Speeds up text comparison expressions when using a collation other
than the database default collation. Does not affect larger operations
such as ORDER BY, because the lookup is only done once.
Discussion: https://postgr.es/m/7bb9f018d20a7b30b9a7f6231efab1b5e50c7720.camel@j-davis.com
Reviewed-by: John Naylor, Andreas Karlsson
pg_size_pretty(bigint) would return the value in bytes rather than PB
for the smallest-most bigint value. This happened due to an incorrect
assumption that the absolute value of -9223372036854775808 could be
stored inside a signed 64-bit type.
Here we fix that by instead storing that value in an unsigned 64-bit type.
This bug does exist in versions prior to 15 but the code there is
sufficiently different and the bug seems sufficiently non-critical that
it does not seem worth risking backpatching further.
Author: Joseph Koshakow <koshy44@gmail.com>
Discussion: https://postgr.es/m/CAAvxfHdTsMZPWEHUrZ=h3cky9Ccc3Mtx2whUHygY+ABP-mCmUw@mail.gmail.com
Backpatch-through: 15
There were quite a few places where we either had a non-NUL-terminated
string or a text Datum which we needed to call escape_json() on. Many of
these places required that a temporary string was created due to the fact
that escape_json() needs a NUL-terminated cstring. For text types, those
first had to be converted to cstring before calling escape_json() on them.
Here we introduce two new functions to make escaping JSON more optimal:
escape_json_text() can be given a text Datum to append onto the given
buffer. This is more optimal as it foregoes the need to convert the text
Datum into a cstring. A temporary allocation is only required if the text
Datum needs to be detoasted.
escape_json_with_len() can be used when the length of the cstring is
already known or the given string isn't NUL-terminated. Having this
allows various places which were creating a temporary NUL-terminated
string to just call escape_json_with_len() without any temporary memory
allocations.
Discussion: https://postgr.es/m/CAApHDvpLXwMZvbCKcdGfU9XQjGCDm7tFpRdTXuB9PVgpNUYfEQ@mail.gmail.com
Reviewed-by: Melih Mutlu, Heikki Linnakangas
The documentation for System V IPC parameters provides complicated
formulas to determine the appropriate values for SEMMNI and SEMMNS.
Furthermore, these formulas have often been wrong because folks
forget to update them (e.g., when adding a new auxiliary process).
This commit introduces a new runtime-computed GUC named
num_os_semaphores that reports the number of semaphores needed for
the configured number of allowed connections, worker processes,
etc. This new GUC allows us to simplify the formulas in the
documentation, and it should help prevent future inaccuracies.
Like the other runtime-computed GUCs, users can view it with
"postgres -C" before starting the server, which is useful for
preconfiguring the necessary operating system resources.
Reviewed-by: Tom Lane, Sami Imseih, Andres Freund, Robert Haas
Discussion: https://postgr.es/m/20240517164452.GA1914161%40nathanxps13
The new test tests the libpq fallback behavior on an early error,
which was fixed in the previous commit.
This adds an IS_INJECTION_POINT_ATTACHED() macro, to allow writing
injected test code alongside the normal source code. In principle, the
new test could've been implemented by an extra test module with a
callback that sets the FrontendProtocol global variable, but I think
it's more clear to have the test code right where the injection point
is, because it has pretty intimate knowledge of the surrounding
context it runs in.
Reviewed-by: Michael Paquier
Discussion: https://www.postgresql.org/message-id/CAOYmi%2Bnwvu21mJ4DYKUa98HdfM_KZJi7B1MhyXtnsyOO-PB6Ww%40mail.gmail.com
Commit 86db52a506 changed the locking of injection points to use only
atomic ops and spinlocks, to make it possible to define injection
points in processes that don't have a PGPROC entry (yet). However, it
didn't work in EXEC_BACKEND mode, because the pointer to shared memory
area was not initialized until the process "attaches" to all the
shared memory structs. To fix, pass the pointer to the child process
along with other global variables that need to be set up early.
Backpatch-through: 17
populate_domain() didn't take into account the omit_quotes flag passed
down to json_populate_type() by ExecEvalJsonCoercion() and that led
to incorrect behavior when the RETURNING type is a domain over
jsonb. Fix that by passing the flag by adding a new function
parameter to populate_domain().
Reported-by: Jian He <jian.universality@gmail.com>
Discussion: https://postgr.es/m/CACJufxEo4sUjKCYtda0_qt9tazqqKPmF1cqhW9KBOUeJFqQd2g@mail.gmail.com
Backpatch-through: 17
"path" provides a reliable method of determining the parent/child
relationships between memory contexts. Previously this could be done in
a non-reliable way by writing a recursive query and joining the "parent"
and "name" columns. This wasn't reliable as the names were not unique,
which could result in joining to the wrong parent.
To make this reliable, "path" stores an array of numerical identifiers
starting with the identifier for TopLevelMemoryContext. It contains an
element for each intermediate parent between that and the current context.
Incompatibility: Here we also adjust the "level" column to make it
1-based rather than 0-based. A 1-based level provides a convenient way
to access elements in the "path" array. e.g. path[level] gives the
identifier for the current context.
Identifiers are not stable across multiple evaluations of the view. In
an attempt to make these more stable for ad-hoc queries, the identifiers
are assigned breadth-first. Contexts closer to TopLevelMemoryContext
are less likely to change between queries and during queries.
Author: Melih Mutlu <m.melihmutlu@gmail.com>
Discussion: https://postgr.es/m/CAGPVpCThLyOsj3e_gYEvLoHkr5w=tadDiN_=z2OwsK3VJppeBA@mail.gmail.com
Reviewed-by: Andres Freund, Stephen Frost, Atsushi Torikoshi,
Reviewed-by: Michael Paquier, Robert Haas, David Rowley
Add extern declarations in appropriate header files for global
variables related to GUC. In many cases, this was handled quite
inconsistently before, with some GUC variables declared in a header
file and some only pulled in via ad-hoc extern declarations in various
.c files.
Also add PGDLLIMPORT qualifications to those variables. These were
previously missing because src/tools/mark_pgdllimport.pl has only been
used with header files.
This also fixes -Wmissing-variable-declarations warnings for GUC
variables (not yet part of the standard warning options).
Reviewed-by: Andres Freund <andres@anarazel.de>
Discussion: https://www.postgresql.org/message-id/flat/e0a62134-83da-4ba4-8cdb-ceb0111c95ce@eisentraut.org
When provided an empty initial array, array_set_slice() fails to
check for overflow when computing the new array's dimensions.
While such overflows are ordinarily caught by ArrayGetNItems(),
commands with the following form are accepted:
INSERT INTO t (i[-2147483648:2147483647]) VALUES ('{}');
To fix, perform the hazardous computations using overflow-detecting
arithmetic routines. As with commit 18b585155a, the added test
cases generate errors that include a platform-dependent value, so
we again use psql's VERBOSITY parameter to suppress printing the
message text.
Reported-by: Alexander Lakhin
Author: Joseph Koshakow
Reviewed-by: Jian He
Discussion: https://postgr.es/m/31ad2cd1-db94-bdb3-f91a-65ffdb4bef95%40gmail.com
Backpatch-through: 12
strtok() considers adjacent delimiters to be one delimiter, which is
arguably the wrong behavior in some cases. Replace with strsep(),
which has the right behavior: Adjacent delimiters create an empty
token.
Affected by this are parsing of:
- Stored SCRAM secrets
("SCRAM-SHA-256$<iterations>:<salt>$<storedkey>:<serverkey>")
- ICU collation attributes
("und@colStrength=primary;colCaseLevel=yes") for ICU older than
version 54
- PG_COLORS environment variable
("error=01;31:warning=01;35:note=01;36:locus=01")
- pg_regress command-line options with comma-separated list arguments
(--dbname, --create-role) (currently only used pg_regress_ecpg)
Reviewed-by: Kyotaro Horiguchi <horikyota.ntt@gmail.com>
Reviewed-by: David Steele <david@pgmasters.net>
Discussion: https://www.postgresql.org/message-id/flat/79692bf9-17d3-41e6-b9c9-fc8c3944222a@eisentraut.org
This new error code, named file_name_too_long, maps internally to the
errno ENAMETOOLONG to produce a proper error code rather than an
internal code under errcode_for_file_access(). This error code can be
reached with some SQL command patterns, like a snapshot file name.
Reported-by: Alexander Lakhin
Reviewed-by: Daniel Gustafsson
Discussion: https://postgr.es/m/Zo4ROR9mgy8bowMo@paquier.xyz
This new macro is able to perform a direct lookup from the local cache
of injection points (refreshed each time a point is loaded or run),
without touching the shared memory state of injection points at all.
This works in combination with INJECTION_POINT_LOAD(), and it is better
than INJECTION_POINT() in a critical section due to the fact that it
would avoid all memory allocations should a concurrent detach happen
since a LOAD(), as it retrieves a callback from the backend-private
memory.
The documentation is updated to describe in more details how to use this
new macro with a load. Some tests are added to the module
injection_points based on a new SQL function that acts as a wrapper of
INJECTION_POINT_CACHED().
Based on a suggestion from Heikki Linnakangas.
Author: Heikki Linnakangas, Michael Paquier
Discussion: https://postgr.es/m/58d588d0-e63f-432f-9181-bed29313dece@iki.fi
Commit f4b54e1ed9, which introduced macros for protocol characters,
missed updating a few places. It also did not introduce macros for
messages sent from parallel workers to their leader processes.
This commit adds a new section in protocol.h for those.
Author: Aleksander Alekseev
Discussion: https://postgr.es/m/CAJ7c6TNTd09AZq8tGaHS3LDyH_CCnpv0oOz2wN1dGe8zekxrdQ%40mail.gmail.com
Backpatch-through: 17
This switches the pgstats write code to use durable_rename() rather than
rename(). This ensures that the stats file's data is durable when the
statistics are written, which is something only happening at shutdown
now with the checkpointer doing the job.
This could cause the statistics to be lost even after PostgreSQL is shut
down, should a host failure happen, for example.
Suggested-by: Konstantin Knizhnik
Reviewed-by: Bertrand Drouvot
Discussion: https://postgr.es/m/ZpDQTZ0cAz0WEbh7@paquier.xyz
Before this change guc_var_compare() cast the input arguments to
const struct config_generic *. That's not quite right however, as the input
on one side is often just a char * on one side.
Instead just use char *, the first field in config_generic.
This fixes a -Warray-bounds warning with some versions of gcc. While the
warning is only known to be triggered for <= 15, the issue the warning points
out seems real, so apply the fix everywhere.
Author: Nazir Bilal Yavuz <byavuz81@gmail.com>
Reported-by: Erik Rijkers <er@xs4all.nl>
Suggested-by: Andres Freund <andres@anarazel.de>
Discussion: https://postgr.es/m/a74a1a0d-0fd2-3649-5224-4f754e8f91aa%40xs4all.nl
This routine can currently only be called from the postmaster in
single-user mode or the checkpointer, but there was no sanity check to
make sure that this was always the case.
This has proved to be useful when hacking the zone (at least to me), to
make sure that the write of the pgstats file happens at shutdown, as
wanted by design, in the correct process context.
Discussion: https://postgr.es/m/ZnEiqAITL-VgZDoY@paquier.xyz
The comment at the top of pgstat_read_statsfile() mentioned that the
stats are read from the on-disk file into the pgstats dshash. This is
incorrect for fix-numbered stats as these are loaded directly into
shared memory. This commit simplifies the comment to be more general.
Author: Bertrand Drouvot
Discussion: https://postgr.es/m/Zo/eJIHUcqKxeSgv@ip-10-97-1-34.eu-west-3.compute.internal
This new entry type is used for all the fixed-numbered statistics,
making possible support for custom pluggable stats. In short, we need
to be able to detect more easily if a stats kind exists or not when
reading back its data from the pgstats file without a dependency on the
order of the entries read. The kind ID of the stats is added to the
data written.
The data is written in the same fashion as previously, with the
fixed-numbered stats first and the dshash entries next. The read part
becomes more flexible, loading fixed-numbered stats into shared memory
based on the new entry type found.
Bump PGSTAT_FILE_FORMAT_ID.
Reviewed-by: Bertrand Drouvot
Discussion: https://postgr.es/m/Zot5bxoPYdS7yaoy@paquier.xyz
This new callback gives fixed-numbered stats the possibility to take
actions based on the area of shared memory allocated for them.
This removes from pgstat_shmem.c any knowledge specific to the types
of fixed-numbered stats, and the initializations happen in their own
files. Like b68b29bc8f, this change is useful to make this area of
the code more pluggable, so as custom fixed-numbered stats can take
actions after their shared memory area is initialized.
Reviewed-by: Bertrand Drouvot
Discussion: https://postgr.es/m/Zot5bxoPYdS7yaoy@paquier.xyz
Formerly, the computation of the bucket index involved calling
div_var() with a scale determined by select_div_scale(), and then
taking the floor of the result. That involved computing anything from
16 to 1000 digits after the decimal point, only for floor_var() to
throw them away. In addition, the quotient was computed with rounding
in the final digit, which meant that in rare cases the whole result
could round up to the wrong bucket, and could exceed count. Thus it
was also necessary to clamp the result to the range [1, count], though
that didn't prevent the result being in the wrong internal bucket.
Instead, compute the quotient using floor division, which guarantees
the correct result, as specified by the SQL spec, and doesn't need to
be clamped. This is both much simpler and more efficient, since it no
longer computes any quotient digits after the decimal point.
In addition, it is not necessary to have separate code to handle
reversed bounds, since the signs cancel out when dividing.
As with b0e9e4d76c and a2a0c7c29e, no back-patch.
Dean Rasheed, reviewed by Joel Jacobson.
Discussion: https://postgr.es/m/CAEZATCVbJH%2BLE9EXW8Rk3AxLe%3DjbOk2yrT_AUJGGh5Rah6zoeg%40mail.gmail.com
libxml2 2.13 has an entirely different rule than earlier versions
about when to emit "chunk is not well balanced" errors. This
causes regression test output discrepancies for three test cases
that formerly provoked that error (along with others) and now don't.
Closer inspection shows that at least in 2.13, this error is pretty
useless because it can only be emitted after some other more-relevant
error. So let's get rid of the cross-version discrepancy by just
suppressing it. In case some older libxml2 version is capable of
emitting this error by itself, suppress only when some other error
has already been captured.
Like 066e8ac6e and 6082b3d5d, this will need to be back-patched,
but let's check the results in HEAD first. (The patch for xml_2.out,
in particular, is blind since I can't test it here.)
Erik Wienhold and Tom Lane, per report from Frank Streitzig.
Discussion: https://postgr.es/m/trinity-b0161630-d230-4598-9ebc-7a23acdb37cb-1720186432160@3c-app-gmx-bap25
Discussion: https://postgr.es/m/trinity-361ba18b-541a-4fe7-bc63-655ae3a7d599-1720259822452@3c-app-gmx-bs01
When either input has a small number of digits, and the exact product
is requested, the speed of numeric multiplication can be increased
significantly by using a faster direct multiplication algorithm. This
works by fully computing each result digit in turn, starting with the
least significant, and propagating the carry up. This save cycles by
not requiring a temporary buffer to store digit products, not making
multiple passes over the digits of the longer input, and not requiring
separate carry-propagation passes.
For now, this is used when the shorter input has 1-4 NBASE digits (up
to 13-16 decimal digits), and the longer input is of any size, which
covers a lot of common real-world cases. Also, the relative benefit
increases as the size of the longer input increases.
Possible future work would be to try extending the technique to larger
numbers of digits in the shorter input.
Joel Jacobson and Dean Rasheed.
Discussion: https://postgr.es/m/44d2ffca-d560-4919-b85a-4d07060946aa@app.fastmail.com
This is similar to 9004abf620, but this time for the write part of the
stats file. The code is changed so as, rather than referring to
individual members of PgStat_Snapshot in an order based on their
PgStat_Kind value, a loop based on pgstat_kind_infos is used to retrieve
the contents to write from the snapshot structure, for a size of
PgStat_KindInfo's shared_data_len.
This requires the addition to PgStat_KindInfo of an offset to track the
location of each fixed-numbered stats in PgStat_Snapshot. This change
is useful to make this area of the code more easily pluggable, and
reduces the knowledge of specific fixed-numbered kinds in pgstat.c.
Reviewed-by: Bertrand Drouvot
Discussion: https://postgr.es/m/Zot5bxoPYdS7yaoy@paquier.xyz
xmlParseInNodeContext has basically the same functionality with
a different API: we have to supply an xmlNode that's attached to a
document rather than just the document. That's not hard though.
The benefits are two:
* Early 2.13.x releases of libxml2 contain a bug that causes
xmlParseBalancedChunkMemory to return the wrong status value in some
cases. This breaks our regression tests. While that bug is now fixed
upstream and will probably never be seen in any production-oriented
distro, it is currently a problem on some more-bleeding-edge-friendly
platforms.
* xmlParseBalancedChunkMemory is considered to depend on libxml2's
semi-deprecated SAX1 APIs, and will go away when and if they do.
There may already be libxml2 builds out there that lack this function.
So there are both short- and long-term reasons to make this change.
While here, avoid allocating an xmlParserCtxt in DOCUMENT parse mode,
since that code path is not going to use it.
Like 066e8ac6e, this will need to be back-patched. This is just a
trial commit to see if the buildfarm agrees that we can use
xmlParseInNodeContext unconditionally.
Erik Wienhold and Tom Lane, per report from Frank Streitzig.
Discussion: https://postgr.es/m/trinity-b0161630-d230-4598-9ebc-7a23acdb37cb-1720186432160@3c-app-gmx-bap25
Discussion: https://postgr.es/m/trinity-361ba18b-541a-4fe7-bc63-655ae3a7d599-1720259822452@3c-app-gmx-bs01
The numeric round() and trunc() functions clamp the scale argument to
the range between +/- NUMERIC_MAX_RESULT_SCALE (2000), which is much
smaller than the actual allowed range of type numeric. As a result,
they return incorrect results when asked to round/truncate more than
2000 digits before or after the decimal point.
Fix by using the correct upper and lower scale limits based on the
actual allowed (and documented) range of type numeric.
While at it, use the new NUMERIC_WEIGHT_MAX constant instead of
SHRT_MAX in all other overflow checks, and fix a comment thinko in
power_var() introduced by e54a758d24 -- the minimum value of
ln_dweight is -NUMERIC_DSCALE_MAX (-16383), not -SHRT_MAX, though this
doesn't affect the point being made in the comment, that the resulting
local_rscale value may exceed NUMERIC_MAX_DISPLAY_SCALE (1000).
Back-patch to all supported branches.
Dean Rasheed, reviewed by Joel Jacobson.
Discussion: https://postgr.es/m/CAEZATCXB%2BrDTuMjhK5ZxcouufigSc-X4tGJCBTMpZ3n%3DxxQuhg%40mail.gmail.com
590b045c3 made it so tuplestore.c would store tuples inside a
generation.c memory context. After fixing a bug report in 97651b013, it
seems that it's probably best not to allocate BufFile related
allocations in that context. Let's keep it just for tuple data.
This adjusts the code to switch to the Tuplestorestate.context's parent,
which is the MemoryContext that tuplestore_begin_common() was called in.
It does not seem worth adding a new field in Tuplestorestate to store
this when we can access it by looking at the Tuplestorestate's
context's parent.
Discussion: https://postgr.es/m/CAApHDvqFt_CdJtSr+E9YLZb7jZAyRCy3hjQ+ktM+dcOFVq-xkg@mail.gmail.com
This only affects MEMORY_CONTEXT_CHECKING builds.
This fixes an off-by-one issue in GenerationRealloc() where the
fast-path code which tries to reuse the existing allocation if the
existing chunk is >= the new requested size. The code there thought it
was always ok to use the existing chunk, but when oldsize == size there
isn't enough space to store the sentinel byte. If both sizes matched
exactly set_sentinel() would overwrite the first byte beyond the chunk
and then subsequent GenerationRealloc() calls could then fail the
Assert(chunk->requested_size < oldsize) check which is trying to ensure
the chunk is large enough to store the sentinel.
The same issue does not exist in aset.c as the sentinel checking code
only adds a sentinel byte if there's enough space in the chunk.
Reported-by: Alexander Lakhin <exclusion@gmail.com>
Discussion: https://postgr.es/m/49275921-7b39-41af-5eb8-97b50ce3312e@gmail.com
Backpatch-through: 16, where the problem was introduced by 0e480385e
Each of max_connections, max_worker_processes,
autovacuum_max_workers, and max_wal_senders has a GUC check hook
that verifies the sum of those GUCs does not exceed a hard-coded
limit (see the comment for MAX_BACKENDS in postmaster.h). In
general, the hooks effectively guard against egregious
misconfigurations.
However, this approach has some problems. Since these check hooks
are called as each GUC is assigned its user-specified value, only
one of the hooks will be called with all the relevant GUCs set. If
one or more of the user-specified values are less than the initial
values of the GUCs' underlying variables, false positives can
occur.
Furthermore, the error message emitted when one of the check hooks
fails is not tremendously helpful. For example, the command
$ pg_ctl -D . start -o "-c max_connections=262100 -c max_wal_senders=10000"
fails with the following error:
FATAL: invalid value for parameter "max_wal_senders": 10000
Fortunately, there is an extra copy of this check in
InitializeMaxBackends() that we can rely on, so this commit removes
the aforementioned GUC check hooks in favor of that one. It also
enhances the error message to clearly show the values of the
relevant GUCs and the hard-coded limit their sum may not exceed.
The downside of this change is that server startup progresses
further before failing due to such misconfigurations (thus taking
longer), but these failures are expected to be rare, so we don't
anticipate any real harm in practice.
Reviewed-by: Tom Lane
Discussion: https://postgr.es/m/ZnMr2k-Nk5vj7T7H%40nathan
This can be used to load an injection point and prewarm the
backend-level cache before running it, to avoid issues if the point
cannot be loaded due to restrictions in the code path where it would be
run, like a critical section where no memory allocation can happen
(load_external_function() can do allocations when expanding a library
name).
Tests can use a macro called INJECTION_POINT_LOAD() to load an injection
point. The test module injection_points gains some tests, and a SQL
function able to load an injection point.
Based on a request from Andrey Borodin, who has implemented a test for
multixacts requiring this facility.
Reviewed-by: Andrey Borodin
Discussion: https://postgr.es/m/ZkrBE1e2q2wGvsoN@paquier.xyz
Here we make tuplestore.c use a generation.c memory context rather than
allocating tuples into the CurrentMemoryContext, which primarily is the
ExecutorState or PortalHoldContext memory context. Not having a
dedicated context can cause the CurrentMemoryContext context to become
bloated when pfree'd chunks are not reused by future tuples. Using
generation speeds up users of tuplestore.c, such as the Materialize,
WindowAgg and CTE Scan executor nodes. The main reason for the speedup is
due to generation.c being more memory efficient than aset.c memory
contexts. Specifically, generation does not round sizes up to the next
power of 2 value. This both saves memory, allowing more tuples to fit in
work_mem, but also makes the memory usage more compact and fit on fewer
cachelines. One benchmark showed up to a 22% performance increase in a
query containing a Materialize node. Much higher gains are possible if
the memory reduction prevents tuplestore.c from spilling to disk. This is
especially true for WindowAgg nodes where improvements of several thousand
times are possible if the memory reductions made here prevent tuplestore
from spilling to disk.
Additionally, a generation.c memory context is much better suited for this
job as it works well with FIFO palloc/pfree patterns, which is exactly how
tuplestore.c uses it. Because of the way generation.c allocates memory,
tuples consecutively stored in tuplestores are much more likely to be
stored consecutively in memory. This allows the CPU's hardware prefetcher
to work more efficiently as it provides a more predictable pattern to
allow cachelines for the next tuple to be loaded from RAM in advance of
them being needed by the executor.
Using a dedicated memory context for storing tuples also allows us to more
efficiently clean up the memory used by the tuplestore as we can reset or
delete the context rather than looping over all stored tuples and
pfree'ing them one by one.
Also, remove a badly placed USEMEM call in readtup_heap(). The tuple
wasn't being allocated in the Tuplestorestate's context, so no need to
adjust the memory consumed by the tuplestore there.
Author: David Rowley
Reviewed-by: Matthias van de Meent, Dmitry Dolgov
Discussion: https://postgr.es/m/CAApHDvp5Py9g4Rjq7_inL3-MCK1Co2CRt_YWFwTU2zfQix0p4A@mail.gmail.com
Up until now, there was no ability to easily determine if a Material
node caused the underlying tuplestore to spill to disk or even see how
much memory the tuplestore used if it didn't.
Here we add some new functions to tuplestore.c to query this information
and add some additional output in EXPLAIN ANALYZE to display this
information for the Material node.
There are a few other executor node types that use tuplestores, so we
could also consider adding these details to the EXPLAIN ANALYZE for
those nodes too. Let's consider those independently from this. Having
the tuplestore.c infrastructure in to allow that is step 1.
Author: David Rowley
Reviewed-by: Matthias van de Meent, Dmitry Dolgov
Discussion: https://postgr.es/m/CAApHDvp5Py9g4Rjq7_inL3-MCK1Co2CRt_YWFwTU2zfQix0p4A@mail.gmail.com
All the errors triggered in the code paths patched here would cause the
backend to issue an internal_error errcode, which is a state that should
be used only for "can't happen" situations. However, these code paths
are reachable by the regression tests, and could be seen by users in
valid cases. Some regression tests expect internal errcodes as they
manipulate the backend state to cause corruption (like checksums), or
use elog() because it is more convenient (like injection points), these
have no need to change.
This reduces the number of internal failures triggered in a check-world
by more than half, while providing correct errcodes for these valid
cases.
Reviewed-by: Robert Haas
Discussion: https://postgr.es/m/Zic_GNgos5sMxKoa@paquier.xyz
This changes pgstat.c so as the three types of entries that can exist in
a pgstats file are not hardcoded anymore, replacing them with
descriptively-named macros, when reading and writing stats files:
- 'N' for named entries, like replication slot stats.
- 'S' for entries identified by a hash.
- 'E' for the end-of-file
This has come up while working on making this area of the code more
pluggable. The format of the stats file is unchanged, hence there is no
need to bump PGSTAT_FILE_FORMAT_ID.
Reviewed-by: Bertrand Drouvot
Discussion: https://postgr.es/m/Zmqm9j5EO0I4W8dx@paquier.xyz
We have in launch_backend.c:
/*
* The following need to be available to the save/restore_backend_variables
* functions. They are marked NON_EXEC_STATIC in their home modules.
*/
extern slock_t *ShmemLock;
extern slock_t *ProcStructLock;
extern PGPROC *AuxiliaryProcs;
extern PMSignalData *PMSignalState;
extern pg_time_t first_syslogger_file_time;
extern struct bkend *ShmemBackendArray;
extern bool redirection_done;
That comment is not completely true: ShmemLock, ShmemBackendArray, and
redirection_done are not in fact NON_EXEC_STATIC. ShmemLock once was,
but was then needed elsewhere. ShmemBackendArray was static inside
postmaster.c before launch_backend.c was created. redirection_done
was never static.
This patch moves the declaration of ShmemLock and redirection_done to
a header file.
ShmemBackendArray gets a NON_EXEC_STATIC. This doesn't make a
difference, since it only exists if EXEC_BACKEND anyway, but it makes
it consistent.
After that, the comment is now correct.
Reviewed-by: Andres Freund <andres@anarazel.de>
Discussion: https://www.postgresql.org/message-id/flat/e0a62134-83da-4ba4-8cdb-ceb0111c95ce@eisentraut.org
src/backend/libpq/pqcomm.c: "postmaster/postmaster.h" for Unix_socket_group, Unix_socket_permissions
src/backend/utils/init/globals.c: "postmaster/postmaster.h" for MyClientSocket
src/backend/utils/misc/guc_tables.c: "utils/rls.h" for row_security
src/backend/utils/sort/tuplesort.c: "utils/guc.h" for trace_sort
Nothing currently diagnoses missing includes for global variables, but
this is being cleaned up, and these ones had an obvious header file
available.
Reviewed-by: Andres Freund <andres@anarazel.de>
Discussion: https://www.postgresql.org/message-id/flat/e0a62134-83da-4ba4-8cdb-ceb0111c95ce@eisentraut.org
This commit removes unused variables and routines from some perl code
that have accumulated across the years. This touches the following
areas:
- Wait event generation script.
- AdjustUpgrade.pm.
- TAP perl code
Author: Alexander Lakhin
Reviewed-by: Dagfinn Ilmari Mannsåker
Discussion: https://postgr.es/m/70b340bc-244a-589d-ef8b-d8aebb707a84@gmail.com
Up to now, committing a transaction has caused CurrentMemoryContext to
get set to TopMemoryContext. Most callers did not pay any particular
heed to this, which is problematic because TopMemoryContext is a
long-lived context that never gets reset. If the caller assumes it
can leak memory because it's running in a limited-lifespan context,
that behavior translates into a session-lifespan memory leak.
The first-reported instance of this involved ProcessIncomingNotify,
which is called from the main processing loop that normally runs in
MessageContext. That outer-loop code assumes that whatever it
allocates will be cleaned up when we're done processing the current
client message --- but if we service a notify interrupt, then whatever
gets allocated before the next switch to MessageContext will be
permanently leaked in TopMemoryContext. sinval catchup interrupts
have a similar problem, and I strongly suspect that some places in
logical replication do too.
To fix this in a generic way, let's redefine the behavior as
"CommitTransactionCommand restores the memory context that was current
at entry to StartTransactionCommand". This clearly fixes the issue
for the notify and sinval cases, and it seems to match the mental
model that's in use in the logical replication code, to the extent
that anybody thought about it there at all.
For consistency, likewise make subtransaction exit restore the context
that was current at subtransaction start (rather than always selecting
the CurTransactionContext of the parent transaction level). This case
has less risk of resulting in a permanent leak than the outer-level
behavior has, but it would not meet the principle of least surprise
for some CommitTransactionCommand calls to restore the previous
context while others don't.
While we're here, also change xact.c so that we reset
TopTransactionContext at transaction exit and then re-use it in later
transactions, rather than dropping and recreating it in each cycle.
This probably doesn't save a lot given the context recycling mechanism
in aset.c, but it should save a little bit. Per suggestion from David
Rowley. (Parenthetically, the text in src/backend/utils/mmgr/README
implies that this is how I'd planned to implement it as far back as
commit 1aebc3618 --- but the code actually added in that commit just
drops and recreates it each time.)
This commit also cleans up a few places outside xact.c that were
needlessly making TopMemoryContext current, and thus risking more
leaks of the same kind. I don't think any of them represent
significant leak risks today, but let's deal with them while the
issue is top-of-mind.
Per bug #18512 from wizardbrony. Commit to HEAD only, as this change
seems to have some risk of breaking things for some callers. We'll
apply a narrower fix for the known-broken cases in the back branches.
Discussion: https://postgr.es/m/3478884.1718656625@sss.pgh.pa.us
The standby_slot_names GUC allows the specification of physical standby
slots that must be synchronized before the logical walsenders associated
with logical failover slots. However, for this purpose, the GUC name is
too generic.
Author: Hou Zhijie
Reviewed-by: Bertrand Drouvot, Masahiko Sawada
Backpatch-through: 17
Discussion: https://postgr.es/m/ZnWeUgdHong93fQN@momjian.us
