Robert Haas [Fri, 30 Aug 2024 19:57:47 +0000 (15:57 -0400)] Also subdivide merge join.
Robert Haas [Fri, 30 Aug 2024 19:11:23 +0000 (15:11 -0400)] Split up nested loops.
Robert Haas [Fri, 30 Aug 2024 14:27:31 +0000 (10:27 -0400)] New contrib module: alphabet_join.
This forces joins to be done alphabetically by alias name. It demonstrates
that join_path_setup_hook is sufficient to control the join order, and
is not intended for commit.
Robert Haas [Thu, 29 Aug 2024 19:38:58 +0000 (15:38 -0400)] Allow extensions to control join strategy.
At the start of join planning, we build a bitmask called jsa_mask based on
the values of the various enable_* planner GUCs, indicating which join
strategies are allowed. Extensions can override this mask on a plan-wide
basis using join_search_hook, or, generaly more usefully, they can change
the mask for each call to add_paths_to_joinrel using a new hook called
join_path_setup_hook. This is sufficient to allow an extension to force
the use of particular join strategies either in general or for specific
joins, and it is also sufficient to allow an extension to force the join
order.
If you want to control some aspect of planner behavior that is not
join-related, this won't help, but the same concepts could be
applied to scans, appendrels, and aggregation.
Notes:
- Materialization and memoization are optional sub-strategies when
performing a nested loop. You might want to avoid these strategies, allow
them, or force them. I don't know how this should be controlled.
- join_path_setup_hook will see JOIN_UNIQUE_INNER or JOIN_UNIQUE_OUTER as the
jointype and can reject if it wants. Conversely, you could allow that
case and reject the JOIN_SEMI/JOIN_RIGHT_SEMI case. However, if you wanted
to force the uniquification to use a sort or a hash rather than the
other, more would be needed (though it looks like you could kludge things
by modifying the SpecialJoinInfo).
- To fully control merge joins produced by sort_outer_and_inner(),
you would need control over which merge clauses are selected.
In match_unsorted_outer(), it's just a function of the input paths.
- You could potentially want to disallow certain strategies at greater
granularity than is possible here - e.g. unparameterized paths aren't
disabled, but making a nested loop out of them is.
- There are going to be residual references to various planner GUCs in
the code path, such as enable_parallel_hash and enable_sort. That seems
fine.
Peter Eisentraut [Wed, 28 Aug 2024 05:26:48 +0000 (07:26 +0200)] Add prefetching support on macOS
macOS doesn't have posix_fadvise(), but fcntl() with the F_RDADVISE
command does the same thing.
Some related documentation has been generalized to not mention
posix_advise() specifically anymore.
Reviewed-by: Thomas Munro <[email protected]>Discussion: https://www.postgresql.org/message-id/flat/
0827edec-1317-4917-a186-
035eb1e3241d%40eisentraut.org
Peter Eisentraut [Tue, 27 Aug 2024 14:54:10 +0000 (16:54 +0200)] Message style improvements
Peter Eisentraut [Tue, 27 Aug 2024 14:15:28 +0000 (16:15 +0200)] Fix misplaced translator comments
They did not immediately precede the code they were applying to.
Masahiko Sawada [Mon, 26 Aug 2024 23:16:12 +0000 (16:16 -0700)] Fix identation.
Masahiko Sawada [Mon, 26 Aug 2024 18:00:07 +0000 (11:00 -0700)] Fix memory counter update in ReorderBuffer.
Commit
5bec1d6bc5e changed the memory usage updates of the
ReorderBufferTXN to zero all at once by subtracting txn->size, rather
than updating it for each change. However, if TOAST reconstruction
data remained in the transaction when freeing it, there were cases
where it further subtracted the memory counter from zero, resulting in
an assertion failure.
This change calculates the memory size for each change and updates the
memory usage to precisely the amount that has been freed.
Back to v17, where this was introducd.
Reviewed-by: Amit Kapila, Shlok KyalDiscussion: https://postgr.es/m/CAD21AoAqkNUvicgKPT_dXzNoOwpPkVTg0QPPxEcWmzT0moCJ1g%40mail.gmail.com
Back-through: 17
Peter Geoghegan [Mon, 26 Aug 2024 15:29:15 +0000 (11:29 -0400)] Fix nbtree lookahead overflow bug.
Add bounds checking to nbtree's lookahead/skip-within-a-page mechanism.
Otherwise it's possible for cases with lots of before-array-keys tuples
to overflow an int16 variable, causing the mechanism to generate an out
of bounds page offset number.
Oversight in commit
5bf748b8, which enhanced nbtree ScalarArrayOp
execution.
Reported-By: Alexander Lakhin <[email protected]>Discussion: https://postgr.es/m/
6c68ac42-bbb5-8b24-103e-
af0e279c536f@gmail.com
Back: 17-, where nbtree SAOP execution was enhanced.
Peter Eisentraut [Mon, 26 Aug 2024 12:38:59 +0000 (14:38 +0200)] pg_upgrade: Message style improvements
Dean Rasheed [Mon, 26 Aug 2024 10:00:20 +0000 (11:00 +0100)] Fix compiler warning in mul_var_short().
Some compilers (e.g., gcc before version 7) mistakenly think "carry"
might be used uninitialized.
Reported by Tom Lane, per various buildfarm members, e.g. arowana.
Alexander Korotkov [Sun, 25 Aug 2024 21:22:44 +0000 (00:22 +0300)] Revert: Avoid looping over all type cache entries in TypeCacheRelCallback()
This commit reverts
c14d4acb8 as the design didn't take into account
that TypeCacheEntry could be invalidated during the lookup_type_cache() call.
Reported-by: Alexander LakhinDiscussion: https://postgr.es/m/
1927cba4-177e-5c23-cbcc-
d444a850304f%40gmail.com
Alexander Korotkov [Sun, 25 Aug 2024 00:21:23 +0000 (03:21 +0300)] Avoid looping over all type cache entries in TypeCacheRelCallback()
Currently when a single relcache entry gets invalidated,
TypeCacheRelCallback() has to loop over all type cache entries to find
appropriate typentry to invalidate. Unfortunately, using the syscache here
is impossible, because this callback could be called outside a transaction
and this makes impossible catalog lookups. This is why present commit
introduces RelIdToTypeIdCacheHash to map relation OID to its composite type
OID.
We are keeping RelIdToTypeIdCacheHash entry while corresponding type cache
entry have something to clean. Therefore, RelIdToTypeIdCacheHash shouldn't
get bloat in the case of temporary tables flood.
Discussion: https://postgr.es/m/
5812a6e5-68ae-4d84-9d85-
b443176966a1%40sigaev.ru
Author: Teodor Sigaev
Reviewed-by: Aleksander Alekseev, Tom Lane, Michael Paquier, Roman ZharkovReviewed-by: Andrei Lepikhov, Pavel BorisovAlexander Korotkov [Sat, 24 Aug 2024 15:48:48 +0000 (18:48 +0300)] Revert support for ALTER TABLE ... MERGE/SPLIT PARTITION(S) commands
This commit reverts
1adf16b8fb,
87c21bb941, and subsequent fixes and
improvements including
df64c81ca9,
c99ef1811a,
9dfcac8e15,
885742b9f8,
842c9b2705,
fcf80c5d5f,
96c7381c4c,
f4fc7cb54b,
60ae37a8bc,
259c96fa8f,
449cdcd486,
3ca43dbbb6,
2a679ae94e,
3a82c689fd,
fbd4321fd5,
d53a4286d7,
c086896625,
4e5d6c4091,
04158e7fa3.
The reason for reverting is security issues related to repeatable name lookups
(CVE-2014-0062). Even though
04158e7fa3 solved part of the problem, there
are still remaining issues, which aren't feasible to even carefully analyze
before the RC deadline.
Reported-by: Noah Misch, Robert HaasDiscussion: https://postgr.es/m/
20240808171351.a9.nmisch%40google.com
Back-through: 17
Peter Eisentraut [Sat, 24 Aug 2024 13:56:32 +0000 (15:56 +0200)] pg_createsubscriber: Message style improvements
Tom Lane [Fri, 23 Aug 2024 14:12:56 +0000 (10:12 -0400)] Provide feature-test macros for libpq features added in v17.
As per the policy established in commit
6991e774e, invent macros
that can be tested at compile time to detect presence of new libpq
features. This should make calling code more readable and less
error-prone than checking the libpq version would be (especially
since we don't expose that at compile time; the server version is
an unreliable substitute).
Discussion: https://postgr.es/m/
2042418.
1724346970@sss.pgh.pa.us
Peter Eisentraut [Fri, 23 Aug 2024 05:07:53 +0000 (07:07 +0200)] thread-safety: gmtime_r(), localtime_r()
Use gmtime_r() and localtime_r() instead of gmtime() and localtime(),
for thread-safety.
There are a few affected calls in libpq and ecpg's libpgtypes, which
are probably effectively bugs, because those libraries already claim
to be thread-safe.
There is one affected call in the backend. Most of the backend
otherwise uses the custom functions pg_gmtime() and pg_localtime(),
which are implemented differently.
While we're here, change the call in the backend to gmtime*() instead
of localtime*(), since for that use time zone behavior is irrelevant,
and this side-steps any questions about when time zones are
initialized by localtime_r() vs localtime().
Portability: gmtime_r() and localtime_r() are in POSIX but are not
available on Windows. Windows has functions gmtime_s() and
localtime_s() that can fulfill the same purpose, so we add some small
wrappers around them. (Note that these *_s() functions are also
different from the *_s() functions in the bounds-checking extension of
C11. We are not using those here.)
On MinGW, you can get the POSIX-style *_r() functions by defining
_POSIX_C_SOURCE appropriately before including <time.h>. This leads
to a conflict at least in plpython because apparently _POSIX_C_SOURCE
gets defined in some header there, and then our replacement
definitions conflict with the system definitions. To avoid that sort
of thing, we now always define _POSIX_C_SOURCE on MinGW and use the
POSIX-style functions here.
Reviewed-by: Stepan Neretin <[email protected]>Reviewed-by: Heikki Linnakangas <[email protected]>Reviewed-by: Thomas Munro <[email protected]>Discussion: https://www.postgresql.org/message-id/flat/
eba1dc75-298e-4c46-8869-
48ba8aad7d70@eisentraut.org
Michael Paquier [Fri, 23 Aug 2024 03:11:36 +0000 (12:11 +0900)] Rework new SLRU test with injection points
Rather than the SQL injection_points_load(), this commit changes the
injection point test introduced in
768a9fd5535f to rely on the two
macros INJECTION_POINT_LOAD() and INJECTION_POINT_CACHED(), that have
been originally introduced for the sake of this test.
This runs the test as a two-step process: load the injection point, then
run its callback directly from the local cache loaded. What the test
did originally was also fine, but the point here is to have an example
in core of how to use these new macros.
While on it, fix the header ordering in multixact.c, as pointed out by
Alexander Korotkov. This was an oversight in
768a9fd5535f.
Per discussion with Álvaro Herrera.
Author: Michael Paquier
Discussion: https://postgr.es/m/
[email protected]Discussion: https://postgr.es/m/CAPpHfduzaBz7KMhwuVOZMTpG=JniPG4aUosXPZCxZydmzq_oEQ@mail.gmail.com
Michael Paquier [Fri, 23 Aug 2024 02:36:41 +0000 (11:36 +0900)] injection_point: Add injection_points.stats
This GUC controls if cumulative statistics are enabled or not in the
module. Custom statistics require the module to be loaded with
shared_preload_libraries, hence this GUC is made PGC_POSTMASTER. By
default, the stats are disabled. 001_stats.pl is updated to enable the
statistics, as it is the only area where these are required now.
This will be used by an upcoming change for the injection point test
added by
768a9fd5535f where stats should not be used, as the test runs a
point callback in a critical section. And the module injection_points
will need to be loaded with shared_preload_libraries there.
Per discussion with Álvaro Herrera.
Author: Michael Paquier
Discussion: https://postgr.es/m/
[email protected]Michael Paquier [Fri, 23 Aug 2024 01:12:58 +0000 (10:12 +0900)] injection_points: Add initialization of shmem state when loading module
This commits adds callbacks to initialize the shared memory state of the
module when loaded with shared_preload_libraries. This is necessary to
be able to update the test introduced in
768a9fd5535f to use the macros
INJECTION_POINT_{LOAD,CACHED}() rather than a SQL function in the module
injection_points forcing a load, as this test runs a callback in a
critical section where no memory allocation should happen.
Initializing the shared memory state of the module while loading
provides a strict control on the timing of its allocation. If the
module is not loaded at startup, it will use a GetNamedDSMSegment()
instead to initialize its shmem state on-the-fly.
Per discussion with Álvaro Herrera.
Author: Michael Paquier
Discussion: https://postgr.es/m/
[email protected]Amit Kapila [Thu, 22 Aug 2024 08:41:50 +0000 (14:11 +0530)] Doc: explain the log format of logical replication conflicts.
This commit adds a detailed explanation of the log format for logical
replication conflicts.
Author: Hou Zhijie
Reviewed-by: Shveta Malik, Peter Smith, Hayato Kuroda
Discussion: https://postgr.es/m/OS0PR01MB5716352552DFADB8E9AD1D8994C92@OS0PR01MB5716.jpnprd01.prod.outlook.com
Discussion: https://postgr.es/m/OS0PR01MB57162EDE8BA17F3EE08A24CA948D2@OS0PR01MB5716.jpnprd01.prod.outlook.com
Michael Paquier [Thu, 22 Aug 2024 07:25:57 +0000 (16:25 +0900)] psql: Add more meta-commands able to use the extended protocol
Currently, only unnamed prepared statement are supported by psql with
the meta-command \bind. With only this command, it is not possible to
test named statement creation, execution or close through the extended
protocol.
This commit introduces three additional commands:
* \parse creates a prepared statement using the extended protocol,
acting as a wrapper of libpq's PQsendPrepare().
* \bind_named binds and executes an existing prepared statement using
the extended protocol, for PQsendQueryPrepared().
* \close closes an existing prepared statement using the extended
protocol, for PQsendClosePrepared().
This is going to be useful to add regression tests for the extended
query protocol, and I have some plans for that on separate threads.
Note that \bind relies on PQsendQueryParams().
The code of psql is refactored so as bind_flag is replaced by an enum in
_psqlSettings that tracks the type of libpq routine to execute, based on
the meta-command involved, with the default being PQsendQuery(). This
refactoring piece has been written by me, while Anthonin has implemented
the rest.
Author: Anthonin Bonnefoy, Michael Paquier
Reviewed-by: Aleksander Alekseev, Jelte Fennema-Nio
Discussion: https://postgr.es/m/CAO6_XqpSq0Q0kQcVLCbtagY94V2GxNP3zCnR6WnOM8WqXPK4nw@mail.gmail.com
Noah Misch [Thu, 22 Aug 2024 07:07:04 +0000 (00:07 -0700)] Fix attach of a previously-detached injection point.
It's normal for the name in a free slot to match the new name. The
max_inuse mechanism kept simple cases from reaching the problem. The
problem could appear when index 0 was the previously-detached entry and
index 1 is in use. Back- to v17, where this code first appeared.
Alexander Korotkov [Thu, 22 Aug 2024 06:50:48 +0000 (09:50 +0300)] Avoid repeated table name lookups in createPartitionTable()
Currently, createPartitionTable() opens newly created table using its name.
This approach is prone to privilege escalation attack, because we might end
up opening another table than we just created.
This commit address the issue above by opening newly created table by its
OID. It appears to be tricky to get a relation OID out of ProcessUtility().
We have to extend TableLikeClause with new newRelationOid field, which is
filled within ProcessUtility() to be further accessed by caller.
Security: CVE-2014-0062
Reported-by: Noah MischDiscussion: https://postgr.es/m/
20240808171351.a9.nmisch%40google.com
Reviewed-by: Pavel Borisov, Dmitry KovalRichard Guo [Thu, 22 Aug 2024 02:41:08 +0000 (11:41 +0900)] Small code simplification
Apply the same code simplification to ATExecAddColumn as was done in
7ff9afbbd: apply GETSTRUCT() once instead of doing it repeatedly in
the same function.
Author: Tender Wang
Discussion: https://postgr.es/m/CAHewXNkO9+U437jvKT14s0MCu6Qpf6G-p2mZK5J9mAi4cHDgpQ@mail.gmail.com
Michael Paquier [Thu, 22 Aug 2024 01:48:25 +0000 (10:48 +0900)] Create syscache entries for pg_extension
Two syscache identifiers are added for extension names and OIDs.
Shared libraries of extensions might want to invalidate or update their
own caches whenever a CREATE, ALTER or DROP EXTENSION command is run for
their extension (in any backend). Right now this is non-trivial to do
correctly and efficiently, but, if an extension catalog is part of a
syscache, this could simply be done by registering an callback using
CacheRegisterSyscacheCallback for the relevant syscache.
Another case where this is useful is a loaded library where some of its
code paths rely on some objects of the extension to exist; it can be
simpler and more efficient to do an existence check directly on the
extension through the syscache.
Author: Jelte Fennema-Nio
Reviewed-by: Alexander Korotkov, Pavel Stehule
Discussion: https://postgr.es/m/CAGECzQTWm9sex719Hptbq4j56hBGUti7J9OWjeMobQ1ccRok9w@mail.gmail.com
Jeff Davis [Wed, 21 Aug 2024 16:19:21 +0000 (09:19 -0700)] Fix obsolete comments in varstr_cmp().
Tom Lane [Wed, 21 Aug 2024 16:00:03 +0000 (12:00 -0400)] Disallow creating binary-coercible casts involving range types.
For a long time we have forbidden binary-coercible casts to or from
composite and array types, because such a cast cannot work correctly:
the type OID embedded in the value would need to change, but it won't
in a binary coercion. That reasoning applies equally to range types,
but we overlooked installing a similar restriction here when we
invented range types. Do so now.
Given the lack of field complaints, we won't change this in stable
branches, but it seems not too late for v17.
Per discussion of a problem noted by Peter Eisentraut.
Discussion: https://postgr.es/m/
076968e1-0852-40a9-bc0b-
117cd3f0e43c@eisentraut.org
Robert Haas [Wed, 21 Aug 2024 14:14:35 +0000 (10:14 -0400)] Show number of disabled nodes in EXPLAIN ANALYZE output.
Now that disable_cost is not included in the cost estimate, there's
no visible sign in EXPLAIN output of which plan nodes are disabled.
Fix that by propagating the number of disabled nodes from Path to
Plan, and then showing it in the EXPLAIN output.
There is some question about whether this is a desirable change.
While I personally believe that it is, it seems best to make it a
separate commit, in case we decide to back out just this part, or
rework it.
Reviewed by Andres Freund, Heikki Linnakangas, and David Rowley.
Discussion: http://postgr.es/m/CA+TgmoZ_+MS+o6NeGK2xyBv-xM+w1AfFVuHE4f_aq6ekHv7YSQ@mail.gmail.com
Robert Haas [Wed, 21 Aug 2024 14:12:30 +0000 (10:12 -0400)] Treat number of disabled nodes in a path as a separate cost metric.
Previously, when a path type was disabled by e.g. enable_seqscan=false,
we either avoided generating that path type in the first place, or
more commonly, we added a large constant, called disable_cost, to the
estimated startup cost of that path. This latter approach can distort
planning. For instance, an extremely expensive non-disabled path
could seem to be worse than a disabled path, especially if the full
cost of that path node need not be paid (e.g. due to a Limit).
Or, as in the regression test whose expected output changes with this
commit, the addition of disable_cost can make two paths that would
normally be distinguishible in cost seem to have fuzzily the same cost.
To fix that, we now count the number of disabled path nodes and
consider that a high-order component of both the startup cost and the
total cost. Hence, the path list is now sorted by disabled_nodes and
then by total_cost, instead of just by the latter, and likewise for
the partial path list. It is important that this number is a count
and not simply a Boolean; else, as soon as we're unable to respect
disabled path types in all portions of the path, we stop trying to
avoid them where we can.
Because the path list is now sorted by the number of disabled nodes,
the join prechecks must compute the count of disabled nodes during
the initial cost phase instead of postponing it to final cost time.
Counts of disabled nodes do not cross subquery levels; at present,
there is no reason for them to do so, since the we do not postpone
path selection across subquery boundaries (see make_subplan).
Reviewed by Andres Freund, Heikki Linnakangas, and David Rowley.
Discussion: http://postgr.es/m/CA+TgmoZ_+MS+o6NeGK2xyBv-xM+w1AfFVuHE4f_aq6ekHv7YSQ@mail.gmail.com
Robert Haas [Wed, 21 Aug 2024 13:58:11 +0000 (09:58 -0400)] Peter Eisentraut [Wed, 21 Aug 2024 13:11:21 +0000 (15:11 +0200)] doc: remove llvm-config search from configure documentation
As of
4dd29b6833, we no longer attempt to locate any other llvm-config
variant than plain llvm-config in configure-based builds; update the
documentation accordingly. (For Meson-based builds, we still use Meson's
LLVMDependencyConfigTool [0], which runs through a set of possible
suffixes [1], so no need to update the documentation there.)
[0]: https://.com/mesonbuild/meson/blob/
7d28ff29396f9d7043204de8ddc52226b9903811/mesonbuild/dependencies/dev.py#L184
[1]: https://.com/mesonbuild/meson/blob/
7d28ff29396f9d7043204de8ddc52226b9903811/mesonbuild/environment.py#L183
Author: Ole Peder Brandtzæg <
[email protected]>
Discussion: https://www.postgresql.org/message-id/
20240518224601.gtisttjerylukjr5%40samfundet.no
Amit Kapila [Wed, 21 Aug 2024 11:15:36 +0000 (16:45 +0530)] Fix typos in
9758174e2e.
Reported off-list by Erik Rijkers
Peter Eisentraut [Wed, 21 Aug 2024 07:21:25 +0000 (09:21 +0200)] Small code simplification
Apply GETSTRUCT() once instead of doing it repeatedly in the same
function. This simplifies the notation and makes the function's
structure more similar to the surrounding ones.
Discussion: https://www.postgresql.org/message-id/flat/
a368248e-69e4-40be-9c07-
6c3b5880b0a6@eisentraut.org
Amit Kapila [Wed, 21 Aug 2024 03:52:32 +0000 (09:22 +0530)] Don't advance origin during apply failure.
We advance origin progress during abort on successful and
application of ROLLBACK in parallel mode. But the origin
shouldn't be advanced during an error or unsuccessful apply due to
shutdown. Otherwise, it will result in a transaction loss as such a
transaction won't be sent again by the server.
Reported-by: Hou Zhijie
Author: Hayato Kuroda and Shveta Malik
Reviewed-by: Amit Kapila
Back-through: 16
Discussion: https://postgr.es/m/TYAPR01MB5692FAC23BE40C69DA8ED4AFF5B92@TYAPR01MB5692.jpnprd01.prod.outlook.com
Jeff Davis [Tue, 20 Aug 2024 21:29:34 +0000 (14:29 -0700)] Slightly refactor varstr_sortsupport() to improve readability.
Author: Andreas Karlsson
Discussion: https://postgr.es/m/
69c2a864-846f-4309-bd5a-
aaa1c34f9a11@proxel.se
Michael Paquier [Tue, 20 Aug 2024 22:24:03 +0000 (07:24 +0900)] Remove _PG_fini()
ab02d702ef08 has removed from the backend the code able to support the
unloading of modules, because this has never worked. This removes the
last references to _PG_fini(), that could be used as a callback for
modules to manipulate the stack when unloading a library.
The test module ldap_password_func had the idea to declare it, doing
nothing. The function declaration in fmgr.h is gone.
It was left around in 2022 to avoid breaking extension code, but at this
stage there are also benefits in letting extension developers know that
keeping the unloading code is pointless and this move leads to less
maintenance.
Reviewed-by: Tom Lane, Heikki LinnakangasDiscussion: https://postgr.es/m/
[email protected]Alvaro Herrera [Tue, 20 Aug 2024 21:53:40 +0000 (17:53 -0400)] Minor wording change in table "JSON Creation Functions"
For readability. Back to 16.
Author: Erik Wienhold <
[email protected]>
Discussion: https://postgr.es/m/
8ddac732-d650-4958-b9c9-
ea8e6116251e@ewie.name
Jeff Davis [Tue, 20 Aug 2024 18:24:00 +0000 (11:24 -0700)] Improve configure error for ICU libraries if pkg-config is absent.
If pkg-config is not installed, the ICU libraries cannot be found, but
the custom configure error message did not mention this. This might
lead to confusion about the actual problem. To improve this, remove
the explicit error message and rely on PKG_CHECK_MODULES' generic
error message.
Author: Michael Banck
Reported-by: Holger JakobsDiscussion: https://postgr.es/m/
ccd579ed-4949-d3de-ab13-
9e6456fd2caf%40jakobs.com
Discussion: https://postgr.es/m/
66b5d05c.
050a0220[email protected]Nathan Bossart [Tue, 20 Aug 2024 18:43:20 +0000 (13:43 -0500)] Fix a couple of wait event descriptions.
The descriptions for ProcArrayGroupUpdate and XactGroupUpdate claim
that these events mean we are waiting for the group leader "at end
of a parallel operation," but neither pertains to parallel
operations. This commit reverts these descriptions to their
wording before commit
3048898e73, i.e., "end of a parallel
operation" is changed to "transaction end."
Author: Sameer Kumar
Reviewed-by: Amit KapilaDiscussion: https://postgr.es/m/CAGPeHmh6UMrKQHKCmX%2B5vV5TH9P%3DKw9en3k68qEem6J%3DyrZPUA%40mail.gmail.com
Back-through: 13
Alvaro Herrera [Tue, 20 Aug 2024 18:20:48 +0000 (14:20 -0400)] Add injection-point test for new multixact CV usage
Before commit
a0e0fb1ba56f, multixact.c contained a case in the
multixact-read path where it would loop sleeping 1ms each time until
another multixact-create path completed, which was uncovered by any
tests. That commit changed the code to rely on a condition variable
instead. Add a test now, which relies on injection points and "loading"
thereof (because of it being in a critical section), per commit
4b211003ecc2.
Author: Andrey Borodin <
[email protected]>
Reviewed-by: Michaël Paquier <[email protected]>Discussion: https://postgr.es/m/
0925F9A9-4D53-4B27-A87E-
3D83A757B0E0@yandex-team.ru
John Naylor [Tue, 20 Aug 2024 03:02:34 +0000 (10:02 +0700)] Document limit on the number of out-of-line values per table
Document the hard limit stemming from the size of an OID, and also
mention the perfomance impact that occurs before the hard limit
is reached.
Jakub Wartak and Robert Haas
Back to all supported versions
Discussion: https://postgr.es/m/CAKZiRmwWhp2yxjqJLwbBjHdfbJBcUmmKMNAZyBjjtpgM9AMatQ%40mail.gmail.com
Amit Kapila [Tue, 20 Aug 2024 03:05:11 +0000 (08:35 +0530)] Log the conflicts while applying changes in logical replication.
This provides the additional logging information in the following
conflict scenarios while applying changes:
insert_exists: Inserting a row that violates a NOT DEFERRABLE unique constraint.
update_differ: Updating a row that was previously modified by another origin.
update_exists: The updated row value violates a NOT DEFERRABLE unique constraint.
update_missing: The tuple to be updated is missing.
delete_differ: Deleting a row that was previously modified by another origin.
delete_missing: The tuple to be deleted is missing.
For insert_exists and update_exists conflicts, the log can include the origin
and commit timestamp details of the conflicting key with track_commit_timestamp
enabled.
update_differ and delete_differ conflicts can only be detected when
track_commit_timestamp is enabled on the subscriber.
We do not offer additional logging for exclusion constraint violations because
these constraints can specify rules that are more complex than simple equality
checks. Resolving such conflicts won't be straightforward. This area can be
further enhanced if required.
Author: Hou Zhijie
Reviewed-by: Shveta Malik, Amit Kapila, Nisha Moond, Hayato Kuroda, Dilip Kumar
Discussion: https://postgr.es/m/OS0PR01MB5716352552DFADB8E9AD1D8994C92@OS0PR01MB5716.jpnprd01.prod.outlook.com
David Rowley [Tue, 20 Aug 2024 01:38:22 +0000 (13:38 +1200)] Speed up Hash Join by making ExprStates support hashing
Here we add ExprState support for obtaining a 32-bit hash value from a
list of expressions. This allows both faster hashing and also JIT
compilation of these expressions. This is especially useful when hash
joins have multiple join keys as the previous code called ExecEvalExpr on
each hash join key individually and that was inefficient as tuple
deformation would have only taken into account one key at a time, which
could lead to walking the tuple once for each join key. With the new
code, we'll determine the maximum attribute required and deform the tuple
to that point only once.
Some performance tests done with this change have shown up to a 20%
performance increase of a query containing a Hash Join without JIT
compilation and up to a 26% performance increase when JIT is enabled and
optimization and inlining were performed by the JIT compiler. The
performance increase with 1 join column was less with a 14% increase
with and without JIT. This test was done using a fairly small hash
table and a large number of hash probes. The increase will likely be
less with large tables, especially ones larger than L3 cache as memory
pressure is more likely to be the limiting factor there.
This commit only addresses Hash Joins, but lays expression evaluation
and JIT compilation infrastructure for other hashing needs such as Hash
Aggregate.
Author: David Rowley
Reviewed-by: Alexey Dvoichenkov <[email protected]>Reviewed-by: Tels <[email protected]>Discussion: https://postgr.es/m/CAApHDvoexAxgQFNQD_GRkr2O_eJUD1-wUGm%3Dm0L%2BGc%3DT%3DkEa4g%40mail.gmail.com
Bruce Momjian [Tue, 20 Aug 2024 00:18:03 +0000 (20:18 -0400)] doc: improve create/alter sequence CYCLE syntax
Reported-by: Peter Smith
Discussion: https://postgr.es/m/CAHut+PtqwZwPfGq62xq2614_ce2ejDmbB9CfP+a1azxpneFRBQ@mail.gmail.com
Author: Peter Smith
Back-through: master
Bruce Momjian [Mon, 19 Aug 2024 23:54:39 +0000 (19:54 -0400)] doc: mention of postpostgres_fdw INSERT ON CONFLICT limitation
Reported-by: Fujii MasaoDiscussion: https://postgr.es/m/
47801526-d017-4c89-9f52-
c02c449a139b@oss.nttdata.com
Author: Fujii Masao
Back-through: master
Bruce Momjian [Mon, 19 Aug 2024 23:22:10 +0000 (19:22 -0400)] doc: clarify create database in start docs uses command line
Reported-by: [email protected]Discussion: https://postgr.es/m/
172251463564.915373.
17748961617119647662@wrigleys.postgresql.org
Back-through: master
Bruce Momjian [Mon, 19 Aug 2024 22:27:22 +0000 (18:27 -0400)] doc: Improve vague pg_createsubscriber description
Discussion: https://postgr.es/m/
[email protected]Author: Euler Taveira
Back-through: 17
Alvaro Herrera [Mon, 19 Aug 2024 20:09:10 +0000 (16:09 -0400)] Avoid failure to open dropped detached partition
When a partition is detached and immediately dropped, a prepared
statement could try to compute a new partition descriptor that includes
it. This leads to this kind of error:
ERROR: could not open relation with OID 457639
Avoid this by skipping the partition in expand_partitioned_rtentry if it
doesn't exist.
Noted by me while investigating bug #18559. Kuntal Gosh helped to
identify the exact failure.
Back to 14, where DETACH CONCURRENTLY was introduced.
Author: Álvaro Herrera <
[email protected]>
Reviewed-by: Kuntal Ghosh <[email protected]>Reviewed-by: Junwang Zhao <[email protected]>Discussion: https://postgr.es/m/
202408122233[email protected]Tomas Vondra [Mon, 19 Aug 2024 17:46:05 +0000 (19:46 +0200)] Document that search_path is reported by the server
Commit
28a1121fd912 marked search_path as GUC_REPORT, but failed to
update the relevant places in docs. There are two places listing the GUC
options reported to the client, so update both.
Reported-by: Tom LaneDiscussion: https://postgr.es/m/CAFh8B=k8s7WrcqhafmYhdN1+E5LVzZi_QaYDq8bKvrGJTAhY2Q@mail.gmail.com
Tomas Vondra [Mon, 19 Aug 2024 15:04:09 +0000 (17:04 +0200)] Mark search_path as GUC_REPORT
Report search_path changes to the client. Multi-tenant applications
often map tenants to schemas, and use search_path to pick the tenant a
given connection works with. This breaks when a connection pool (like
PgBouncer), because the search_path may change unexpectedly.
There are other GUCs we might want reported (e.g. various timeouts), but
search_path is by far the biggest foot gun that can lead either to
puzzling failures during query execution (when objects are missing or
are defined differently), or even to accessing incorrect data.
Many existing tools modify search_path, pg_dump being a notable example.
Ideally, clients could specify which GUCs are interesting and should be
subject to this reporting, but we don't support that. GUC_REPORT is what
connection pools rely on for other interesting GUCs, so just use that.
When this change was initially proposed in 2014, one of the concerns was
impact on performance. But this was addressed by commit
2432b1a04087,
which ensures we report each GUC at most once per query, no matter how
many times it changed during execution.
Eventually, this might be replaced / superseded by allowing doing this
by making the protocol extensible in this direction, but it's unclear
when (or if) that happens. Until then, we can leverage GUC_REPORT.
Author: Alexander Kukushkin, Jelte Fennema-Nio
Discussion: https://postgr.es/m/CAFh8B=k8s7WrcqhafmYhdN1+E5LVzZi_QaYDq8bKvrGJTAhY2Q@mail.gmail.com
Tomas Vondra [Mon, 19 Aug 2024 11:31:51 +0000 (13:31 +0200)] Explain dropdb can't use syscache because of TOAST
Add a comment explaining dropdb() can't rely on syscache. The issue with
flattened rows was fixed by commit
0f92b230f88b, but better to have
a clear explanation why the systable scan is necessary. The other places
doing in-place updates on pg_database have the same comment.
Suggestion and by Yugo Nagata. Back to 12, same as the fix.
Author: Yugo Nagata
Back-through: 12
Discussion: https://postgr.es/m/CAJTYsWWNkCt+-UnMhg=BiCD3Mh8c2JdHLofPxsW3m2dkDFw8RA@mail.gmail.com
Daniel Gustafsson [Mon, 19 Aug 2024 10:55:11 +0000 (12:55 +0200)] Fix regression in TLS session ticket disabling
Commit
274bbced disabled session tickets for TLSv1.3 on top of the
already disabled TLSv1.2 session tickets, but accidentally caused
a regression where TLSv1.2 session tickets were incorrectly sent.
Fix by unconditionally disabling TLSv1.2 session tickets and only
disable TLSv1.3 tickets when the right version of OpenSSL is used.
Back to all supported branches.
Reported-by: Cameron Vogt <[email protected]>Reported-by: Fire Emerald <[email protected]>Reviewed-by: Jacob Champion <[email protected]>Discussion: https://postgr.es/m/DM6PR16MB3145CF62857226F350C710D1AB852@DM6PR16MB3145.namprd16.prod.outlook.com
Back-through: v12
Thomas Munro [Mon, 19 Aug 2024 09:21:03 +0000 (21:21 +1200)] Fix harmless LC_COLLATE[_MASK] confusion.
Commit
ca051d8b101 called newlocale(LC_COLLATE, ...) instead of
newlocale(LC_COLLATE_MASK, ...), in code reached only on FreeBSD. They
have the same value on that OS, explaining why it worked. Fix.
Back- to 14, where
ca051d8b101 landed.
Heikki Linnakangas [Mon, 19 Aug 2024 06:48:25 +0000 (09:48 +0300)] Fix garbled process name on backend crash
The log message on backend crash used wrong variable, which could be
uninitialized. Introduced in commit
28a520c0b7.
Reported-by: Alexander LakhinDiscussion: https://www.postgresql.org/message-id/
451b0797-83b8-cdbc-727f-
8d7a7b0e3bca@gmail.com
Michael Paquier [Mon, 19 Aug 2024 03:34:18 +0000 (12:34 +0900)] Fix more holes with SLRU code in need of int64 for segment numbers
This is a continuation of
c9e24573905b, containing changes included into
the proposed that have been missed in the actual commit. I have
managed to miss these diffs while doing a rebase of the original .
Thanks to Noah Misch, Peter Eisentraut and Alexander Korotkov for the
pokes.
Discussion: https://postgr.es/m/
92fe572d-638e-4162-aef6-
1c42a2936f25@eisentraut.org
Discussion: https://postgr.es/m/
20240810175055[email protected]Back-through: 17
Alvaro Herrera [Mon, 19 Aug 2024 00:49:57 +0000 (20:49 -0400)] Search for SLRU page only in its own bank
One of the two slot scans in SlruSelectLRUPage was not walking only the
slots in the specific bank where the buffer could be; change it to do
that.
Oversight in
53c2a97a9266.
Author: Sergey Sargsyan <
[email protected]>
Discussion: https://postgr.es/m/18582-
5f301dd30ba91a38@postgresql.org
Michael Paquier [Mon, 19 Aug 2024 00:03:52 +0000 (09:03 +0900)] injection_points: Add stats for point caching and loading
This adds two counters to the fixed-numbered stats of injection points
to track the number of times injection points have been cached and
loaded from the cache, as of the additions coming from
a0a5869a8598 and
4b211003ecc2.
These should have been part of
f68cd847fa40, but I have lacked time and
energy back then, and it did not prevent the code to be a useful
template.
While on it, this commit simplifies the description of a few tests while
adding coverage for the new stats data.
Author: Yogesh Sharma
Discussion: https://postgr.es/m/
3a6977f7-54ab-43ce-8806-
11d5e15526a2@catprosystems.com
Thomas Munro [Sun, 18 Aug 2024 23:47:37 +0000 (11:47 +1200)] ci: Upgrade MacPorts version to 2.10.1.
MacPorts version 2.9.3 started failing in our ci_macports_packages.sh
script, for reasons not fully determined, but plausibly linked to the
release of 2.10.1. 2.10.1 seems to work, so let's switch to it.
Back- to 15, where CI began.
Reported-by: Peter Eisentraut <[email protected]>Discussion: https://postgr.es/m/
81f104e8-f0a9-43c0-85bd-
2bbbf590a5b8%40eisentraut.org
Michael Paquier [Sun, 18 Aug 2024 22:53:47 +0000 (07:53 +0900)] doc: Fix typo in section for custom pgstats
Per offline report from Erik Rijkers.
Tomas Vondra [Sun, 18 Aug 2024 22:04:41 +0000 (00:04 +0200)] Fix DROP DATABASE for databases with many ACLs
Commit
c66a7d75e652 modified DROP DATABASE so that if interrupted, the
database is known to be in an invalid state and can only be dropped.
This is done by setting a flag using an in-place update, so that it's
not lost in case of rollback.
For databases with many ACLs, this may however fail like this:
ERROR: wrong tuple length
This happens because with many ACLs, the pg_database.datacl attribute
gets TOASTed. The dropdb() code reads the tuple from the syscache, which
means it's detoasted. But the in-place update expects the tuple length
to match the on-disk tuple.
Fixed by reading the tuple from the catalog directly, not from syscache.
Report and fix by Ayush Tiwari. Back to 12. The DROP DATABASE fix
was backed to 11, but 11 is EOL at this point.
Reported-by: Ayush TiwariAuthor: Ayush Tiwari
Reviewed-by: Tomas VondraBack-through: 12
Discussion: https://postgr.es/m/CAJTYsWWNkCt+-UnMhg=BiCD3Mh8c2JdHLofPxsW3m2dkDFw8RA@mail.gmail.com
Thomas Munro [Sun, 18 Aug 2024 19:59:16 +0000 (07:59 +1200)] Fix cpluspluscheck for pg_verifybackup.h.
simplehash.h references pg_fatal(), which cpluspluscheck says is
undeclared, causing the CI CompilerWarnings task to fail since commit
aa2d6b15. Include the header it needs.
Discussion: https://postgr.es/m/CA%2BhUKGJC3d4PXkErpfOWrzQqcq6MLiCv0%2BAH0CMQnB6hdLUFEw%40mail.gmail.com
Noah Misch [Sun, 18 Aug 2024 19:03:59 +0000 (12:03 -0700)] Fix comments on wal_level=minimal, CREATE TABLESPACE and CREATE DATABASE.
Commit
97ddda8a82ac470ae581d0eb485b6577707678bc removed the rmtree()
behavior from XLOG_TBLSPC_CREATE, obsoleting that part of the comment.
The comment's point about XLOG_DBASE_CREATE was wrong when commit
fa0f466d5329e10b16f3b38c8eaf5306f7e234e8 introduced the point. (It
would have been accurate if that commit had predated commit
fbcbc5d06f53aea412130deb52e216aa3883fb8d introducing the second
checkpoint of CREATE DATABASE.) Nothing can skip log_smgrcreate() on
the basis of wal_level=minimal, so don't comment on that.
Commit
c6b92041d38512a4176ed76ad06f713d2e6c01a8 expanded WAL skipping
from five specific operations to relfilenodes generally, hence the
CreateDatabaseUsingFileCopy() comment change.
Discussion: https://postgr.es/m/
20231008022204[email protected]Bruce Momjian [Sat, 17 Aug 2024 02:50:54 +0000 (22:50 -0400)] docs: fix incorrect plpgsql error message
Change "$1" to "username".
Reported-by: [email protected]Discussion: https://postgr.es/m/
172112109590.736590.
12219129462878821880@wrigleys.postgresql.org
Back-through: 12
Bruce Momjian [Sat, 17 Aug 2024 01:11:55 +0000 (21:11 -0400)] C comment: fix for commit
b5a9b18cd0bThe commit was "Provide API for relation data.".
Reported-by: Nazir Bilal YavuzDiscussion: https://postgr.es/m/CAN55FZ3KsZ2faZs1sK5J0W+_8B3myB232CfLYGie4u4BBMwP3g@mail.gmail.com
Back-through: master
David Rowley [Fri, 16 Aug 2024 22:36:23 +0000 (10:36 +1200)] Relocate a badly placed Assert in COPY FROM code
There's not much point in asserting a pointer isn't NULL after some code
has already dereferenced that pointer.
Adjust the code so that the Assert occurs before the pointer dereference.
The Assert probably has questionable value in the first place, but it
seems worth keeping around to document the contract between
CopyMultiInsertInfoNextFreeSlot() and its callers.
Author: Amul Sul <
[email protected]>
Discussion: https://postgr.es/m/CAAJ_b94hXQzXaJxTLShkxQUgezf_SUxhzX9TH2f-g6gP7bne7g@mail.gmail.com
Nathan Bossart [Fri, 16 Aug 2024 20:06:40 +0000 (15:06 -0500)] Further reduce dependence on -fwrapv semantics in jsonb.
Commit
108d2adb9e missed updating a few places in the jsonb code
that rely on signed integer wrapping for correctness. These can
also be fixed by using pg_abs_s32() to negate a signed integer
(that is known to be negative) for comparison with an unsigned
integer.
Reported-by: Alexander LakhinDiscussion: https://postgr.es/m/
bfff906f-300d-81ea-83b7-
f2c93845e7f2%40gmail.com
Robert Haas [Fri, 16 Aug 2024 19:09:42 +0000 (15:09 -0400)] pg_verifybackup: Move some declarations to new pg_verifybackup.h
This is in preparation for adding a second source file to this
directory.
Amul Sul, reviewed by Sravan Kumar and revised a bit by me.
Discussion: http://postgr.es/m/CAAJ_b95mcGjkfAf1qduOR97CokW8-_i-dWLm3v6x1w2-OW9M+A@mail.gmail.com
Robert Haas [Fri, 16 Aug 2024 18:52:52 +0000 (14:52 -0400)] pg_verifybackup: Move skip_checksums into verifier_context.
This is in preparation for adding a second source file to this
directory. It will need access to this value. Also, fewer global
variables is usually a good thing.
Amul Sul, reviewed by Sravan Kumar and revised a bit by me.
Discussion: http://postgr.es/m/CAAJ_b95mcGjkfAf1qduOR97CokW8-_i-dWLm3v6x1w2-OW9M+A@mail.gmail.com
Robert Haas [Fri, 16 Aug 2024 17:34:18 +0000 (13:34 -0400)] Improve more comments in astreamer_gzip.c.
Duplicate the comment from astreamer_plain_writer_new instead of just
referring to it. Add a further note to mention that there are dangers
if anything else is written to the same FILE. Also add a comment where
we dup() the filehandle, referring to the existing comment in
astreamer_gzip_writer_finalize(), because the dup() looks wrong on
first glance without that comment to clarify.
Per concerns expressed by Tom Lane on pgsql-security, and using
some wording suggested by him.
Discussion: http://postgr.es/m/CA+TgmoYTFAD0YTh4HC1Nuhn0YEyoQi0_CENFgVzAY_YReiSksQ@mail.gmail.com
Alvaro Herrera [Fri, 16 Aug 2024 17:23:18 +0000 (13:23 -0400)] libpq: Trace all messages received from the server
Not all messages that libpq received from the server would be sent
through our message tracing logic. This commit tries to fix that by
introducing a new function pqParseDone which make it harder to forget
about doing so.
The messages that we now newly send through our tracing logic are:
- CopyData (received by COPY TO STDOUT)
- Authentication requests
- NegotiateProtocolVersion
- Some ErrorResponse messages during connection startup
- ReadyForQuery when received after a FunctionCall message
Author: Jelte Fennema-Nio <
[email protected]>
Discussion: https://postgr.es/m/CAGECzQSoPHtZ4xe0raJ6FYSEiPPS+YWXBhOGo+Y1YecLgknF3g@mail.gmail.com
Tom Lane [Fri, 16 Aug 2024 16:35:50 +0000 (12:35 -0400)] Fix extraction of week and quarter fields from intervals.
"EXTRACT(WEEK FROM interval_value)" formerly threw an error.
Define it as "tm->tm_mday / 7". (With C99 division semantics,
this gives consistent results for negative intervals.)
"EXTRACT(QUARTER FROM interval_value)" has been implemented
all along, but it formerly gave extremely strange results for
negative intervals. Fix it so that the output for -N months
is the negative of the output for N months.
Per bug #18348 from Michael Bondarenko and subsequent discussion.
Discussion: https://postgr.es/m/18348-
b097a3587dfde8a4@postgresql.org
Nathan Bossart [Fri, 16 Aug 2024 16:24:44 +0000 (11:24 -0500)] Remove dependence on -fwrapv semantics in jsonb.
This commit updates a couple of places in the jsonb code to no
longer rely on signed integer wrapping for correctness. Like
commit
9e9a2b7031, this is intended to move us closer towards
removing -fwrapv, which may enable some compiler optimizations.
However, there is presently no plan to actually remove that
compiler option in the near future.
This commit makes use of the newly introduced pg_abs_s32() routine
to negate a signed integer (that is known to be negative) for
comparison with an unsigned integer. In passing, change one use of
INT_MIN to the more portable PG_INT32_MIN.
Reported-by: Alexander LakhinAuthor: Joseph Koshakow
Reviewed-by: Jian HeDiscussion: https://postgr.es/m/CAAvxfHdBPOyEGS7s%2Bxf4iaW0-cgiq25jpYdWBqQqvLtLe_t6tw%40mail.gmail.com
Peter Eisentraut [Fri, 16 Aug 2024 15:14:32 +0000 (17:14 +0200)] Remove incidental md5() function use from test
To allow test to pass in OpenSSL FIPS mode, similar to
657f5f223e, for
a new test that has been added since.
Reviewed-by: Tomas Vondra <[email protected]>Discussion: https://www.postgresql.org/message-id/
86763810-70a1-4872-8ba7-
1676f788e5a2@eisentraut.org
Heikki Linnakangas [Fri, 16 Aug 2024 11:45:37 +0000 (14:45 +0300)] Relax fsyncing at end of a bulk load that was not WAL-logged
And improve the comments.
Back to v17 where this was introduced.
Reviewed-by: Noah MischDiscussion: https://www.postgresql.org/message-id/
cac7d1b6-8358-40be-af0b-
21bc9b27d34c@iki.fi
Heikki Linnakangas [Fri, 16 Aug 2024 10:48:10 +0000 (13:48 +0300)] Refactor CopyOneRowTo
The handling of binary and text formats are quite different here, so
it's more clear to check for the format first and have two separate
loops.
Author: jian he <
[email protected]>
Reviewed-by: Ilia Evdokimov, Junwang ZhaoDiscussion: https://www.postgresql.org/message-id/CACJufxFzHCeFBQF0M%
[email protected]Heikki Linnakangas [Fri, 16 Aug 2024 10:13:43 +0000 (13:13 +0300)] Remove unused 'cur_skey' argument from IndexScanOK()
Commit
a78fcfb51243 removed the last use of it.
Author: Hugo Zhang, Aleksander Alekseev
Reviewed-by: Daniel GustafssonDiscussion: https://www.postgresql.org/message-id/NT0PR01MB129459E243721B954611938F9CDD2%40NT0PR01MB1294.CHNPR01.prod.partner.outlook.cn
Peter Eisentraut [Fri, 16 Aug 2024 04:41:17 +0000 (06:41 +0200)] libpq: Fix minor TOCTOU violation
libpq checks the permissions of the password file before opening it.
The way this is done in two separate operations, a static analyzer
would flag as a time-of-check-time-of-use violation. In practice, you
can't do anything with that, but it still seems better style to fix
it.
To fix it, open the file first and then check the permissions on the
opened file handle.
Reviewed-by: Aleksander Alekseev <[email protected]>Reviewed-by: Andreas Karlsson <[email protected]>Discussion: https://www.postgresql.org/message-id/flat/
a3356054-14ae-4e7a-acc6-
249d19dac20b%40eisentraut.org
Alexander Korotkov [Thu, 15 Aug 2024 21:58:32 +0000 (00:58 +0300)] Add missing wait_for_catchup() to pg_visibility tap test
e2ed7e32271a introduced check of pg_visibility on standby. This commit adds
missing wait_for_catchup() to synchronize standby before querying it.
Alexander Korotkov [Thu, 15 Aug 2024 21:17:59 +0000 (00:17 +0300)] Fix GetStrictOldestNonRemovableTransactionId() on standby
e85662df44 implemented GetStrictOldestNonRemovableTransactionId() function
for computation of xid horizon that avoid reporting of false errors.
However, GetStrictOldestNonRemovableTransactionId() uses
GetRunningTransactionData() even on standby leading to an assertion failure.
Given that we decided to ignore KnownAssignedXids and standby can't have
own running xids, we switch to use TransamVariables->nextXid as a xid horizon.
Also, revise the comment regarding ignoring KnownAssignedXids with more
detailed reasoning provided by Heikki.
Reported-by: Heikki LinnakangasDiscussion: https://postgr.es/m/
42218c4f-2c8d-40a3-8743-
4d34dd0e4cce%40iki.fi
Reviewed-by: Heikki LinnakangasNathan Bossart [Thu, 15 Aug 2024 20:47:31 +0000 (15:47 -0500)] Remove dependence on -fwrapv semantics in a few places.
This commit attempts to update a few places, such as the money,
numeric, and timestamp types, to no longer rely on signed integer
wrapping for correctness. This is intended to move us closer
towards removing -fwrapv, which may enable some compiler
optimizations. However, there is presently no plan to actually
remove that compiler option in the near future.
Besides using some of the existing overflow-aware routines in
int.h, this commit introduces and makes use of some new ones.
Specifically, it adds functions that accept a signed integer and
return its absolute value as an unsigned integer with the same
width (e.g., pg_abs_s64()). It also adds functions that accept an
unsigned integer, store the result of negating that integer in a
signed integer with the same width, and return whether the negation
overflowed (e.g., pg_neg_u64_overflow()).
Finally, this commit adds a couple of tests for timestamps near
POSTGRES_EPOCH_JDATE.
Author: Joseph Koshakow
Reviewed-by: Tom Lane, Heikki Linnakangas, Jian He
Discussion: https://postgr.es/m/CAAvxfHdBPOyEGS7s%2Bxf4iaW0-cgiq25jpYdWBqQqvLtLe_t6tw%40mail.gmail.com
Tom Lane [Thu, 15 Aug 2024 15:43:55 +0000 (11:43 -0400)] Add
97add39c0 to .git-blame-ignore-revs.
Tom Lane [Thu, 15 Aug 2024 15:41:46 +0000 (11:41 -0400)] Clean up indentation and whitespace inconsistencies in ecpg.
ecpg's lexer and parser files aren't normally processed by
pgindent, and unsurprisingly there's a lot of code in there
that doesn't really match project style. I spent some time
running pgindent over the fragments of these files that are
C code, and this is the result. This is in the same spirit
as commit
30ed71e42, though apparently Peter used a different
method for that one, since it didn't find these problems.
Discussion: https://postgr.es/m/
2011420.
1713493114@sss.pgh.pa.us
Robert Haas [Thu, 15 Aug 2024 14:44:15 +0000 (10:44 -0400)] Do not hardcode PG_PROTOCOL_LATEST in NegotiateProtocolVersion
We shouldn't ask the client to use a protocol version later than the
one that they requested. To avoid that, if the client requests a
version newer than the latest one we support, set FrontendProtocol
to the latest version we support, not the requested version. Then,
use that value when building the NegotiateProtocolVersion message.
(It seems good on general principle to avoid setting FrontendProtocol
to a version we don't support, anyway.)
None of this really matters right now, because we only support a
single protocol version, but if that ever changes, we'll need this.
Jelte Fennema-Nio, reviewed by me and incorporating some of my
proposed wording
Discussion: https://postgr.es/m/CAGECzQTyXDNtMXdq2L-Wp=OvOCPa07r6+U_MGb==h90MrfT+fQ@mail.gmail.com
Dean Rasheed [Thu, 15 Aug 2024 09:36:17 +0000 (10:36 +0100)] Optimise numeric multiplication using base-NBASE^2 arithmetic.
Currently mul_var() uses the schoolbook multiplication algorithm,
which is O(n^2) in the number of NBASE digits. To improve performance
for large inputs, convert the inputs to base NBASE^2 before
multiplying, which effectively halves the number of digits in each
input, theoretically speeding up the computation by a factor of 4. In
practice, the actual speedup for large inputs varies between around 3
and 6 times, depending on the system and compiler used. In turn, this
significantly reduces the runtime of the numeric_big regression test.
For this to work, 64-bit integers are required for the products of
base-NBASE^2 digits, so this works best on 64-bit machines, on which
it is faster whenever the shorter input has more than 4 or 5 NBASE
digits. On 32-bit machines, the additional overheads, especially
during carry propagation and the final conversion back to base-NBASE,
are significantly higher, and it is only faster when the shorter input
has more than around 50 NBASE digits. When the shorter input has more
than 6 NBASE digits (so that mul_var_short() cannot be used), but
fewer than around 50 NBASE digits, there may be a noticeable slowdown
on 32-bit machines. That seems to be an acceptable tradeoff, given the
performance gains for other inputs, and the effort that would be
required to maintain code specifically targeting 32-bit machines.
Joel Jacobson and Dean Rasheed.
Discussion: https://postgr.es/m/
9d8a4a42-c354-41f3-bbf3-
199e1957db97%40app.fastmail.com
Dean Rasheed [Thu, 15 Aug 2024 09:33:12 +0000 (10:33 +0100)] Extend mul_var_short() to 5 and 6-digit inputs.
Commit
ca481d3c9a introduced mul_var_short(), which is used by
mul_var() whenever the shorter input has 1-4 NBASE digits and the
exact product is requested. As speculated on in that commit, it can be
extended to work for more digits in the shorter input. This commit
extends it up to 6 NBASE digits (up to 24 decimal digits), for which
it also gives a significant speedup. This covers more cases likely to
occur in real-world queries, for which using base-NBASE^2 arithmetic
provides little benefit.
To avoid code bloat and duplication, refactor it a bit using macros
and exploiting the fact that some portions of the code are shared
between the different cases.
Dean Rasheed, reviewed by Joel Jacobson.
Discussion: https://postgr.es/m/
9d8a4a42-c354-41f3-bbf3-
199e1957db97%40app.fastmail.com
Peter Eisentraut [Thu, 15 Aug 2024 05:08:12 +0000 (07:08 +0200)] Variable renaming in dbcommands.c
There were several sets of very similar local variable names, such as
"downer" and "dbowner", which was very confusing and error-prone.
Rename the former to "ownerEl" and so on, similar to collationcmds.c
and typecmds.c.
Reviewed-by: Daniel Gustafsson <[email protected]>Discussion: https://www.postgresql.org/message-id/flat/
e5bce225-ee04-40c7-a280-
ea7214318048%40eisentraut.org
Jeff Davis [Thu, 15 Aug 2024 02:05:39 +0000 (19:05 -0700)] Fix doc typo: unicode_assigned() return type.
Reported-by: Hironobu SUZUKIDiscussion: https://postgr.es/m/
5dd88820-bb00-4b90-904b-
738ea2e4ee2e@interdb.jp
Back-through: 17
David Rowley [Thu, 15 Aug 2024 01:10:25 +0000 (13:10 +1200)] Improve ALTER PUBLICATION validation and error messages
Attempting to add a system column for a table to an existing publication
would result in the not very intuitive error message of:
ERROR: negative bitmapset member not allowed
Here we improve that to have it display the same error message as a user
would see if they tried adding a system column for a table when adding
it to the publication in the first place.
Doing this requires making the function which validates the list of
columns an extern function. The signature of the static function wasn't
an ideal external API as it made the code more complex than it needed to be.
Here we adjust the function to have it populate a Bitmapset of attribute
numbers. Doing it this way allows code simplification.
There was no particular bug here other than the weird error message, so
no back.
Bug: #18558
Reported-by: Alexander Lakhin <[email protected]>Author: Peter Smith, David Rowley
Discussion: https://postgr.es/m/18558-
411bc81b03592125@postgresql.org
Nathan Bossart [Wed, 14 Aug 2024 19:25:54 +0000 (14:25 -0500)] Add a couple of recent commits to .git-blame-ignore-revs.
Alvaro Herrera [Wed, 14 Aug 2024 18:53:55 +0000 (14:53 -0400)] libpq: Trace responses to SSLRequest and GSSENCRequest
Since these are single bytes instead of v2 or v3 messages they need
custom tracing logic. These "messages" don't even have official names
in the protocol specification, so I (Jelte) called them SSLResponse and
GSSENCResponse here.
Author: Jelte Fennema-Nio <
[email protected]>
Discussion: https://postgr.es/m/CAGECzQSoPHtZ4xe0raJ6FYSEiPPS+YWXBhOGo+Y1YecLgknF3g@mail.gmail.com
Peter Eisentraut [Wed, 14 Aug 2024 09:36:12 +0000 (11:36 +0200)] Apply PGDLLIMPORT markings to some GUC variables
According to the commit message in
8ec569479, we must have all variables
in header files marked with PGDLLIMPORT. In commit
d3cc5ffe81f6 some
variables were moved from launch_backend.c file to several header files.
This adds PGDLLIMPORT to moved variables.
Author: Sofia Kopikova <
[email protected]>
Reviewed-by: Robert Haas <[email protected]>Discussion: https://www.postgresql.org/message-id/flat/
e0b17014-5319-4dd6-91cd-
93d9c8fc9539%40postgrespro.ru
Peter Eisentraut [Wed, 14 Aug 2024 06:02:32 +0000 (08:02 +0200)] Remove TRACE_SORT macro
The TRACE_SORT macro guarded the availability of the trace_sort GUC
setting. But it has been enabled by default ever since it was
introduced in PostgreSQL 8.1, and there have been no reports that
someone wanted to disable it. So just remove the macro to simplify
things. (For the avoidance of doubt: The trace_sort GUC is still
there. This only removes the rarely-used macro guarding it.)
Reviewed-by: Heikki Linnakangas <[email protected]>Discussion: https://www.postgresql.org/message-id/flat/
be5f7162-7c1d-44e3-9a78-
74dcaa6529f2%40eisentraut.org
Thomas Munro [Wed, 14 Aug 2024 03:02:12 +0000 (15:02 +1200)] Harmonize MinGW CODESET lookup with MSVC.
Historically, MinGW environments lacked some Windows API calls, so we
took a different code path in win32_langinfo(). Somehow, the code
change in commit
35eeea62 (removing setlocale() calls) caused one
particular 001_initdb.pl test to fail on MinGW + ICU builds, because
pg_import_system_collations() found no collations. It might take a
MinGW user to discover the exact reason.
Updating that function to use the same code as MSVC seems to fix that
test, so lets do that. (There are plenty more places that test for MSVC
unnecessarily, to be investigated later.)
While here, also rename the helper function win32_langinfo() to
win32_get_codeset(), to explain what it does less confusingly; it's not
really a general langinfo() substitute.
Noticed by triggering the optional MinGW CI task; no build farm animals
failed.
Discussion: https://postgr.es/m/CA%2BhUKGKBWfhXQ3J%2B2Lj5PhKvQnGD%3DsywA0XQcb7boTCf%3DerVLg%40mail.gmail.com
Masahiko Sawada [Wed, 14 Aug 2024 02:23:56 +0000 (19:23 -0700)] Add resource statistics reporting to ANALYZE VERBOSE.
Previously, log_autovacuum_min_duration utilized dedicated code for
logging resource statistics, such as system and buffer usage during
autoanalyze. However, this logging functionality was not utilized by
ANALYZE VERBOSE.
This commit adds resource statistics reporting to ANALYZE VERBOSE by
reusing the same logging code as autoanalyze.
Author: Anthonin Bonnefoy
Reviewed-by: Masahiko Sawada
Discussion: https://postgr.es/m/CAO6_Xqr__kTTCLkftqS0qSCm-J7_xbRG3Ge2rWhucxQJMJhcRA%40mail.gmail.com
Masahiko Sawada [Wed, 14 Aug 2024 01:49:45 +0000 (18:49 -0700)] Use pgBufferUsage for buffer usage tracking in analyze.
Previously, (auto)analyze used global variables VacuumPageHit,
VacuumPageMiss, and VacuumPageDirty to track buffer usage. However,
pgBufferUsage provides a more generic way to track buffer usage with
support functions.
This change replaces those global variables with pgBufferUsage in
analyze. Since analyze was the sole user of those variables, it
removes their declarations. Vacuum previously used those variables but
replaced them with pgBufferUsage as part of a bug fix, commit
5cd72cc0c.
Additionally, it adjusts the buffer usage message in both vacuum and
analyze for better consistency.
Author: Anthonin Bonnefoy
Reviewed-by: Masahiko Sawada, Michael PaquierDiscussion: https://postgr.es/m/CAO6_Xqr__kTTCLkftqS0qSCm-J7_xbRG3Ge2rWhucxQJMJhcRA%40mail.gmail.com
Thomas Munro [Tue, 13 Aug 2024 11:42:58 +0000 (23:42 +1200)] Include <xlocale.h> for macOS, take II.
Fix typo in macro name.
Discussion: https://postgr.es/m/CA%2BhUKG%2Bk-o3N_SyNJNJpAcdtMo_HheN30miAeXehk9yw%3D9WYzA%40mail.gmail.com
Thomas Munro [Tue, 13 Aug 2024 11:02:05 +0000 (23:02 +1200)] Include <xlocale.h> for older macOS.
Commit
35eeea62 forgot to include <xlocale.h> when using locale_t
(which didn't seem to be required on newer Apple SDK as used by CI,
hence mistake). Let's see if this fixes build farm animals longfin and
sifika.
Thomas Munro [Tue, 13 Aug 2024 10:27:16 +0000 (22:27 +1200)] Use thread-safe nl_langinfo_l(), not nl_langinfo().
This gets rid of some setlocale() calls. The remaining call to
setlocale() in pg_get_encoding_from_locale() is a query of the name
of the current locale when none was provided (in a multi-threaded future
that would need more work).
All known non-Windows targets have nl_langinfo_l(), from POSIX 2008, and
for Windows we already do something thread-safe.
Reviewed-by: Heikki Linnakangas <[email protected]>Discussion: https://postgr.es/m/CA%2BhUKGJqVe0%2BPv9dvC9dSums_PXxGo9SWcxYAMBguWJUGbWz-A%40mail.gmail.com