aboutsummaryrefslogtreecommitdiff
path: root/src
Commit message (Collapse)AuthorAge
* Take care to de-duplicate entries in standby.c's table of locks.Tom Lane2022-10-06
| | | | | | | | | | | | | | | | | | | | | | | | | | | | The RecoveryLockLists data structure, which tracks all exclusive locks that the startup process is holding on behalf of transactions being replayed, did not have any provision for avoiding duplicate entries for the same lock. Maybe that was okay when the code was first written. However, modern practice is for checkpoints to write fresh lists of all active exclusive locks into the WAL. Thus, an exclusive lock that survives across multiple checkpoints causes bloat in standbys' startup processes. If there are a lot of such locks this can look like a memory leak, and it's even possible to drive the startup process into a palloc failure from an over-length List. To fix, use a hash table instead of simple lists to track the locks being held. Allowing for dynahash overhead, this requires a little more space per lock than the old way (although it's the same size as what we were allocating prior to c6e0fe1f2). It's probably a shade slower too. However, testing indicates that the penalty is negligible on ordinary workloads, so let's make this change to improve robustness in extreme cases. Patch by me, per report from Dmitriy Kuzmin. No back-patch (for now anyway), since it seems that a significant improvement would only occur in corner cases. Discussion: https://postgr.es/m/CAHLDt=_ts0A7Agn=hCpUh+RCFkxd+G6uuT=kcTfqFtGur0dp=A@mail.gmail.com
* Introduce t_isalnum() to replace t_isalpha() || t_isdigit() tests.Tom Lane2022-10-06
| | | | | | | | | | | | ts_locale.c omitted support for "isalnum" tests, perhaps on the grounds that there were initially no use-cases for that. However, both ltree and pg_trgm need such tests, and we do also have one use-case now in the core backend. The workaround of testing isalpha and isdigit separately seems quite inefficient, especially when dealing with multibyte characters; so let's fill in the missing support. Discussion: https://postgr.es/m/2548310.1664999615@sss.pgh.pa.us
* Fix comment in xlogprefetcher.cMichael Paquier2022-10-06
| | | | | Author: Sho Kato Discussion: https://postgr.es/m/TYCPR01MB684954052EC534A3261B29249F5C9@TYCPR01MB6849.jpnprd01.prod.outlook.com
* Refactor TAP test authentication/001_password.plMichael Paquier2022-10-06
| | | | | | | | | | | The test is changed to test for connection strings rather than specific roles, and the reset logic of pg_hba.conf is extended so as the database and user name entries can be directly specified. This is aimed at being used as a base for more test scenarios of pg_hba.conf and authentication paths. Author: Bertrand Drouvot, Michael Paquier Discussion: https://postgr.es/m/Yz0xO0emJ+mxtj2a@paquier.xyz
* Fix final compiler warning produced by -Wshadow=compatible-localDavid Rowley2022-10-06
| | | | | | | | We're now able to compile the entire tree with -Wshadow=compatible-local without any compiler warnings. Author: David Rowley Discussion: https://postgr.es/m/CAApHDvqWGMdB_pATeUqE=JCtNqNxObPOJ00jFEa2_sZ20j_Wvg@mail.gmail.com
* Add optional parameter to PG_TRY() macrosDavid Rowley2022-10-06
| | | | | | | | | | | | | | | | | | | This optional parameter can be specified in cases where there are nested PG_TRY() statements within a function in order to stop the compiler from issuing warnings about shadowed local variables when compiling with -Wshadow. The optional parameter is used as a suffix on the variable names declared within the PG_TRY(), PG_CATCH(), PG_FINALLY() and PG_END_TRY() macros. The parameter, if specified, must be the same in each component macro of the given PG_TRY() block. This also adjusts the single case where we have nested PG_TRY() statements to add a parameter to the inner-most PG_TRY(). This reduces the number of compiler warnings when compiling with -Wshadow=compatible-local from 5 down to 1. Author: David Rowley Discussion: https://postgr.es/m/CAApHDvqWGMdB_pATeUqE=JCtNqNxObPOJ00jFEa2_sZ20j_Wvg@mail.gmail.com
* tests: Restrict pg_locks queries in advisory_locks.sql to current databaseAndres Freund2022-10-05
| | | | | | | | Otherwise testing an existing installation can fail, if there are other locks, e.g. from one of the isolation tests. Reviewed-by: Michael Paquier <michael@paquier.xyz> Discussion: https://postgr.es/m/20221003234111.4ob7yph6r4g4ywhu@awork3.anarazel.de
* tests: Rename conflicting role namesAndres Freund2022-10-05
| | | | | | | These cause problems when running installcheck-world USE_MODULE_DB=1 with -j. Reviewed-by: Michael Paquier <michael@paquier.xyz> Discussion: https://postgr.es/m/20221003234111.4ob7yph6r4g4ywhu@awork3.anarazel.de
* meson: Add windows resource filesAndres Freund2022-10-05
| | | | | | | | | | | | | The generated resource files aren't exactly the same ones as the old buildsystems generate. Previously "InternalName" and "OriginalFileName" were mostly wrong / not set (despite being required), but that was hard to fix in at least the make build. Additionally, the meson build falls back to a "auto-generated" description when not set, and doesn't set it in a few cases - unlikely that anybody looks at these descriptions in detail. Author: Andres Freund <andres@anarazel.de> Author: Nazir Bilal Yavuz <byavuz81@gmail.com> Reviewed-by: Peter Eisentraut <peter.eisentraut@enterprisedb.com>
* meson: ecpg: Split definition of static and shared librariesAndres Freund2022-10-05
| | | | | | | | | | | | | | Required for correct resource file generation, as the resource files should only be added to the shared library. This also fixes a bunch of issues in the .pc files. Previously I tried to avoid building sources twice, once for the static and once for the shared libraries. We could still do so, but it's not clear that it's worth the complication. Reviewed-by: Peter Eisentraut <peter.eisentraut@enterprisedb.com> Discussion: https://postgr.es/m/20220927011951.j3h4o7n6bhf7dwau@awork3.anarazel.de
* meson: libpq: Revise static / shared library setupAndres Freund2022-10-05
| | | | | | | | | | | | | | | Improvements: - we don't need -DFRONTEND for libpq anymore since 1d77afefbd1 - the .pc file contents for a static libpq were wrong (referencing {pgport, common}_shlib) - incidentally fixes meson not supporting link_whole on AIX yet - added explanatory comments Previously I tried to avoid building libpq's sources twice, once for the static and once for the shared library. We could still do so, but it's not clear that it's worth the complication. Reviewed-by: Peter Eisentraut <peter.eisentraut@enterprisedb.com>
* Rename shadowed local variablesDavid Rowley2022-10-05
| | | | | | | | | | | | In a similar effort to f01592f91, here we mostly rename shadowed local variables to remove the warnings produced when compiling with -Wshadow=compatible-local. This fixes 63 warnings and leaves just 5. Author: Justin Pryzby, David Rowley Reviewed-by: Justin Pryzby Discussion https://postgr.es/m/20220817145434.GC26426%40telsasoft.com
* Remove definition of JUMBLE_SIZE from queryjumble.hMichael Paquier2022-10-05
| | | | | | | | | The same exists in queryjumble.c, and it is used only locally in this file so let's remove the definition in the header. Author: Tatsu Nakamori Reviewed-by: Tom Lane, Julien Rouhaud Discussion: https://postgr.es/m/bb4ebd0412da9b1ac87a5eb2a3646bf1@oss.nttdata.com
* Use macros from xlog_internal.h for WAL segment logic in pg_resetwalMichael Paquier2022-10-05
| | | | | | | | | | | | | | | | | | | | | When scanning for the end of WAL, pg_resetwal has been maintaining its own internal logic to calculate segment numbers and to parse a WAL segment name for its timeline and segment number. The code claimed for example that XLogFromFileName() cannot be used because it lacks the possibility of specifying a WAL segment size, which is not the case since fc49e24, that has made the WAL segment size configurable at initialization time, extending this routine to do so. Similarly, this switches one segment number calculation to use XLByteToSeg() rather than the same logic present in xlog_internal.h. While on it, switch to TimeLineID in pg_resetwal.c for TLI numbers parsed from segment names, to be more consistent with XLogFromFileName(). The maths are exactly the same, but the code gets simplified. Author: Bharath Rupireddy Reviewed-by: Kyotaro Horiguchi Discussion: https://postgr.es/m/CALj2ACX+E_jnwqH_jmjhNG8BczJTNRTOLpw8K1CB1OcB48MJ8w@mail.gmail.com
* Add a few new patterns to the tab completion of psqlMichael Paquier2022-10-05
| | | | | | | | | | | | This improves the tab completion of psql on a few points: - Provide a list of subscriptions on \dRs. - Provide a list of publications on \dRp. - Add CURRENT_ROLE, CURRENT_USER, SESSION_USER when OWNER TO is provided at the end of a query (as defined by RoleSpec in gram.y). Author: Vignesh C Reviewed-by: Dagfinn Ilmari Mannsåker Discussion: https://postgr.es/m/CALDaNm3toRBt6c6saY3174f7CsGztXRvVvfWbikkJEXY7x5WAA@mail.gmail.com
* Fix comment in guc_tables.cMichael Paquier2022-10-04
| | | | | | | s/ERROR_HANDLING/ERROR_HANDLING_OPTIONS/. Author: Peter Smith Discussion: https://postgr.es/m/CAHut+PtDj3CV+f0pVisc0XYMi2LHGBpQxQWtF0FjiSVN_nV17Q@mail.gmail.com
* Cleanup useless assignments and checksMichael Paquier2022-10-04
| | | | | | | | | | | | | | This cleans up a couple of areas: - Remove XLogSegNo calculation for the last WAL segment in backup in xlog.c (7d70809 has moved this logic entirely to xlogbackup.c when building the contents of the backup history file). - Remove check on log_min_duration in analyze.c, as it is already true where this code path is reached. - Simplify call to find_option() in guc.c. Author: Ranier Vilela Reviewed-by: Masahiko Sawada Discussion: https://postgr.es/m/CAEudQArCDQQiPiFR16=yu9k5s2tp4tgEe1U1ZbkW4ofx81AWWQ@mail.gmail.com
* Add filtering capability for cross-version pg_upgrade testsMichael Paquier2022-10-04
| | | | | | | | | | | | | | | | | | | | | | | | | This commit expands the TAP tests of pg_upgrade when running these with different major versions for the "old" cluster (to-be-upgraded) and the "new" cluster (upgraded-to), by backporting some of the buildfarm facilities directory into the script: - Remove comments from the dump files, avoiding version-dependent information. - Remove empty lines from the dump files. - Use --extra-float-digits=0 in the pg_dump command, when using an "old" cluster with version equal to or lower than v11. - Use --wal-segsize and --allow-group-access in initdb only when the "old" cluster is equal to or higher than v11. This allows the tests to pass down to v14 with the main regression test suite, while v9.5~13 still generate some diffs, but these are minimal compared to what happened before this commit. Much more could be done, especially around dump differences with function and procedures (these can also be avoided with direct manipulation of the dumps loaded, for example, in a way similar to the buildfarm), but at least the basics are in place now. Reviewed-by: Justin Pryzby, Anton A. Melnikov Discussion: https://postgr.es/m/Yox1ME99GhAemMq1@paquier.xyz
* meson: llvm: Use llvm-config's --cxxflags when building llvmjitAndres Freund2022-10-03
| | | | | | | Otherwise we don't use LLVM's flags when building llvmjit_wrap.cpp and llvmjit_inline.cpp. That can cause compile time failures if the C++ compiler doesn't default to a new enough C++ standards version and link time failures due to ABI influencing flags like -fno-rtti.
* Fix psql's behavior with \g for a multiple-command string.Tom Lane2022-10-03
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The pre-v15 behavior was to discard all but the last result, but with the new behavior of printing all results by default, we will send each such result to the \g file. However, we're still opening and closing the \g file for each result, so you lose all but the last result anyway. Move the output-file state up to ExecQueryAndProcessResults so that we open/close the \g file only once per command string. To support this without changing other behavior, we must adjust PrintQueryResult to have separate FILE * arguments for query and status output (since status output has never gone to the \g file). That in turn makes it a good idea to push the responsibility for fflush'ing output down to PrintQueryTuples and PrintQueryStatus. Also fix an infinite loop if COPY IN/OUT is attempted in \watch. We used to reject that, but that error exit path got broken somewhere along the line in v15. There seems no real reason to reject it anyway as the code now stands, so just remove the error exit and make sure that COPY OUT data goes to the right place. Also remove PrintQueryResult's unused is_watch parameter, and make some other cosmetic cleanups (adjust obsolete comments, break some overly-long lines). Daniel Vérité and Tom Lane Discussion: https://postgr.es/m/4333844c-2244-4d6e-a49a-1d483fbe304f@manitou-mail.org
* Revert "Optimize order of GROUP BY keys".Tom Lane2022-10-03
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This reverts commit db0d67db2401eb6238ccc04c6407a4fd4f985832 and several follow-on fixes. The idea of making a cost-based choice of the order of the sorting columns is not fundamentally unsound, but it requires cost information and data statistics that we don't really have. For example, relying on procost to distinguish the relative costs of different sort comparators is pretty pointless so long as most such comparator functions are labeled with cost 1.0. Moreover, estimating the number of comparisons done by Quicksort requires more than just an estimate of the number of distinct values in the input: you also need some idea of the sizes of the larger groups, if you want an estimate that's good to better than a factor of three or so. That's data that's often unknown or not very reliable. Worse, to arrive at estimates of the number of calls made to the lower-order-column comparison functions, the code needs to make estimates of the numbers of distinct values of multiple columns, which are necessarily even less trustworthy than per-column stats. Even if all the inputs are perfectly reliable, the cost algorithm as-implemented cannot offer useful information about how to order sorting columns beyond the point at which the average group size is estimated to drop to 1. Close inspection of the code added by db0d67db2 shows that there are also multiple small bugs. These could have been fixed, but there's not much point if we don't trust the estimates to be accurate in-principle. Finally, the changes in cost_sort's behavior made for very large changes (often a factor of 2 or so) in the cost estimates for all sorting operations, not only those for multi-column GROUP BY. That naturally changes plan choices in many situations, and there's precious little evidence to show that the changes are for the better. Given the above doubts about whether the new estimates are really trustworthy, it's hard to summon much confidence that these changes are better on the average. Since we're hard up against the release deadline for v15, let's revert these changes for now. We can always try again later. Note: in v15, I left T_PathKeyInfo in place in nodes.h even though it's unreferenced. Removing it would be an ABI break, and it seems a bit late in the release cycle for that. Discussion: https://postgr.es/m/TYAPR01MB586665EB5FB2C3807E893941F5579@TYAPR01MB5866.jpnprd01.prod.outlook.com
* Add authentication TAP test for peer authenticationMichael Paquier2022-10-03
| | | | | | | | | | | This commit introduces an authentication test for the peer method, as of a set of scenarios with and without a user name map. The script is automatically skipped if peer is not supported in the environment where this test is run, checking this behavior by attempting a connection first on a cluster up and running. Author: Bertrand Drouvot Discussion: https://postgr.es/m/aa60994b-1c66-ca7a-dab9-9a200dbac3d2@amazon.com
* Fix tiny memory leaksPeter Eisentraut2022-10-01
| | | | | | | | | | | | | Both check_application_name() and check_cluster_name() use pg_clean_ascii() but didn't release the memory. Depending on when the GUC is set, this might be cleaned up at some later time or it would leak postmaster memory once. In any case, it seems better not to have to rely on such analysis and make the code locally robust. Also, this makes Valgrind happier. Author: Masahiko Sawada <sawada.mshk@gmail.com> Reviewed-by: Jacob Champion <jchampion@timescale.com> Discussion: https://www.postgresql.org/message-id/CAD21AoBmFNy9MPfA0UUbMubQqH3AaK5U3mrv6pSeWrwCk3LJ8g@mail.gmail.com
* doc: Fix some grammar and typosMichael Paquier2022-10-01
| | | | | | | | This fixes some areas related to logical replication and custom RMGRs. Author: Ekaterina Kiryanova Discussion: https://postgr.es/m/fa4773f1-1396-384a-bcd7-85b5e013f399@postgrespro.ru Backpatch-through: 15
* Avoid improbable PANIC during heap_update, redux.Tom Lane2022-09-30
| | | | | | | | | | | | | | | | | | | | Commit 34f581c39 intended to ensure that RelationGetBufferForTuple would acquire a visibility-map page pin in case the otherBuffer's all-visible bit had become set since we last had lock on that page. But I missed a case: when we're extending the relation, VM concerns were dealt with only in the relatively-less-likely case that we fail to conditionally lock the otherBuffer. I think I'd believed that we couldn't need to worry about it if the conditional lock succeeds, which is true for the target buffer; but the otherBuffer was unlocked for awhile so its bit might be set anyway. So we need to do the GetVisibilityMapPins dance, and then also recheck the page's free space, in both cases. Per report from Jaime Casanova. Back-patch to v12 as the previous patch was (although there's still no evidence that the bug is reachable pre-v14). Discussion: https://postgr.es/m/E1lWLjP-00006Y-Ml@gemulon.postgresql.org
* mingw: Define PGDLLEXPORT as __declspec (dllexport) as done for msvcAndres Freund2022-09-30
| | | | | | | | | | | | | While mingw would otherwise fall back to __attribute__((visibility("default"))), that appears to only work as long as no symbols are declared with __declspec(dllexport). But we can end up with some, e.g. plpython's Py_Init. It's quite possible we should do the same for cygwin, but I don't have a test environment for that... Discussion: http://postgr.es/m/20220928022724.erzuk5v4ai4b53do@awork3.anarazel.de Discussion: http://postgr.es/m/20220928025242.ugf7t5ugxxgmkraa@awork3.anarazel.de
* Adjust PQsslAttributeNames() to match PQsslAttribute().Tom Lane2022-09-30
| | | | | | | | | | | | | | | | Currently, PQsslAttributeNames() returns the same list of attribute names regardless of its conn parameter. This patch changes it to have behavior parallel to what 80a05679d installed for PQsslAttribute: you get OpenSSL's attributes if conn is NULL or is an SSL-encrypted connection, or an empty list if conn is a non-encrypted connection. The point of this is to have sensible connection-dependent behavior in case we ever support multiple SSL libraries. The behavior for NULL can be defined as "the attributes for the default SSL library", parallel to what PQsslAttribute(NULL, "library") does. Since this is mostly just future-proofing, no back-patch. Discussion: https://postgr.es/m/17625-fc47c78b7d71b534@postgresql.org
* Fix tab-completion after commit 790bf615ddbaAlvaro Herrera2022-09-30
| | | | | | | | | | | | | | | | | I (Álvaro) broke tab-completion for GRANT .. ALL TABLES IN SCHEMA while removing ALL from the publication syntax for schemas in the aforementioned commit. I also missed to update a bunch of tab-completion rules for ALTER/CREATE PUBLICATION that match each individual piece of ALL TABLES IN SCHEMA. Repair those bugs. While fixing up that commit, update a couple of outdated comments related to the same change. Backpatch to 15. Author: Shi yu <shiy.fnst@fujitsu.com> Reviewed-by: Peter Smith <smithpb2250@gmail.com> Discussion: https://postgr.es/m/OSZPR01MB6310FCE8609185A56344EED2FD559@OSZPR01MB6310.jpnprd01.prod.outlook.com
* Remove useless argument from UnpinBuffer()Michael Paquier2022-09-30
| | | | | | | | | | | The last caller of UnpinBuffer() that did not want to adjust CurrentResourceOwner was removed in 2d115e4, and nothing has been introduced in bufmgr.c to do the same thing since. This simplifies 10 code paths. Author: Aleksander Alekseev Reviewed-by: Nathan Bossart, Zhang Mingli, Bharath Rupireddy Discussion: https://postgr.es/m/CAJ7c6TOmmFpb6ohurLhTC7hKNJWGzdwf8s4EAtAZxD48g-e6Jw@mail.gmail.com
* Fix bogus behavior of PQsslAttribute(conn, "library").Tom Lane2022-09-29
| | | | | | | | | | | | Commit ebc8b7d44 intended to change the behavior of PQsslAttribute(NULL, "library"), but accidentally also changed what happens with a non-NULL conn pointer. Undo that so that only the intended behavior change happens. Clarify some associated documentation. Per bug #17625 from Heath Lord. Back-patch to v15. Discussion: https://postgr.es/m/17625-fc47c78b7d71b534@postgresql.org
* Improve wording of log messages triggered by max_slot_wal_keep_size.Tom Lane2022-09-29
| | | | | | | | | | | | | | | | | | | The one about "terminating process to release replication slot" told you nothing about why that was happening. The one about "invalidating slot because its restart_lsn exceeds max_slot_wal_keep_size" told you what was happening, but violated our message style guideline about keeping the primary message short. Add DETAIL/HINT lines to carry the appropriate detail and make the two cases more uniform. While here, fix bogus test logic in 019_replslot_limit.pl: if it timed out without seeing the expected log message, no test failure would be reported. This is flat broken since commit 549ec201d removed the test counts; even before that it was horribly bad style, since you'd only get told that not all tests had been run. Kyotaro Horiguchi, reviewed by Bertrand Drouvot; test fixes by me Discussion: https://postgr.es/m/20211214.130456.2233153190058148084.horikyota.ntt@gmail.com
* Use actual backend IDs in pg_stat_get_backend_idset() and friends.Tom Lane2022-09-29
| | | | | | | | | | | | | | | | | | | | | | | Up to now, the ID values returned by pg_stat_get_backend_idset() and used by pg_stat_get_backend_activity() and allied functions were just indexes into a local array of sessions seen by the last stats refresh. This is problematic for a few reasons. The "ID" of a session can vary over its existence, which is surprising. Also, while these numbers often match the "backend ID" used for purposes like temp schema assignment, that isn't reliably true. We can fairly cheaply switch things around to make these numbers actually be the sessions' backend IDs. The added test case illustrates that with this definition, the temp schema used by a given session can be obtained given its PID. While here, delete some dead code that guarded against getting a NULL return from pgstat_fetch_stat_local_beentry(). That can't happen as long as the caller is careful to pass an in-range array index, as all the callers are. (This code may not have been dead when written, but it surely is now.) Nathan Bossart Discussion: https://postgr.es/m/20220815205811.GA250990@nathanxps13
* Update comment in ExecInsert() regarding batch insertion.Etsuro Fujita2022-09-29
| | | | | | | | | | | | Remove the stale text that is a leftover from an earlier version of the patch to add support for batch insertion, and adjust the wording in the remaining text. Back-patch to v14 where batch insertion came in. Review and wording adjustment by Tom Lane. Discussion: https://postgr.es/m/CAPmGK14goatHPHQv2Aeu_UTKqZ%2BBO%2BP%2Bzd3HKv5D%2BdyyfWKDSw%40mail.gmail.com
* Introduce SYSTEM_USERMichael Paquier2022-09-29
| | | | | | | | | | | | | | | | | | | | | | | | SYSTEM_USER is a reserved keyword of the SQL specification that, roughly described, is aimed at reporting some information about the system user who has connected to the database server. It may include implementation-specific information about the means by the user connected, like an authentication method. This commit implements SYSTEM_USER as of auth_method:identity, where "auth_method" is a keyword about the authentication method used to log into the server (like peer, md5, scram-sha-256, gss, etc.) and "identity" is the authentication identity as introduced by 9afffcb (peer sets authn to the OS user name, gss to the user principal, etc.). This format has been suggested by Tom Lane. Note that thanks to d951052, SYSTEM_USER is available to parallel workers. Bump catalog version. Author: Bertrand Drouvot Reviewed-by: Jacob Champion, Joe Conway, Álvaro Herrera, Michael Paquier Discussion: https://postgr.es/m/7e692b8c-0b11-45db-1cad-3afc5b57409f@amazon.com
* Mark sigint_interrupt_enabled as sig_atomic_tMichael Paquier2022-09-29
| | | | | | | | | | | This is a continuation of 78fdb1e, where this flag is set in the psql callback handler used for SIGINT. This was previously a boolean but the C standard recommends the use of sig_atomic_t. Note that this influences PromptInterruptContext in string.h, where the same flag is tracked. Author: Hayato Kuroda Discussion: https://postgr.es/m/TYAPR01MB58669A9EC96AA3078C2CD938F5549@TYAPR01MB5866.jpnprd01.prod.outlook.com
* windows: Set UMDF_USING_NTSTATUS globally, include ntstatus.hAndres Freund2022-09-28
| | | | | | | | | | | | | | | | We'd like to use precompiled headers on windows to reduce compile times. Right now we rely on defining UMDF_USING_NTSTATUS before including postgres.h in a few select places - which doesn't work with precompiled headers. Instead define it globally. When UMDF_USING_NTSTATUS is defined we need to explicitly include ntstatus.h, winternl.h to get a comparable set of symbols. Right now these includes would be required in a number of non-platform-specific .c files - to avoid that, include them in win32_port.h. Based on my measurements that doesn't increase compile times measurably. Reviewed-by: Thomas Munro <thomas.munro@gmail.com> Discussion: https://postgr.es/m/20220927011951.j3h4o7n6bhf7dwau@awork3.anarazel.de
* meson: pg_regress: Define a HOST_TUPLE sufficient to make resultmap workAndres Freund2022-09-28
| | | | | | | | | | | | | This doesn't end up with a triple that's exactly the same as config.guess - it'd be hard to achieve that and it doesn't seem required. We can't rely on config.guess as we don't necessarily have a /bin/sh on windows, e.g., when building on windows with msvc. This isn't perfect, e.g., clang works on windows as well. But I suspect we'd need a bunch of other changes to make clang on windows work, and we haven't supported it historically. Discussion: http://postgr.es/m/20220928022724.erzuk5v4ai4b53do@awork3.anarazel.de
* Map ERROR_INVALID_NAME to ENOENT in mapping table of win32error.cMichael Paquier2022-09-29
| | | | | | | | | This error can be reached when sending an incorrect file name to open() on Windows, resulting in a confusing errno reported. This has been seen in the development of a different patch by the same author. Author: Bharath Rupireddy Discussion: https://postgr.es/m/CALj2ACWet-b8Juba0DiXwfGCyyOcohzwksahE5ebB9rcbLZKCQ@mail.gmail.com
* Restore pg_pread and friends.Thomas Munro2022-09-29
| | | | | | | | | | | | | | | | | | | | Commits cf112c12 and a0dc8271 were a little too hasty in getting rid of the pg_ prefixes where we use pread(), pwrite() and vectored variants. We dropped support for ancient Unixes where we needed to use lseek() to implement replacements for those, but it turns out that Windows also changes the current position even when you pass in an offset to ReadFile() and WriteFile() if the file handle is synchronous, despite its documentation saying otherwise. Switching to asynchronous file handles would fix that, but have other complications. For now let's just put back the pg_ prefix and add some comments to highlight the non-standard side-effect, which we can now describe as Windows-only. Reported-by: Bharath Rupireddy <bharath.rupireddyforpostgres@gmail.com> Reviewed-by: Bharath Rupireddy <bharath.rupireddyforpostgres@gmail.com> Discussion: https://postgr.es/m/20220923202439.GA1156054%40nathanxps13
* Restrict Datum sort optimization to byval types onlyDavid Rowley2022-09-29
| | | | | | | | | | | | | | | | | | | 91e9e89dc modified nodeSort.c so that it used datum sorts when the targetlist of the outer node contained only a single column. That commit failed to recognise that the Datum returned by tuplesort_getdatum() must be pfree'd when the type is a byref type. Ronan Dunklau did originally propose the patch with that restriction, but that, probably through my own fault, got lost during further development work. Due to the timing of this report (PG15 RC1 is almost out the door), let's just restrict the datum sort optimization to apply for byval types only. We might want to look harder into making this work for byref types in PG16. Reported-by: Önder Kalacı Diagnosis-by: Tom Lane Discussion: https://postgr.es/m/CACawEhVxe0ufR26UcqtU7GYGRuubq3p6ZWPGXL4cxy_uexpAAQ@mail.gmail.com Backpatch-through: 15, where 91e9e89dc was introduced.
* Allow callback functions to deregister themselves during a call.Tom Lane2022-09-28
| | | | | | | | | | | | | | | | | | Fetch the next-item pointer before the call not after, so that we aren't dereferencing a dangling pointer if the callback deregistered itself during the call. The risky coding pattern appears in CallXactCallbacks, CallSubXactCallbacks, and ResourceOwnerReleaseInternal. (There are some other places that might be at hazard if they offered deregistration functionality, but they don't.) I (tgl) considered back-patching this, but desisted because it wouldn't be very safe for extensions to rely on this working in pre-v16 branches. Hao Wu Discussion: https://postgr.es/m/CAH+9SWXTiERkmhRke+QCcc+jRH8d5fFHTxh8ZK0-Yn4BSpyaAg@mail.gmail.com
* Change some errdetail() to errdetail_internal()Alvaro Herrera2022-09-28
| | | | | | | | | | | | This prevents marking the argument string for translation for gettext, and it also prevents the given string (which is already translated) from being translated at runtime. Also, mark the strings used as arguments to check_rolespec_name for translation. Backpatch all the way back as appropriate. None of this is caught by any tests (necessarily so), so I verified it manually.
* Fix bug in DROP OWNED BY.Robert Haas2022-09-28
| | | | | | | | | | Commit 6566133c5f52771198aca07ed18f84519fac1be7 broke the case where the role passed to DROP OWNED BY owns a database. Report by Rushabh Lathia, who also provided a patch, but this patch takes a slightly different approach to fixing the problem. Discussion: http://postgr.es/m/CAGPqQf2vO+nbo=3yAdZ8v26Rbug7bY4YjPaPLZx=L1NZ9-CC3w@mail.gmail.com
* Revert 56-bit relfilenode change and follow-up commits.Robert Haas2022-09-28
| | | | | | | | There are still some alignment-related failures in the buildfarm, which might or might not be able to be fixed quickly, but I've also just realized that it increased the size of many WAL records by 4 bytes because a block reference contains a RelFileLocator. The effect of that hasn't been studied or discussed, so revert for now.
* Fix InitializeRelfilenumberMap for 05d4cbf9b6ba708858984b01ca0fc56d59d4ec7cRobert Haas2022-09-28
| | | | | | | | | | | Since relfilenodes are now 56-bits, we use bigint as the SQL type to represent them, which means F_INT8EQ must be used here rather than F_OIDEQ. On 64-bit machines this doesn't matter, but 32-bit machines are unhappy. Dilip Kumar Discussion: http://postgr.es/m/CAFiTN-t71ciSckMzixAhrF9py7oRO6xszKi4mTRwjuucXr5tpw@mail.gmail.com
* Fix alignment problems with SharedInvalSmgrMsg.Robert Haas2022-09-28
| | | | | | | | | | | | SharedInvalSmgrMsg can't require 8-byte alignment, because then SharedInvalidationMessage will require 8-byte alignment, which will then cause ParseCommitRecord to fail on machines that are picky about alignment, because it assumes that everything that gets packed into a commit record requires only 4-byte alignment. Another problem with 05d4cbf9b6ba708858984b01ca0fc56d59d4ec7c. Discussion: http://postgr.es/m/3825454.1664310917@sss.pgh.pa.us
* Remove publicationcmds.c's expr_allowed_in_node as a functionAlvaro Herrera2022-09-28
| | | | | | | | Its API is quite strange, and since there's only one caller, there's no reason for it to be a separate function in the first place. Inline it instead. Discussion: https://postgr.es/m/20220927124249.4zdzzlz6had7k3x2@alvherre.pgsql
* Fix some comments of do_pg_backup_start() and do_pg_backup_stop()Michael Paquier2022-09-28
| | | | | | | | | | Both functions referred to an incorrect variable name, so make the whole more consistent. Oversight in 7d70809. Author: Kyotaro Horiguchi, Bharath Rupireddy Discussion: https://postgr.es/m/20220927.172427.467118514018439476.horikyota.ntt@gmail.com
* In BufTagGetForkNum, cast to the correct type.Robert Haas2022-09-27
| | | | | | | | Another defect in 05d4cbf9b6ba708858984b01ca0fc56d59d4ec7c. Per CI, via Justin Pryzby. Discussion: http://postgr.es/m/20220927200712.GH6256@telsasoft.com
* Fix typos in commit 05d4cbf9b6ba708858984b01ca0fc56d59d4ec7c.Robert Haas2022-09-27
| | | | | | Reported by Justin Pryzby. Discussion: http://postgr.es/m/20220927185121.GE6256@telsasoft.com
generated by cgit v1.2.3 (git 2.25.1) at 2025-05-31 08:28:02 +0000