folded
into a constant, we end up processing it twice -- once directly in
eval_const_expressions_mutator() and again recursively via
ece_generic_processing(). This recursive processing is required to
handle raw_expr. To avoid the redundant processing of formatted_expr,
we now process raw_expr directly in eval_const_expressions_mutator().
Finally, update the comment of JsonValueExpr to describe the roles of
raw_expr and formatted_expr more clearly.
Bug: #18657
Reported-by: Alexander Lakhin <exclusion@gmail.com>
Diagnosed-by: Fabio R. Sluzala <fabio3rs@gmail.com>
Diagnosed-by: Tender Wang <tndrwang@gmail.com>
Reviewed-by: Tom Lane <tgl@sss.pgh.pa.us>
Discussion: https://postgr.es/m/18657-1b90ccce2b16bdb8@postgresql.org
Backpatch-through: 16
After repartitioning the inner side of a hash join that would have
exceeded the allowed size, we check if all the tuples from a parent
partition moved to one child partition. That is evidence that it
contains duplicate keys and later attempts to repartition will also
fail, so we should give up trying to limit memory (for lack of a better
fallback strategy).
A thinko prevented the check from working correctly in partition 0 (the
one that is partially loaded into memory already). After
repartitioning, we should check for extreme skew if the *parent*
partition's space_exhausted flag was set, not the child partition's.
The consequence was repeated futile repartitioning until per-partition
data exceeded various limits including "ERROR: invalid DSA memory alloc
request size 1811939328", OS allocation failure, or temporary disk space
errors. (We could also do something about some of those symptoms, but
that's material for separate patches.)
This problem only became likely when PostgreSQL 16 introduced support
for Parallel Hash Right/Full Join, allowing NULL keys into the hash
table. Repartitioning always leaves NULL in partition 0, no matter how
many times you do it, because the hash value is all zero bits. That's
unlikely for other hashed values, but they might still have caused
wasted extra effort before giving up.
Back-patch to all supported releases.
Reported-by: Craig Milhiser <craig@milhiser.com>
Reviewed-by: Andrei Lepikhov <lepihov@gmail.com>
Discussion: https://postgr.es/m/CA%2BwnhO1OfgXbmXgC4fv_uu%3DOxcDQuHvfoQ4k0DFeB0Qqd-X-rQ%40mail.gmail.com
adf97c156 made it so ExprStates could support hashing and changed Hash
Join to use that instead of manually extracting Datums from tuples and
hashing them one column at a time.
When hashing multiple columns or expressions, the code added in that
commit stored the intermediate hash value in the ExprState's resvalue
field. That was a mistake as steps may be injected into the ExprState
between each hashing step that look at or overwrite the stored
intermediate hash value. EEOP_PARAM_SET is an example of such a step.
Here we fix this by adding a new dedicated field for storing
intermediate hash values and adjust the code so that all apart from the
final hashing step store their result in the intermediate field.
In passing, rename a variable so that it's more aligned to the
surrounding code and also so a few lines stay within the 80 char margin.
Reported-by: Andres Freund
Reviewed-by: Alena Rybakina <a.rybakina@postgrespro.ru>
Discussion: https://postgr.es/m/CAApHDvqo9eenEFXND5zZ9JxO_k4eTA4jKMGxSyjdTrsmYvnmZw@mail.gmail.com
Commit 2dc1deaea turns out to have been still a brick shy of a load,
because CALL statements executing within a plpgsql exception block
could still pass the wrong snapshot to stable functions within the
CALL's argument list. That happened because standard_ProcessUtility
forces isAtomicContext to true if IsTransactionBlock is true, which
it always will be inside a subtransaction. Then ExecuteCallStmt
would think it does not need to push a new snapshot --- but
_SPI_execute_plan didn't do so either, since it thought it was in
nonatomic mode.
The best fix for this seems to be for _SPI_execute_plan to operate
in atomic execution mode if IsSubTransaction() is true, even when the
SPI context as a whole is non-atomic. This makes _SPI_execute_plan
have the same rules about when non-atomic execution is allowed as
_SPI_commit/_SPI_rollback have about when COMMIT/ROLLBACK are allowed,
which seems appropriately symmetric. (If anyone ever tries to allow
COMMIT/ROLLBACK inside a subtransaction, this would all need to be
rethought ... but I'm unconvinced that such a thing could be logically
consistent at all.)
For further consistency, also check IsSubTransaction() in
SPI_inside_nonatomic_context. That does not matter for its
one present-day caller StartTransaction, which can't be reached
inside a subtransaction. But if any other callers ever arise,
they'd presumably want this definition.
Per bug #18656 from Alexander Alehin. Back-patch to all
supported branches, like previous fixes in this area.
Discussion: https://postgr.es/m/18656-cade1780866ef66c@postgresql.org
The MergeJoin struct was tracking "mergeStrategies", which were an
array of btree strategy numbers, purely for the purpose of comparing
it later against btree strategies to determine if the scan direction
was forward or reverse. Change that. Instead, track
"mergeReversals", an array of bool, to indicate the same without an
unfortunate assumption that a strategy number refers specifically to a
btree strategy.
Author: Mark Dilger <mark.dilger@enterprisedb.com>
Discussion: https://www.postgresql.org/message-id/flat/E72EAA49-354D-4C2E-8EB9-255197F55330@enterprisedb.com
These fields can be set by executor nodes to record how many parallel
workers were planned to be launched and how many of them have been
actually launched within the number initially planned. This data is
able to give an approximation of the parallel worker draught a system
is facing, making easier the tuning of related configuration parameters.
These fields will be used by some follow-up patches to populate other
parts of the system with their data.
Author: Guillaume Lelarge, Benoit Lobréau
Discussion: https://postgr.es/m/783bc7f7-659a-42fa-99dd-ee0565644e25@dalibo.com
Discussion: https://postgr.es/m/CAECtzeWtTGOK0UgKXdDGpfTVSa5bd_VbUt6K6xn8P7X+_dZqKw@mail.gmail.com
This assertion has been added by 24f5205948, but Alexander Lakhin has
proved that the ExecutorRun() one can be broken by using a PL function
that manipulates compute_query_id and track_activities, while the ones
in ExecutorFinish() and ExecutorEnd() could be triggered when cleaning
up portals at the beginning of a new query execution.
Discussion: https://postgr.es/m/b37d8e6c-e83d-e157-8865-1b2460a6aef2@gmail.com
As formulated, the assertion added in the executor by 24f5205948 to
check that a query ID is set had two problems:
- track_activities may be disabled while compute_query_id is enabled,
causing the query ID to not be reported to pg_stat_activity.
- debug_query_string may not be set in some context. The only path
where this would matter is visibly autovacuum, should parallel workers
be enabled there at some point. This is not the case currently.
There was no test showing the interactions between the query ID and
track_activities, so let's add one based on a scan of pg_stat_activity.
This assertion is still an experimentation at this stage, but let's see
if this shows more paths where query IDs are not properly set while they
should.
Discussion: https://postgr.es/m/Zvn5616oYXmpXyHI@paquier.xyz
The previous commit fixed some ways of losing an inplace update. It
remained possible to lose one when a backend working toward a
heap_update() copied a tuple into memory just before inplace update of
that tuple. In catalogs eligible for inplace update, use LOCKTAG_TUPLE
to govern admission to the steps of copying an old tuple, modifying it,
and issuing heap_update(). This includes MERGE commands. To avoid
changing most of the pg_class DDL, don't require LOCKTAG_TUPLE when
holding a relation lock sufficient to exclude inplace updaters.
Back-patch to v12 (all supported versions). In v13 and v12, "UPDATE
pg_class" or "UPDATE pg_database" can still lose an inplace update. The
v14+ UPDATE fix needs commit 86dc90056d,
and it wasn't worth reimplementing that fix without such infrastructure.
Reviewed by Nitin Motiani and (in earlier versions) Heikki Linnakangas.
Discussion: https://postgr.es/m/20231027214946.79.nmisch@google.com
nodeRecursiveunion.c makes use of two tuplestores and, until now, would
delete and recreate one of these tuplestores after every recursive
iteration.
Here we adjust that behavior and instead reuse one of the existing
tuplestores and just empty it of all tuples using tuplestore_clear().
This saves some free/malloc roundtrips and has shown a 25-30% performance
improvement for queries that perform very little work between recursive
iterations.
This also paves the way to add some EXPLAIN ANALYZE telemetry output for
recursive common table expressions, similar to what was done in 1eff8279d
and 95d6e9af0. Previously calling tuplestore_end() would have caused
the maximum storage space used to be lost.
Reviewed-by: Tatsuo Ishii
Discussion: https://postgr.es/m/CAApHDvr9yW0YRiK8A2J7nvyT8g17YzbSfOviEWrghazKZbHbig@mail.gmail.com
This commit adds three sanity checks in code paths of the executor where
it is possible to use hooks, checking that a query ID is reported in
pg_stat_activity if compute_query_id is enabled:
- ExecutorRun()
- ExecutorFinish()
- ExecutorEnd()
This causes the test in pg_stat_statements added in 933848d16d to
complain immediately in ExecutorRun(). The idea behind this commit is
to help extensions to detect if they are missing query ID reports when a
query goes through the executor. Perhaps this will prove to be a bad
idea, but let's see where this experience goes in v18 and newer
versions.
Reviewed-by: Sami Imseih
Discussion: https://postgr.es/m/ZuJb5xCKHH0A9tMN@paquier.xyz
This commit adds query ID reports for two code paths when processing
extended query protocol messages:
- When receiving a bind message, setting it to the first Query retrieved
from a cached cache.
- When receiving an execute message, setting it to the first PlannedStmt
stored in a portal.
An advantage of this method is that this is able to cover all the types
of portals handled in the extended query protocol, particularly these
two when the report done in ExecutorStart() is not enough (neither is an
addition in ExecutorRun(), actually, for the second point):
- Multiple execute messages, with multiple ExecutorRun().
- Portal with execute/fetch messages, like a query with a RETURNING
clause and a fetch size that stores the tuples in a first execute
message going though ExecutorStart() and ExecuteRun(), followed by one
or more execute messages doing only fetches from the tuplestore created
in the first message. This corresponds to the case where
execute_is_fetch is set, for example.
Note that the query ID reporting done in ExecutorStart() is still
necessary, as an EXECUTE requires it. Query ID reporting is optimistic
and more calls to pgstat_report_query_id() don't matter as the first
report takes priority except if the report is forced. The comment in
ExecutorStart() is adjusted to reflect better the reality with the
extended query protocol.
The test added in pg_stat_statements is a courtesy of Robert Haas. This
uses psql's \bind metacommand, hence this part is backpatched down to
v16.
Reported-by: Kaido Vaikla, Erik Wienhold
Author: Sami Imseih
Reviewed-by: Jian He, Andrei Lepikhov, Michael Paquier
Discussion: https://postgr.es/m/CA+427g8DiW3aZ6pOpVgkPbqK97ouBdf18VLiHFesea2jUk3XoQ@mail.gmail.com
Discussion: https://postgr.es/m/CA+TgmoZxtnf_jZ=VqBSyaU8hfUkkwoJCJ6ufy4LGpXaunKrjrg@mail.gmail.com
Discussion: https://postgr.es/m/1391613709.939460.1684777418070@office.mailbox.org
Backpatch-through: 14
Add WITHOUT OVERLAPS clause to PRIMARY KEY and UNIQUE constraints.
These are backed by GiST indexes instead of B-tree indexes, since they
are essentially exclusion constraints with = for the scalar parts of
the key and && for the temporal part.
(previously committed as 46a0cd4cef, reverted by 46a0cd4cefb; the new
part is this:)
Because 'empty' && 'empty' is false, the temporal PK/UQ constraint
allowed duplicates, which is confusing to users and breaks internal
expectations. For instance, when GROUP BY checks functional
dependencies on the PK, it allows selecting other columns from the
table, but in the presence of duplicate keys you could get the value
from any of their rows. So we need to forbid empties.
This all means that at the moment we can only support ranges and
multiranges for temporal PK/UQs, unlike the original patch (above).
Documentation and tests for this are added. But this could
conceivably be extended by introducing some more general support for
the notion of "empty" for other types.
Author: Paul A. Jungwirth <pj@illuminatedcomputing.com>
Reviewed-by: Peter Eisentraut <peter@eisentraut.org>
Reviewed-by: jian he <jian.universality@gmail.com>
Discussion: https://www.postgresql.org/message-id/flat/CA+renyUApHgSZF9-nd-a0+OPGharLQLO=mDHcY4_qQ0+noCUVg@mail.gmail.com
