aboutsummaryrefslogtreecommitdiff
path: root/src/backend
Commit message (Collapse)AuthorAge
...
* Account explicitly for long-lived FDs that are allocated outside fd.c.Tom Lane2020-02-24
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The comments in fd.c have long claimed that all file allocations should go through that module, but in reality that's not always practical. fd.c doesn't supply APIs for invoking some FD-producing syscalls like pipe() or epoll_create(); and the APIs it does supply for non-virtual FDs are mostly insistent on releasing those FDs at transaction end; and in some cases the actual open() call is in code that can't be made to use fd.c, such as libpq. This has led to a situation where, in a modern server, there are likely to be seven or so long-lived FDs per backend process that are not known to fd.c. Since NUM_RESERVED_FDS is only 10, that meant we had *very* few spare FDs if max_files_per_process is >= the system ulimit and fd.c had opened all the files it thought it safely could. The contrib/postgres_fdw regression test, in particular, could easily be made to fall over by running it under a restrictive ulimit. To improve matters, invent functions Acquire/Reserve/ReleaseExternalFD that allow outside callers to tell fd.c that they have or want to allocate a FD that's not directly managed by fd.c. Add calls to track all the fixed FDs in a standard backend session, so that we are honestly guaranteeing that NUM_RESERVED_FDS FDs remain unused below the EMFILE limit in a backend's idle state. The coding rules for these functions say that there's no need to call them in code that just allocates one FD over a fairly short interval; we can dip into NUM_RESERVED_FDS for such cases. That means that there aren't all that many places where we need to worry. But postgres_fdw and dblink must use this facility to account for long-lived FDs consumed by libpq connections. There may be other places where it's worth doing such accounting, too, but this seems like enough to solve the immediate problem. Internally to fd.c, "external" FDs are limited to max_safe_fds/3 FDs. (Callers can choose to ignore this limit, but of course it's unwise to do so except for fixed file allocations.) I also reduced the limit on "allocated" files to max_safe_fds/3 FDs (it had been max_safe_fds/2). Conceivably a smarter rule could be used here --- but in practice, on reasonable systems, max_safe_fds should be large enough that this isn't much of an issue, so KISS for now. To avoid possible regression in the number of external or allocated files that can be opened, increase FD_MINFREE and the lower limit on max_files_per_process a little bit; we now insist that the effective "ulimit -n" be at least 64. This seems like pretty clearly a bug fix, but in view of the lack of field complaints, I'll refrain from risking a back-patch. Discussion: https://postgr.es/m/E1izCmM-0005pV-Co@gemulon.postgresql.org
* Adapt hashfn.c and hashutils.h for frontend use.Robert Haas2020-02-24
| | | | | | | | | | | | | | | | | | | | | | | | | hash_any() and its various variants are defined to return Datum, which is a backend-only concept, but the underlying functions actually want to return uint32 and uint64, and only return Datum because it's convenient for callers who are using them to implement a hash function for some SQL datatype. However, changing these functions to return uint32 and uint64 seems like it might lead to programming errors or back-patching difficulties, both because they are widely used and because failure to use UInt{32,64}GetDatum() might not provoke a compilation error. Instead, rename the existing functions as well as changing the return type, and add static inline wrappers for those callers that need the previous behavior. Although this commit adapts hashutils.h and hashfn.c so that they can be compiled as frontend code, it does not actually do anything that would cause them to be so compiled. That is left for another commit. Patch by me, reviewed by Suraj Kharage and Mark Dilger. Discussion: http://postgr.es/m/CA+TgmoaRiG4TXND8QuM6JXFRkM_1wL2ZNhzaUKsuec9-4yrkgw@mail.gmail.com
* Put all the prototypes for hashfn.c into the same header file.Robert Haas2020-02-24
| | | | | | | | | | Previously, some of the prototypes for functions in hashfn.c were in utils/hashutils.h and others were in utils/hsearch.h, but that is confusing and has no particular benefit. Patch by me, reviewed by Suraj Kharage and Mark Dilger. Discussion: http://postgr.es/m/CA+TgmoaRiG4TXND8QuM6JXFRkM_1wL2ZNhzaUKsuec9-4yrkgw@mail.gmail.com
* Move bitmap_hash and bitmap_match to bitmapset.c.Robert Haas2020-02-24
| | | | | | | | | | | The closely-related function bms_hash_value is already defined in that file, and this change means that hashfn.c no longer needs to depend on nodes/bitmapset.h. That gets us closer to allowing use of the hash functions in hashfn.c in frontend code. Patch by me, reviewed by Suraj Kharage and Mark Dilger. Discussion: http://postgr.es/m/CA+TgmoaRiG4TXND8QuM6JXFRkM_1wL2ZNhzaUKsuec9-4yrkgw@mail.gmail.com
* Add prefix checks in exclude lists for pg_rewind, pg_checksums and base backupsMichael Paquier2020-02-24
| | | | | | | | | | | | | | | | | | An instance of PostgreSQL crashing with a bad timing could leave behind temporary pg_internal.init files, potentially causing failures when verifying checksums. As the same exclusion lists are used between pg_rewind, pg_checksums and basebackup.c, all those tools are extended with prefix checks to keep everything in sync, with dedicated checks added for pg_internal.init. Backpatch down to 11, where pg_checksums (pg_verify_checksums in 11) and checksum verification for base backups have been introduced. Reported-by: Michael Banck Author: Michael Paquier Reviewed-by: Kyotaro Horiguchi, David Steele Discussion: https://postgr.es/m/62031974fd8e941dd8351fbc8c7eff60d59c5338.camel@credativ.de Backpatch-through: 11
* Factor out InitControlFile() from BootStrapXLOG()Peter Eisentraut2020-02-22
| | | | | | | Right now this only makes BootStrapXLOG() a bit more manageable, but in the future there may be external callers. Discussion: https://www.postgresql.org/message-id/e8f86ba5-48f1-a80a-7f1d-b76bcb9c5c47@2ndquadrant.com
* Reformat code commentPeter Eisentraut2020-02-22
| | | | Discussion: https://www.postgresql.org/message-id/e8f86ba5-48f1-a80a-7f1d-b76bcb9c5c47@2ndquadrant.com
* Assume that we have <wchar.h>.Tom Lane2020-02-21
| | | | | | | | | | | | | Windows has this, and so do all other live platforms according to the buildfarm; it's been required by POSIX since SUSv2. So remove the configure probe and tests of HAVE_WCHAR_H. This is part of a series of commits to get rid of no-longer-relevant configure checks and dead src/port/ code. I'm committing them separately to make it easier to back out individual changes if they prove less portable than I expect. Discussion: https://postgr.es/m/15379.1582221614@sss.pgh.pa.us
* Assume that we have utime() and <utime.h>.Tom Lane2020-02-21
| | | | | | | | | | | | | | | | | | These are required by POSIX since SUSv2, and no live platforms fail to provide them. On Windows, utime() exists and we bring our own <utime.h>, so we're good there too. So remove the configure probes and ad-hoc substitute code. We don't need to check for utimes() anymore either, since that was only used as a substitute. In passing, make the Windows build include <sys/utime.h> only where we need it, not everywhere. This is part of a series of commits to get rid of no-longer-relevant configure checks and dead src/port/ code. I'm committing them separately to make it easier to back out individual changes if they prove less portable than I expect. Discussion: https://postgr.es/m/15379.1582221614@sss.pgh.pa.us
* Assume that we have cbrt().Tom Lane2020-02-21
| | | | | | | | | | | | Windows has this, and so do all other live platforms according to the buildfarm, so remove the configure probe and float.c's substitute code. This is part of a series of commits to get rid of no-longer-relevant configure checks and dead src/port/ code. I'm committing them separately to make it easier to back out individual changes if they prove less portable than I expect. Discussion: https://postgr.es/m/15379.1582221614@sss.pgh.pa.us
* Fixup for nodeAgg.c refactor.Jeff Davis2020-02-21
| | | | Commit 5b618e1f made an unintended behavior change.
* Avoid redundant checks in partition_bounds_copy().Etsuro Fujita2020-02-21
| | | | | | | | | | | | Previously, partition_bounds_copy() checked whether the strategy for the given partition bounds was hash or not, and then determined the number of elements in the datums in the datums array for the partition bounds, on each iteration of the loop for copying the datums array, but there is no need to do that. Perform the checks only once before the loop iteration. Author: Etsuro Fujita Reported-by: Amit Langote and Julien Rouhaud Discussion: https://postgr.es/m/CAPmGK14Rvxrm8DHWvCjdoks6nwZuHBPvMnWZ6rkEx2KhFeEoPQ@mail.gmail.com
* Remove extra word from comment.Etsuro Fujita2020-02-20
|
* Remove support for upgrading extensions from "unpackaged" state.Tom Lane2020-02-19
| | | | | | | | | | | | | | | | | | | | | | | | | Andres Freund pointed out that allowing non-superusers to run "CREATE EXTENSION ... FROM unpackaged" has security risks, since the unpackaged-to-1.0 scripts don't try to verify that the existing objects they're modifying are what they expect. Just attaching such objects to an extension doesn't seem too dangerous, but some of them do more than that. We could have resolved this, perhaps, by still requiring superuser privilege to use the FROM option. However, it's fair to ask just what we're accomplishing by continuing to lug the unpackaged-to-1.0 scripts forward. None of them have received any real testing since 9.1 days, so they may not even work anymore (even assuming that one could still load the previous "loose" object definitions into a v13 database). And an installation that's trying to go from pre-9.1 to v13 or later in one jump is going to have worse compatibility problems than whether there's a trivial way to convert their contrib modules into extension style. Hence, let's just drop both those scripts and the core-code support for "CREATE EXTENSION ... FROM". Discussion: https://postgr.es/m/20200213233015.r6rnubcvl4egdh5r@alap3.anarazel.de
* Minor refactor of nodeAgg.c.Jeff Davis2020-02-19
| | | | | | | | | | | | * Separate calculation of hash value from the lookup. * Split build_hash_table() into two functions. * Change lookup_hash_entry() to return AggStatePerGroup. That's all the caller needed, anyway. These changes are to support the upcoming Disk-based Hash Aggregation work. Discussion: https://postgr.es/m/31f5ab871a3ad5a1a91a7a797651f20e77ac7ce3.camel%40j-davis.com
* logtape.c: allocate read buffer even for an empty tape.Jeff Davis2020-02-19
| | | | | | | | | | | | | | | | | Prior to this commit, the read buffer was allocated at the time the tape was rewound; but as an optimization, would not be allocated at all if the tape was empty. That optimization meant that it was valid to have a rewound tape with the buffer set to NULL, but only if a number of conditions were met and only if the API was used properly. After 7fdd919a refactored the code to support lazily-allocating the buffer, Coverity started complaining. The optimization for empty tapes doesn't seem important, so just allocate the buffer whether the tape has any data or not. Discussion: https://postgr.es/m/20351.1581868306%40sss.pgh.pa.us
* Fix mesurement of elapsed time during truncating heap in VACUUM.Fujii Masao2020-02-19
| | | | | | | | | | | | | | | | | | | | | | | | | | VACUUM may truncate heap in several batches. The activity report is logged for each batch, and contains the number of pages in the table before and after the truncation, and also the elapsed time during the truncation. Previously the elapsed time reported in each batch was the total elapsed time since starting the truncation until finishing each batch. For example, if the truncation was processed dividing into three batches, the second batch reported the accumulated time elapsed during both first and second batches. This is strange and confusing because the number of pages in the table reported together is not total. Instead, each batch should report the time elapsed during only that batch. The cause of this issue was that the resource usage snapshot was initialized only at the beginning of the truncation and was never reset later. This commit fixes the issue by changing VACUUM so that the resource usage snapshot is reset at each batch. Back-patch to all supported branches. Reported-by: Tatsuhito Kasahara Author: Tatsuhito Kasahara Reviewed-by: Masahiko Sawada, Fujii Masao Discussion: https://postgr.es/m/CAP0=ZVJsf=NvQuy+QXQZ7B=ZVLoDV_JzsVC1FRsF1G18i3zMGg@mail.gmail.com
* Clean up some code, comments and docs referring to Windows 2000 and olderMichael Paquier2020-02-19
| | | | | | | | | | | | This fixes and updates a couple of comments related to outdated Windows versions. Particularly, src/common/exec.c had a fallback implementation to read a file's line from a pipe because stdin/stdout/stderr does not exist in Windows 2000 that is removed to simplify src/common/ as there are unlikely versions of Postgres running on such platforms. Author: Michael Paquier Reviewed-by: Kyotaro Horiguchi, Juan José Santamaría Flecha Discussion: https://postgr.es/m/20191219021526.GC4202@paquier.xyz
* Stop demanding that top xact must be seen before subxact in decoding.Amit Kapila2020-02-19
| | | | | | | | | | | | | | | | | | | | | | | | | | | Manifested as ERROR: subtransaction logged without previous top-level txn record this check forbids legit behaviours like - First xl_xact_assignment record is beyond reading, i.e. earlier restart_lsn. - After restart_lsn there is some change of a subxact. - After that, there is second xl_xact_assignment (for another subxact) revealing the relationship between top and first subxact. Such a transaction won't be streamed anyway because we hadn't seen it in full. Saying for sure whether xact of some record encountered after the snapshot was deserialized can be streamed or not requires to know whether it wrote something before deserialization point --if yes, it hasn't been seen in full and can't be decoded. Snapshot doesn't have such info, so there is no easy way to relax the check. Reported-by: Hsu, John Diagnosed-by: Arseny Sher Author: Arseny Sher, Amit Kapila Reviewed-by: Amit Kapila, Dilip Kumar Backpatch-through: 9.5 Discussion: https://postgr.es/m/AB5978B2-1772-4FEE-A245-74C91704ECB0@amazon.com
* Remove obsolete _bt_compare() comment.Peter Geoghegan2020-02-18
| | | | | | btbuild() has nothing to say about how NULL values compare in nbtree. Besides, there are _bt_compare() header comments that describe how NULL values are handled.
* Make inherited LOCK TABLE perform access permission checks on parent table only.Fujii Masao2020-02-18
| | | | | | | | | | | | | | | | | | Previously, LOCK TABLE command through a parent table checked the permissions on not only the parent table but also the children tables inherited from it. This was a bug and inherited queries should perform access permission checks on the parent table only. This commit fixes LOCK TABLE so that it does not check the permission on children tables. This patch is applied only in the master branch. We decided not to back-patch because it's not hard to imagine that there are some applications expecting the old behavior and the change breaks their security. Author: Amit Langote Reviewed-by: Fujii Masao Discussion: https://postgr.es/m/CAHGQGwE+GauyG7POtRfRwwthAGwTjPQYdFHR97+LzA4RHGnJxA@mail.gmail.com
* Remove duplicated words in commentsMichael Paquier2020-02-18
| | | | | | Author: Daniel Gustafsson Reviewed-by: Vik Fearing Discussion: https://postgr.es/m/EBC3BFEB-664C-4063-81ED-29F1227DB012@yesql.se
* Optimize update of tables with generated columnsPeter Eisentraut2020-02-17
| | | | | | | | | | | When updating a table row with generated columns, only recompute those generated columns whose base columns have changed in this update and keep the rest unchanged. This can result in a significant performance benefit. The required information was already kept in RangeTblEntry.extraUpdatedCols; we just have to make use of it. Reviewed-by: Pavel Stehule <pavel.stehule@gmail.com> Discussion: https://www.postgresql.org/message-id/flat/b05e781a-fa16-6b52-6738-761181204567@2ndquadrant.com
* Fill in extraUpdatedCols in logical replicationPeter Eisentraut2020-02-17
| | | | | | | | | | | | | | The extraUpdatedCols field of the target RTE records which generated columns are affected by an update. This is used in a variety of places, including per-column triggers and foreign data wrappers. When an update was initiated by a logical replication subscription, this field was not filled in, so such an update would not affect generated columns in a way that is consistent with normal updates. To fix, factor out some code from analyze.c to fill in extraUpdatedCols in the logical replication worker as well. Reviewed-by: Pavel Stehule <pavel.stehule@gmail.com> Discussion: https://www.postgresql.org/message-id/flat/b05e781a-fa16-6b52-6738-761181204567@2ndquadrant.com
* Add description about GSSOpenServer wait event into document.Fujii Masao2020-02-17
| | | | | | | | | | | | | This commit also updates wait event enum into alphabetical order. Previously the enum entry for GSSOpenServer was added out-of-order. Back-patch to v12 where commit b0b39f72b9 introduced GSSOpenServer wait event. In v12, the commit doesn't include the update of wait event enum, not to break ABI. Author: Fujii Masao Reviewed-by: Michael Paquier Discussion: https://postgr.es/m/949931aa-4ed4-d867-a7b5-de9c02b2292b@oss.nttdata.com
* Update obsolete comment.Tom Lane2020-02-15
| | | | | | | Noted by Justin Pryzby, though I chose to just rip out the stale text, as it's in no way relevant to this particular function. Discussion: https://postgr.es/m/20200212182337.GZ1412@telsasoft.com
* Clarify coding in Catalog::AddDefaultValues.Tom Lane2020-02-15
| | | | | | | | Make it a bit shorter and better-commented; no functional change. Alvaro Herrera and Tom Lane Discussion: https://postgr.es/m/20200212182337.GZ1412@telsasoft.com
* Don't require pg_class.dat to contain correct relnatts values.Tom Lane2020-02-15
| | | | | | | | | | | | | | | | | | | | | | | Practically everybody who's ever added a column to one of the bootstrap catalogs has been burnt by the need to update the relnatts field in the initial pg_class data to match. Now that we use Perl scripts to generate postgres.bki, we can have the machines take care of that, by filling the field during genbki.pl. While at it, use the BKI_DEFAULTS mechanism to eliminate repetitive specifications of other column values in pg_class.dat, too. They weren't particularly a maintenance problem, but this way is prettier (certainly the spotty previous usage of BKI_DEFAULTS wasn't pretty). No catversion bump needed, since this doesn't actually change the contents of postgres.bki. Per gripe from Justin Pryzby, though this is quite different from his originally proposed solution. Amit Langote, John Naylor, Tom Lane Discussion: https://postgr.es/m/20200212182337.GZ1412@telsasoft.com
* Logical Tape Set: lazily allocate read buffer.Jeff Davis2020-02-13
| | | | | | | | The write buffer was already lazily-allocated, so this is more symmetric. It also means that a freshly-rewound tape (whether for reading or writing) is not consuming memory for the buffer. Discussion: https://postgr.es/m/97c46a59c27f3c38e486ca170fcbc618d97ab049.camel%40j-davis.com
* Avoid a performance regression in float overflow/underflow detection.Tom Lane2020-02-13
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Commit 6bf0bc842 replaced float.c's CHECKFLOATVAL() macro with static inline subroutines, but that wasn't too well thought out. In the original coding, the unlikely condition (isinf(result) or result == 0) was checked first, and the inf_is_valid or zero_is_valid condition only afterwards. The inline-subroutine coding caused that to be swapped around, which is pretty horrid for performance because (a) in common cases the is_valid condition is twice as expensive to evaluate (e.g., requiring two isinf() calls not one) and (b) in common cases the is_valid condition is false, requiring us to perform the unlikely-condition check anyway. Net result is that one isinf() call becomes two or three, resulting in visible performance loss as reported by Keisuke Kuroda. The original fix proposal was to revert the replacement of the macro, but on second thought, that macro was just a bad idea from the beginning: if anything it's a net negative for readability of the code. So instead, let's just open-code all the overflow/underflow tests, being careful to test the unlikely condition first (and mark it unlikely() to help the compiler get the point). Also, rather than having N copies of the actual ereport() calls, collapse those into out-of-line error subroutines to save some code space. This does mean that the error file/line numbers won't be very helpful for figuring out where the issue really is --- but we'd already burned that bridge by putting the ereports into static inlines. In HEAD, check_float[48]_val() are gone altogether. In v12, leave them present in float.h but unused in the core code, just in case some extension is depending on them. Emre Hasegeli, with some kibitzing from me and Andres Freund Discussion: https://postgr.es/m/CANDwggLe1Gc1OrRqvPfGE=kM9K0FSfia0hbeFCEmwabhLz95AA@mail.gmail.com
* Remove long-dead comments.Tom Lane2020-02-12
| | | | | These should've been dropped by a8bb8eb58, but evidently were missed. Not important enough to back-patch.
* Use pg_pwrite() in more places.Thomas Munro2020-02-11
| | | | | | | | This removes some lseek() system calls. Author: Thomas Munro Reviewed-by: Andres Freund Discussion: https://postgr.es/m/CA%2BhUKGJ%2BoHhnvqjn3%3DHro7xu-YDR8FPr0FL6LF35kHRX%3D_bUzg%40mail.gmail.com
* Change signature of TupleHashTableHash().Jeff Davis2020-02-10
| | | | | | | | Commit 4eaea3db introduced TupleHashTableHash(), but the signature didn't match the other exposed functions. Separate it into internal and external versions. The external version hides the details behind an API more consistent with the other external functions, and the internal version is still suitable for simplehash.
* Fix priv checks for ALTER <object> DEPENDS ON EXTENSIONAlvaro Herrera2020-02-10
| | | | | | | | | | | | | | Marking an object as dependant on an extension did not have any privilege check whatsoever; this allowed any user to mark objects as droppable by anyone able to DROP EXTENSION, which could be used to cause system-wide havoc. Disallow by checking that the calling user owns the mentioned object. (No constraints are placed on the extension.) Security: CVE-2020-1720 Reported-by: Tom Lane Discussion: 31605.1566429043@sss.pgh.pa.us
* Fix typos.Amit Kapila2020-02-10
| | | | | | Reported-by: Justin Pryzby Author: Justin Pryzby Discussion: https://postgr.es/m/20200206021432.GA24549@telsasoft.com
* Store the deletion horizon XID for a deleted GIN page on the right page.Tom Lane2020-02-09
| | | | | | | | | | | | | | | | | Commit b10714080 moved the GinPageSetDeleteXid() call to a spot where the "page" variable was pointing to the wrong page, causing the XID to be inserted on a page that's not being deleted, thus allowing later GinPageIsRecyclable tests to recycle the deleted page too soon. It might be a good idea to stop using the single "page" variable for multiple purposes in this function. But for the moment I just moved the GinPageSetDeleteXid() call down beside the GinPageSetDeleted() call, which seems like a more logical place for it anyway. Back-patch to v11, as the faulty patch was. (Fortunately, the bug hasn't made it into any release yet.) Discussion: https://postgr.es/m/21620.1581098806@sss.pgh.pa.us
* Fix failure to create FKs correctly in partitionsAlvaro Herrera2020-02-07
| | | | | | | | | | | | | | | | | | | On a multi-level partioned table, when adding a partition not directly connected to the root table, foreign key constraints referencing the root were not cloned to the new partition, leading to the FK being possibly inadvertently violated later on. This was caused by fuzzy thinking in CloneFkReferenced (commit f56f8f8da6af): it was skipping constraints marked as having parents on the theory that cloning those would create duplicates; but that's only correct for the top level of the partitioning hierarchy. For levels below that one, such constraints must still be considered and only skipped if later on we see that we'd create duplicates. Apparently, I (Álvaro) wrote the comments right but the code implemented something slightly different. Author: Jehan-Guillaume de Rorthais Discussion: https://postgr.es/m/20200206004948.238352db@firost
* Fix TRUNCATE .. CASCADE on partitionsAlvaro Herrera2020-02-07
| | | | | | | | | | | | | | | | When running TRUNCATE CASCADE on a child of a partitioned table referenced by another partitioned table, the truncate was not applied to partitions of the referencing table; this could leave rows violating the constraint in the referencing partitioned table. Repair by walking the pg_constraint chain all the way up to the topmost referencing table. Note: any partitioned tables containing FKs that reference other partitioned tables should be checked for possible violating rows, if TRUNCATE has occurred in partitions of the referenced table. Reported-by: Christophe Courtois Author: Jehan-Guillaume de Rorthais Discussion: https://postgr.es/m/20200204183906.115f693e@firost
* Fix bug in Tid scan.Fujii Masao2020-02-07
| | | | | | | | | | | | | | | | | | | | | | | Commit 147e3722f7 changed Tid scan so that it calls table_beginscan() and uses the scan option for seq scan. This change caused two issues. (1) The change caused Tid scan to take a predicate lock on the entire relation in serializable transaction even when relation-level lock is not necessary. This could lead to an unexpected serialization error. (2) The change caused Tid scan to increment the number of seq_scan in pg_stat_*_tables views even though it's not seq scan. This could confuse the users. This commit adds the scan option for Tid scan and makes Tid scan use it, to avoid those issues. Back-patch to v12, where the bug was introduced. Author: Tatsuhito Kasahara Reviewed-by: Kyotaro Horiguchi, Masahiko Sawada, Fujii Masao Discussion: https://postgr.es/m/CAP0=ZVKy+gTbFmB6X_UW0pP3WaeJ-fkUWHoD-pExS=at3CY76g@mail.gmail.com
* jit: Reference expression step functions via llvmjit_types.Andres Freund2020-02-06
| | | | | | | | | | | | | | The main benefit of doing so is that this allows llvm to ensure that types match - previously that'd only be detected by a crash within the called function. There were a number of cases where we passed a superfluous parameter... To avoid needing to add all the functions to llvmjit.{c,h}, instead get them from the llvm module for llvmjit_types.c. Also use that for the functions from llvmjit_types already in llvmjit.h. Author: Soumyadeep Chakraborty and Andres Freund Discussion: https://postgr.es/m/CADwEdooww3wZv-sXSfatzFRwMuwa186LyTwkBfwEW6NjtooBPA@mail.gmail.com
* Introduce TupleHashTableHash() and LookupTupleHashEntryHash().Jeff Davis2020-02-06
| | | | | | | | | | Expose two new entry points: one for only calculating the hash value of a tuple, and another for looking up a hash entry when the hash value is already known. This will be useful for disk-based Hash Aggregation to avoid recomputing the hash value for the same tuple after saving and restoring it from disk. Discussion: https://postgr.es/m/37091115219dd522fd9ed67333ee8ed1b7e09443.camel%40j-davis.com
* jit: Remove redundancies in expression evaluation code generation.Andres Freund2020-02-06
| | | | | | | | | | This merges the code emission for a number of opcodes by handling the behavioural difference more locally. This reduces code, and also improves the generated code a bit in some cases, by removing redundant constants. Author: Andres Freund Discussion: https://postgr.es/m/20191023163849.sosqbfs5yenocez3@alap3.anarazel.de
* jit: Reference functions by name in IOCOERCE steps.Andres Freund2020-02-06
| | | | | | | | Previously we used constant function pointer addresses, which prevents inlining and other related optimizations. Author: Andres Freund Discussion: https://postgr.es/m/20191023163849.sosqbfs5yenocez3@alap3.anarazel.de
* expression eval: Don't redundantly keep track of AggState.Andres Freund2020-02-06
| | | | | | | | | | It's already tracked via ExprState->parent, so we don't need to also include it in ExprEvalStep. When that code originally was written ExprState->parent didn't exist, but it since has been introduced in 6719b238e8f. Author: Andres Freund Discussion: https://postgr.es/m/20191023163849.sosqbfs5yenocez3@alap3.anarazel.de
* expression eval, jit: Minor code cleanups.Andres Freund2020-02-06
| | | | | | | | | | This mostly consists of using C99 style for loops, moving variables into narrower scopes, and a smattering of other minor improvements. Done separately to make it easier to review patches with actual functional changes. Author: Andres Freund Discussion: https://postgr.es/m/20191023163849.sosqbfs5yenocez3@alap3.anarazel.de
* Fix typo in proc.cMichael Paquier2020-02-07
| | | | | Author: Julien Rouhaud Discussion: https://postgr.es/m/20200206082333.GA95343@nol
* Revert "Add GUC checks for ssl_min_protocol_version and ↵Michael Paquier2020-02-07
| | | | | | | | | | | | | ssl_max_protocol_version" This reverts commit 41aadee, as the GUC checks could run on older values with the new values used, and result in incorrect errors if both parameters are changed at the same time. Per complaint from Tom Lane. Discussion: https://postgr.es/m/27574.1581015893@sss.pgh.pa.us Backpatch-through: 12
* Ensure relcache consistency around generated columnsPeter Eisentraut2020-02-06
| | | | | | | | | | | | | In certain transient states, it's possible that a table has attributes with attgenerated set but no default expressions in pg_attrdef yet. In that case, the old code path would not set relation->rd_att->constr->has_generated_stored, unless relation->rd_att->constr was also populated for some other reason. There was probably no practical impact, but it's better to keep this consistent. Reported-by: Andres Freund <andres@anarazel.de> Discussion: https://www.postgresql.org/message-id/flat/20200115181105.ad6ab6dlgyww3lb6%40alap3.anarazel.de
* Refactor hash_agg_entry_size().Jeff Davis2020-02-06
| | | | | | Consolidate the calculations for hash table size estimation. This will help with upcoming Hash Aggregation work that will add additional call sites.
* Logical Tape Set: use min heap for freelist.Jeff Davis2020-02-06
| | | | | | | | Previously, the freelist of blocks was tracked as an occasionally-sorted array. A min heap is more resilient to larger freelists or more frequent changes between reading and writing. Discussion: https://postgr.es/m/97c46a59c27f3c38e486ca170fcbc618d97ab049.camel%40j-davis.com
generated by cgit v1.2.3 (git 2.25.1) at 2025-08-08 09:28:07 +0000