If DSM entry initialization fails, backends could try to use an
uninitialized DSM segment, DSA, or dshash table (since the entry is
still added to the registry). To fix, keep track of whether
initialization completed, and ERROR if a backend tries to attach to
an uninitialized entry. We could instead retry initialization as
needed, but that seemed complicated, error prone, and unlikely to
help most cases. Furthermore, such problems probably indicate a
coding error.
Reported-by: Alexander Lakhin <exclusion@gmail.com>
Reviewed-by: Sami Imseih <samimseih@gmail.com>
Discussion: https://postgr.es/m/dd36d384-55df-4fc2-825c-5bc56c950fa9%40gmail.com
Backpatch-through: 17
Before we started to freeze async notify entries (commit 8eeb4a0f7c),
no one looked at the 'xid' on an entry with invalid 'dboid'. But now
we might actually need to freeze it later. Initialize them with
InvalidTransactionId to begin with, to avoid that work later.
Álvaro pointed this out in review of commit 8eeb4a0f7c, but I forgot
to include this change there.
Author: Álvaro Herrera <alvherre@kurilemu.de>
Discussion: https://www.postgresql.org/message-id/202511071410.52ll56eyixx7@alvherre.pgsql
Backpatch-through: 14
Previous commit fixed a bug where VACUUM would truncate the CLOG
that's still needed to check the commit status of XIDs in the async
notify queue, but as mentioned in the commit message, it wasn't a full
fix. If a backend is executing asyncQueueReadAllNotifications() and
has just made a local copy of an async SLRU page which contains old
XIDs, vacuum can concurrently truncate the CLOG covering those XIDs,
and the backend still gets an error when it calls
TransactionIdDidCommit() on those XIDs in the local copy. This commit
fixes that race condition.
To fix, hold the SLRU bank lock across the TransactionIdDidCommit()
calls in NOTIFY processing.
Per Tom Lane's idea. Backpatch to all supported versions.
Reviewed-by: Joel Jacobson <joel@compiler.org>
Reviewed-by: Arseniy Mukhin <arseniy.mukhin.dev@gmail.com>
Discussion: https://www.postgresql.org/message-id/2759499.1761756503@sss.pgh.pa.us
Backpatch-through: 14
The async notification queue contains the XID of the sender, and when
processing notifications we call TransactionIdDidCommit() on the
XID. But we had no safeguards to prevent the CLOG segments containing
those XIDs from being truncated away. As a result, if a backend didn't
for some reason process its notifications for a long time, or when a
new backend issued LISTEN, you could get an error like:
test=# listen c21;
ERROR: 58P01: could not access status of transaction 14279685
DETAIL: Could not open file "pg_xact/000D": No such file or directory.
LOCATION: SlruReportIOError, slru.c:1087
To fix, make VACUUM "freeze" the XIDs in the async notification queue
before truncating the CLOG. Old XIDs are replaced with
FrozenTransactionId or InvalidTransactionId.
Note: This commit is not a full fix. A race condition remains, where a
backend is executing asyncQueueReadAllNotifications() and has just
made a local copy of an async SLRU page which contains old XIDs, while
vacuum concurrently truncates the CLOG covering those XIDs. When the
backend then calls TransactionIdDidCommit() on those XIDs from the
local copy, you still get the error. The next commit will fix that
remaining race condition.
This was first reported by Sergey Zhuravlev in 2021, with many other
people hitting the same issue later. Thanks to:
- Alexandra Wang, Daniil Davydov, Andrei Varashen and Jacques Combrink
for investigating and providing reproducable test cases,
- Matheus Alcantara and Arseniy Mukhin for review and earlier proposed
patches to fix this,
- Álvaro Herrera and Masahiko Sawada for reviews,
- Yura Sokolov aka funny-falcon for the idea of marking transactions
as committed in the notification queue, and
- Joel Jacobson for the final patch version. I hope I didn't forget
anyone.
Backpatch to all supported versions. I believe the bug goes back all
the way to commit d1e027221d, which introduced the SLRU-based async
notification queue.
Discussion: https://www.postgresql.org/message-id/16961-25f29f95b3604a8a@postgresql.org
Discussion: https://www.postgresql.org/message-id/18804-bccbbde5e77a68c2@postgresql.org
Discussion: https://www.postgresql.org/message-id/CAK98qZ3wZLE-RZJN_Y%2BTFjiTRPPFPBwNBpBi5K5CU8hUHkzDpw@mail.gmail.com
Backpatch-through: 14
Previously, if async notify processing encountered an error, we would
report the error to the client and advance our read position past the
offending entry to prevent trying to process it over and over
again. Trying to continue after an error has a few problems however:
- We have no way of telling the client that a notification was
lost. They get an ERROR, but that doesn't tell you much. As such,
it's not clear if keeping the connection alive after losing a
notification is a good thing. Depending on the application logic,
missing a notification could cause the application to get stuck
waiting, for example.
- If the connection is idle, PqCommReadingMsg is set and any ERROR is
turned into FATAL anyway.
- We bailed out of the notification processing loop on first error
without processing any subsequent notifications. The subsequent
notifications would not be processed until another notify interrupt
arrives. For example, if there were two notifications pending, and
processing the first one caused an ERROR, the second notification
would not be processed until someone sent a new NOTIFY.
This commit changes the behavior so that any ERROR while processing
async notifications is turned into FATAL, causing the client
connection to be terminated. That makes the behavior more consistent
as that's what happened in idle state already, and terminating the
connection is a clear signal to the application that it might've
missed some notifications.
The reason to do this now is that the next commits will change the
notification processing code in a way that would make it harder to
skip over just the offending notification entry on error.
Reviewed-by: Matheus Alcantara <matheusssilv97@gmail.com>
Reviewed-by: Álvaro Herrera <alvherre@kurilemu.de>
Reviewed-by: Arseniy Mukhin <arseniy.mukhin.dev@gmail.com>
Discussion: https://www.postgresql.org/message-id/fedbd908-4571-4bbe-b48e-63bfdcc38f64@iki.fi
Backpatch-through: 14
Previously, error messages for oversized injection point names, libraries,
and functions showed buffer sizes (64, 128, 128) instead of the usable
character limits (63, 127, 127) as it did not count for the
zero-terminated byte, which was confusing. These messages are adjusted
to show better the reality.
The limit enforced for the private area was also too strict by one byte,
as specifying a zone worth exactly INJ_PRIVATE_MAXLEN should be able to
work because three is no zero-terminated byte in this case.
This is a stylistic change (well, mostly, a private_area size of exactly
1024 bytes can be defined with this change, something that nobody seem
to care about based on the lack of complaints). However, this is a
testing facility let's keep the logic consistent across all the branches
where this code exists, as there is an argument in favor of out-of-core
extensions that use injection points.
Author: Xuneng Zhou <xunengzhou@gmail.com>
Co-authored-by: Michael Paquier <michael@paquier.xyz>
Discussion: https://postgr.es/m/CABPTF7VxYp4Hny1h+7ejURY-P4O5-K8WZg79Q3GUx13cQ6B2kg@mail.gmail.com
Backpatch-through: 17
This omission allowed table owners to create statistics in any
schema, potentially leading to unexpected naming conflicts. For
ALTER TABLE commands that require re-creating statistics objects,
skip this check in case the user has since lost CREATE on the
schema. The addition of a second parameter to CreateStatistics()
breaks ABI compatibility, but we are unaware of any impacted
third-party code.
Reported-by: Jelte Fennema-Nio <postgres@jeltef.nl>
Author: Jelte Fennema-Nio <postgres@jeltef.nl>
Co-authored-by: Nathan Bossart <nathandbossart@gmail.com>
Reviewed-by: Noah Misch <noah@leadboat.com>
Reviewed-by: Álvaro Herrera <alvherre@kurilemu.de>
Security: CVE-2025-12817
Backpatch-through: 13
In generate_orderedappend_paths(), there is an assumption that a child
relation's row estimate is always greater than zero. There is an
Assert verifying this assumption, and the estimate is also used to
convert an absolute tuple count into a fraction.
However, this assumption is not always valid -- for example, upper
relations can have their row estimates unset, resulting in a value of
zero. This can cause an assertion failure in debug builds or lead to
the tuple fraction being computed as infinity in production builds.
To fix, use the row estimate from the cheapest_total path to compute
the tuple fraction. The row estimate in this path should already have
been forced to a valid value.
In passing, update the comment for generate_orderedappend_paths() to
note that the function also considers the cheapest-fractional case
when not all tuples need to be retrieved. That is, it collects all
the cheapest fractional paths and builds an ordered append path for
each interesting ordering.
Backpatch to v18, where this issue was introduced.
Bug: #19102
Reported-by: Kuntal Ghosh <kuntalghosh.2007@gmail.com>
Author: Richard Guo <guofenglinux@gmail.com>
Reviewed-by: Kuntal Ghosh <kuntalghosh.2007@gmail.com>
Reviewed-by: Andrei Lepikhov <lepihov@gmail.com>
Discussion: https://postgr.es/m/19102-93480667e1200169@postgresql.org
Backpatch-through: 18
The comment for ChangeVarNodes() refers to a parameter named
change_RangeTblRef, which does not exist in the code.
The comment for ChangeVarNodesExtended() contains an extra space,
while the comment for replace_relid_callback() has an awkward line
break and a typo.
This patch fixes these issues and revises some sentences for smoother
wording.
Oversights in commits ab42d643c and fc069a3a6.
Author: Richard Guo <guofenglinux@gmail.com>
Discussion: https://postgr.es/m/CAMbWs480j16HC1JtjKCgj5WshivT8ZJYkOfTyZAM0POjFomJkg@mail.gmail.com
Backpatch-through: 18
First, the assertions in assign_io_method() were the wrong way round. Second,
the lengthof() assertion checked the length of io_method_options, which is the
wrong array to check and is always longer than pgaio_method_ops_table.
While add it, add a static assert to ensure pgaio_method_ops_table and
io_method_options stay in sync.
Per coverity and Tom Lane.
Reported-by: Tom Lane <tgl@sss.pgh.pa.us>
Backpatch-through: 18
In 2a0faed9d7, which added JIT compilation support for expressions, I
accidentally used sizeof(LLVMBasicBlockRef *) instead of
sizeof(LLVMBasicBlockRef) as part of computing the size of an allocation. That
turns out to have no real negative consequences due to LLVMBasicBlockRef being
a pointer itself (and thus having the same size). It still is wrong and
confusing, so fix it.
Reported by coverity.
Backpatch-through: 13
Allow pg_newlocale_from_collation(C_COLLATION_OID) to work even if
there's no catalog access, which some extensions expect.
Not known to be a bug without extensions involved, but backport to 18.
Also corrects an issue in master with dummy_c_locale (introduced in
commit 5a38104b36) where deterministic was not set. That wasn't a bug,
but could have been if that structure was used more widely.
Reported-by: Alexander Kukushkin <cyberdemn@gmail.com>
Reviewed-by: Alexander Kukushkin <cyberdemn@gmail.com>
Discussion: https://postgr.es/m/CAFh8B=nj966ECv5vi_u3RYij12v0j-7NPZCXLYzNwOQp9AcPWQ@mail.gmail.com
Backpatch-through: 18
This commit adds CHECK_FOR_INTERRUPTS to the shared buffer iteration
loops in EvictRelUnpinnedBuffers and EvictAllUnpinnedBuffers. These
functions, used by pg_buffercache's pg_buffercache_evict_relation and
pg_buffercache_evict_all, can now be interrupted during long-running
operations.
Backpatch to version 18, where these functions and their corresponding
pg_buffercache functions were introduced.
Author: Yuhang Qiu <iamqyh@gmail.com>
Discussion: https://postgr.es/m/8DC280D4-94A2-4E7B-BAB9-C345891D0B78%40gmail.com
Backpatch-through: 18
Commit a95e3d84c0 added ActiveSnapshot push+pop when processing
work-items (BRIN autosummarization), but forgot to handle the case of
a transaction failing during the run, which drops the snapshot untimely.
Fix by making the pop conditional on an element being actually there.
Author: Álvaro Herrera <alvherre@kurilemu.de>
Backpatch-through: 13
Discussion: https://postgr.es/m/202511041648.nofajnuddmwk@alvherre.pgsql
When building intermediate TID lists during parallel GIN builds, split
the sorted lists into smaller chunks, to limit the amount of memory
needed when merging the chunks later.
The leader may need to keep in memory up to one chunk per worker, and
possibly one extra chunk (before evicting some of the data). The code
processing item pointers uses regular palloc/repalloc calls, which means
it's subject to the MaxAllocSize (1GB) limit.
We could fix this by allowing huge allocations, but that'd require
changes in many places without much benefit. Larger chunks do not
actually improve performance, so the memory usage would be wasted.
Fixed by limiting the chunk size to not hit MaxAllocSize. Each worker
gets a fair share.
This requires remembering the number of participating workers, in a
place that can be accessed from the callback. Luckily, the bs_worker_id
field in GinBuildState was unused, so repurpose that.
Report by Greg Smith, investigation and fix by me. Batchpatched to 18,
where parallel GIN builds were introduced.
Reported-by: Gregory Smith <gregsmithpgsql@gmail.com>
Discussion: https://postgr.es/m/CAHLJuCWDwn-PE2BMZE4Kux7x5wWt_6RoWtA0mUQffEDLeZ6sfA@mail.gmail.com
Backpatch-through: 18
It's possible to define BRIN indexes on functions that require a
snapshot to run, but the autosummarization feature introduced by commit
7526e10224 fails to provide one. This causes autovacuum to leave a
BRIN placeholder tuple behind after a failed work-item execution, making
such indexes less efficient. Repair by obtaining a snapshot prior to
running the task, and add a test to verify this behavior.
Author: Álvaro Herrera <alvherre@kurilemu.de>
Reported-by: Giovanni Fabris <giovanni.fabris@icon.it>
Reported-by: Arthur Nascimento <tureba@gmail.com>
Backpatch-through: 13
Discussion: https://postgr.es/m/202511031106.h4fwyuyui6fz@alvherre.pgsql
Commit b4f584f9d2 (affecting v15~, later backpatched down to 13 as of
3635a0a35a) introduced an unconditional WAL receiver shutdown when
switching from streaming to archive WAL sources. This causes problems
during a timeline switch, when a WAL receiver enters WALRCV_WAITING
state but remains alive, waiting for instructions.
The unconditional shutdown can break some monitoring scenarios as the
WAL receiver gets repeatedly terminated and re-spawned, causing
pg_stat_wal_receiver.status to show a "streaming" instead of "waiting"
status, masking the fact that the WAL receiver is waiting for a new TLI
and a new LSN to be able to continue streaming.
This commit changes the WAL receiver behavior so as the shutdown becomes
conditional, with InstallXLogFileSegmentActive being always reset to
prevent the regression fixed by b4f584f9d2: only terminate the WAL
receiver when it is actively streaming (WALRCV_STREAMING,
WALRCV_STARTING, or WALRCV_RESTARTING). When in WALRCV_WAITING state,
just reset InstallXLogFileSegmentActive flag to allow archive
restoration without killing the process. WALRCV_STOPPED and
WALRCV_STOPPING are not reachable states in this code path. For the
latter, the startup process is the one in charge of setting
WALRCV_STOPPING via ShutdownWalRcv(), waiting for the WAL receiver to
reach a WALRCV_STOPPED state after switching walRcvState, so
WaitForWALToBecomeAvailable() cannot be reached while a WAL receiver is
in a WALRCV_STOPPING state.
A regression test is added to check that a WAL receiver is not stopped
on timeline jump, that fails when the fix of this commit is reverted.
Reported-by: Ryan Bird <ryanzxg@gmail.com>
Author: Xuneng Zhou <xunengzhou@gmail.com>
Reviewed-by: Noah Misch <noah@leadboat.com>
Reviewed-by: Michael Paquier <michael@paquier.xyz>
Discussion: https://postgr.es/m/19093-c4fff49a608f82a0@postgresql.org
Backpatch-through: 13
Add comments explaining when and where it is safe for nbtree to treat
row compare keys as if they were simple scalar inequality keys on the
row's most significant column. This is particularly important within
_bt_advance_array_keys, which deals with required inequality keys in a
general and uniform way, without any special handling for row compares.
Also spell out the implications of _bt_check_rowcompare's approach of
_conditionally_ evaluating lower-order row compare subkeys, particularly
when one of its lower-order subkeys might see NULL index tuple values
(these may or may not affect whether the qual as a whole is satisfied).
The behavior in this area isn't particularly intuitive, so these issues
seem worth going into.
In passing, add a few more defensive/documenting row comparison related
assertions to _bt_first and _bt_check_rowcompare.
Follow-up to commits bd3f59fd and ec986020.
Author: Peter Geoghegan <pg@bowt.ie>
Reviewed-By: Victor Yegorov <vyegorov@gmail.com>
Reviewed-By: Chao Li <li.evan.chao@gmail.com>
Discussion: https://postgr.es/m/CAH2-Wznwkak_K7pcAdv9uH8ZfNo8QO7+tHXOaCUddMeTfaCCFw@mail.gmail.com
Backpatch-through: 18
_bt_first reliably uses the same equality key (on each index column) for
initial positioning purposes as the one that _bt_checkkeys can use to
end the scan following commit f09816a0. _bt_first no longer applies its
own independent rules to determine which initial positioning key to use
on each column (for equality and inequality keys alike). Preprocessing
is now fully in control of determining which keys start and end each
scan, ensuring that _bt_first and _bt_checkkeys have symmetric behavior.
Remove obsolete comments that described why _bt_first was expected to
use at least one of the available required equality keys for initial
positioning purposes. The rules in this area are now maximally strict
and uniform, so there's no reason to draw attention to equality keys.
Any column with a required equality key cannot have a redundant required
inequality key (nor can it have a redundant required equality key).
Oversight in commit f09816a0, which removed similar comments from
_bt_first, but missed these comments.
Author: Peter Geoghegan <pg@bowt.ie>
Backpatch-through: 18
This commit reverts 818fefd8fd, that has been introduced to address a
an instability in some of the TAP tests due to the presence of random
standby snapshot WAL records, when slots are invalidated by
InvalidatePossiblyObsoleteSlot().
Anyway, this commit had also the consequence of introducing a behavior
regression. After 818fefd8fd, the code may determine that a slot needs
to be invalidated while it may not require one: the slot may have moved
from a conflicting state to a non-conflicting state between the moment
when the mutex is released and the moment when we recheck the slot, in
InvalidatePossiblyObsoleteSlot(). Hence, the invalidations may be more
aggressive than they actually have to.
105b2cb336 has tackled the test instability in a way that should be
hopefully sufficient for the buildfarm, even for slow members:
- In v18, the test relies on an injection point that bypasses the
creation of the random records generated for standby snapshots,
eliminating the random factor that impacted the test. This option was
not available when 818fefd8fd was discussed.
- In v16 and v17, the problem was bypassed by disallowing a slot to
become active in some of the scenarios tested.
While on it, this commit adds a comment to document that it is fine for
a recheck to use xmin and LSN values stored in the slot, without storing
and reusing them across multiple checks.
Reported-by: "suyu.cmj" <mengjuan.cmj@alibaba-inc.com>
Author: Bertrand Drouvot <bertranddrouvot.pg@gmail.com>
Reviewed-by: Masahiko Sawada <sawada.mshk@gmail.com>
Reviewed-by: Amit Kapila <amit.kapila16@gmail.com>
Discussion: https://postgr.es/m/f492465f-657e-49af-8317-987460cb68b0.mengjuan.cmj@alibaba-inc.com
Backpatch-through: 16
RIGHT_SEMI joins rely on the HEAP_TUPLE_HAS_MATCH flag to guarantee
that only the first match for each inner tuple is considered.
However, in a parallel hash join, the inner relation is stored in a
shared global hash table that can be probed by multiple workers
concurrently. This allows different workers to inspect and set the
match flags of the same inner tuples at the same time.
If two workers probe the same inner tuple concurrently, both may see
the match flag as unset and emit the same tuple, leading to duplicate
output rows and violating RIGHT_SEMI join semantics.
For now, we disable parallel plans for RIGHT_SEMI joins. In the long
term, it may be possible to support parallel execution by performing
atomic operations on the match flag, for example using a CAS or
similar mechanism.
Backpatch to v18, where RIGHT_SEMI join was introduced.
Bug: #19094
Reported-by: Lori Corbani <Lori.Corbani@jax.org>
Diagnosed-by: Tom Lane <tgl@sss.pgh.pa.us>
Author: Richard Guo <guofenglinux@gmail.com>
Reviewed-by: Tom Lane <tgl@sss.pgh.pa.us>
Discussion: https://postgr.es/m/19094-6ed410eb5b256abd@postgresql.org
Backpatch-through: 18
Because long is 32-bit on 64-bit Windows, it isn't a good datatype to
store the difference between 2 pointers. The under-sized type could
overflow and lead to scary warnings in MEMORY_CONTEXT_CHECKING builds,
such as:
WARNING: problem in alloc set ExecutorState: bad single-chunk %p in block %p
However, the problem lies only in the code running the check, not from
an actual memory accounting bug.
Fix by using "Size" instead of "long". This means using an unsigned
type rather than the previous signed type. If the block's freeptr was
corrupted, we'd still catch that if the unsigned type wrapped. Unsigned
allows us to avoid further needless complexities around comparing signed
and unsigned types.
Author: David Rowley <dgrowleyml@gmail.com>
Reviewed-by: Michael Paquier <michael@paquier.xyz>
Reviewed-by: Tom Lane <tgl@sss.pgh.pa.us>
Backpatch-through: 13
Discussion: https://postgr.es/m/CAApHDvo-RmiT4s33J=aC9C_-wPZjOXQ232V-EZFgKftSsNRi4w@mail.gmail.com
Some recent changes were made to remove the explicit dependency on
btree indexes in some parts of the code. One of these changes was
made in commit 9ef1851685, which allows non-btree indexes to be used
in get_actual_variable_range(). A follow-up commit ee1ae8b99f fixes
the cases where an index doesn’t have a sortopfamily as this is a
prerequisite to be used in get_actual_variable_range().
However, it was found that indexes that have amcanorder = true but do
not allow index-only-scans (amcanreturn returns false or is NULL) will
pass all of the conditions, while they should be rejected since
get_actual_variable_range() uses the index-only-scan machinery in
get_actual_variable_endpoint(). Such an index might cause errors like
ERROR: no data returned for index-only scan
during query planning.
The fix is to add a check in get_actual_variable_range() to reject
indexes that do not allow index-only scans.
Author: Maxime Schoemans <maxime.schoemans@enterprisedb.com>
Discussion: https://www.postgresql.org/message-id/flat/20ED852A-C2D9-41EB-8671-8C8B9D418BE9%40enterprisedb.com
Previously, the check_hook for synchronized_standby_slots attempted to
validate that each specified slot existed and was physical. However, these
checks were not performed during server startup. As a result, if users
configured non-existent slots before startup, the misconfiguration would
go undetected initially. This could later cause parallel query failures,
as newly launched workers would detect the issue and raise an ERROR.
This patch improves the check_hook by validating the syntax and format of
slot names. Validation of slot existence and type is deferred to the WAL
sender process, aligning with the behavior of the check_hook for
primary_slot_name.
Reported-by: Fabrice Chapuis <fabrice636861@gmail.com>
Author: Shlok Kyal <shlok.kyal.oss@gmail.com>
Reviewed-by: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Reviewed-by: Amit Kapila <amit.kapila16@gmail.com>
Reviewed-by: Ashutosh Sharma <ashu.coek88@gmail.com>
Reviewed-by: Rahila Syed <rahilasyed90@gmail.com>
Backpatch-through: 17, where it was introduced
Discussion: https://postgr.es/m/CAA5-nLCeO4MQzWipCXH58qf0arruiw0OeUc1+Q=Z=4GM+=v1NQ@mail.gmail.com
When dealing with ResultRelInfos for partitions, there are cases where
there are mixed requirements for the ri_RootResultRelInfo. There are
cases when the partition itself requires a NULL ri_RootResultRelInfo and
in the same query, the same partition may require a ResultRelInfo with
its parent set in ri_RootResultRelInfo. This could cause the column
mapping between the partitioned table and the partition not to be done
which could result in crashes if the column attnums didn't match
exactly.
The fix is simple. We now check that the ri_RootResultRelInfo matches
what the caller passed to ExecGetTriggerResultRel() and only return a
cached ResultRelInfo when the ri_RootResultRelInfo matches what the
caller wants, otherwise we'll make a new one.
Author: David Rowley <dgrowleyml@gmail.com>
Author: Amit Langote <amitlangote09@gmail.com>
Reported-by: Dmitry Fomin <fomin.list@gmail.com>
Discussion: https://postgr.es/m/7DCE78D7-0520-4207-822B-92F60AEA14B4@gmail.com
Backpatch-through: 15
These two functions expect there to be room to insert another item
in the FreePageBtree's array, but their assertions were too weak
to guarantee that. This has little practical effect granting that
the callers are not buggy, but it seems to be misleading late-model
Coverity into complaining about possible array overrun.
Author: Tom Lane <tgl@sss.pgh.pa.us>
Discussion: https://postgr.es/m/799984.1761150474@sss.pgh.pa.us
Backpatch-through: 13
Commit 883a95646a introduced overflow entries in the replication lag tracker
to fix an issue where lag columns in pg_stat_replication could stall when
the replay LSN stopped advancing.
This commit adds comments clarifying the purpose and behavior of overflow
entries to improve code readability and understanding.
Since commit 883a95646a was recently applied and backpatched to all
supported branches, this follow-up commit is also backpatched accordingly.
Author: Xuneng Zhou <xunengzhou@gmail.com>
Reviewed-by: Fujii Masao <masao.fujii@gmail.com>
Discussion: https://postgr.es/m/CABPTF7VxqQA_DePxyZ7Y8V+ErYyXkmwJ1P6NC+YC+cvxMipWKw@mail.gmail.com
Backpatch-through: 13
When JIT deformed tuples (controlled via the jit_tuple_deforming GUC),
types narrower than sizeof(Datum) would be zero-extended up to Datum
width. This wasn't the same as what fetch_att() does in the standard
tuple deforming code. Logically the values are the same when fetching
via the DatumGet*() marcos, but negative numbers are not the same in
binary form.
In the report, the problem was manifesting itself with:
ERROR: could not find memoization table entry
in a query which had a "Cache Mode: binary" Memoize node. However, it's
currently unclear what else is affected. Anything that uses
datum_image_eq() or datum_image_hash() on a Datum from a tuple deformed by
JIT could be affected, but it may not be limited to that.
The fix for this is simple: use signed extension instead of zero
extension.
Many thanks to Emmanuel Touzery for reporting this issue and providing
steps and backup which allowed the problem to easily be recreated.
Reported-by: Emmanuel Touzery <emmanuel.touzery@plandela.si>
Author: David Rowley <dgrowleyml@gmail.com>
Discussion: https://postgr.es/m/DB8P194MB08532256D5BAF894F241C06393F3A@DB8P194MB0853.EURP194.PROD.OUTLOOK.COM
Backpatch-through: 13
Previously, if primary_slot_name was set to an invalid slot name and
the configuration file was reloaded, both the postmaster and all other
backend processes reported a WARNING. With many processes running,
this could produce a flood of duplicate messages. The problem was that
the GUC check hook for primary_slot_name reported errors at WARNING
level via ereport().
This commit changes the check hook to use GUC_check_errdetail() and
GUC_check_errhint() for error reporting. As with other GUC parameters,
this causes non-postmaster processes to log the message at DEBUG3,
so by default, only the postmaster's message appears in the log file.
Backpatch to all supported versions.
Author: Fujii Masao <masao.fujii@gmail.com>
Reviewed-by: Chao Li <lic@highgo.com>
Reviewed-by: Amit Kapila <amit.kapila16@gmail.com>
Reviewed-by: Álvaro Herrera <alvherre@kurilemu.de>
Reviewed-by: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Discussion: https://postgr.es/m/CAHGQGwFud-cvthCTfusBfKHBS6Jj6kdAPTdLWKvP2qjUX6L_wA@mail.gmail.com
Backpatch-through: 13
Previously, when the replay LSN reported in feedback messages from a standby
stopped advancing, for example, due to a recovery conflict, the write_lag and
flush_lag columns in pg_stat_replication would initially update but then stop
progressing. This prevented users from correctly monitoring replication lag.
The problem occurred because when any LSN stopped updating, the lag tracker's
cyclic buffer became full (the write head reached the slowest read head).
In that state, the lag tracker could no longer compute round-trip lag values
correctly.
This commit fixes the issue by handling the slowest read entry (the one
causing the buffer to fill) as a separate overflow entry and freeing space
so the write and other read heads can continue advancing in the buffer.
As a result, write_lag and flush_lag now continue updating even if the reported
replay LSN remains stalled.
Backpatch to all supported versions.
Author: Fujii Masao <masao.fujii@gmail.com>
Reviewed-by: Chao Li <lic@highgo.com>
Reviewed-by: Shinya Kato <shinya11.kato@gmail.com>
Reviewed-by: Xuneng Zhou <xunengzhou@gmail.com>
Discussion: https://postgr.es/m/CAHGQGwGdGQ=1-X-71Caee-LREBUXSzyohkoQJd4yZZCMt24C0g@mail.gmail.com
Backpatch-through: 13
A BlockNumber (32-bit) might not be large enough to add bo_pagesPerRange
to when the table contains close to 2^32 pages. At worst, this could
result in a cancellable infinite loop during the BRIN index scan with
power-of-2 pagesPerRange, and slow (inefficient) BRIN index scans and
scanning of unneeded heap blocks for non power-of-2 pagesPerRange.
Backpatch to all supported versions.
Author: sunil s <sunilfeb26@gmail.com>
Reviewed-by: David Rowley <dgrowleyml@gmail.com>
Reviewed-by: Michael Paquier <michael@paquier.xyz>
Discussion: https://postgr.es/m/CAOG6S4-tGksTQhVzJM19NzLYAHusXsK2HmADPZzGQcfZABsvpA@mail.gmail.com
Backpatch-through: 13
67a54b9e8 taught the planner to push down HAVING clauses even when
grouping sets are present, as long as the clause does not reference
any columns that are nullable by the grouping sets. However, there
was an oversight: if any empty grouping sets are present, the
aggregation node can produce a row that did not come from the input,
and pushing down a HAVING clause in this case may cause us to fail to
filter out that row.
Currently, non-degenerate HAVING clauses are not pushed down when
empty grouping sets are present, since the empty grouping sets would
nullify the vars they reference. However, degenerate (variable-free)
HAVING clauses are not subject to this restriction and may be
incorrectly pushed down.
To fix, explicitly check for the presence of empty grouping sets and
retain degenerate clauses in HAVING when they are present. This
ensures that we don't emit a bogus aggregated row. A copy of each
such clause is also put in WHERE so that query_planner() can use it in
a gating Result node.
To facilitate this check, this patch expands the groupingSets tree of
the query to a flat list of grouping sets before applying the HAVING
pushdown optimization. This does not add any additional planning
overhead, since we need to do this expansion anyway.
In passing, make a small tweak to preprocess_grouping_sets() by
reordering its initial operations a bit.
Backpatch to v18, where this issue was introduced.
Reported-by: Yuhang Qiu <iamqyh@gmail.com>
Author: Richard Guo <guofenglinux@gmail.com>
Author: Tom Lane <tgl@sss.pgh.pa.us>
Discussion: https://postgr.es/m/0879D9C9-7FE2-4A20-9593-B23F7A0B5290@gmail.com
Backpatch-through: 18
This fixes an unlikely issue when fetching GROUPING SET results from
their internally stored hash tables. It was possible in rare cases that
the hash iterator would be set up incorrectly which could result in a
crash.
This was introduced in 4d143509c, so backpatch to v18.
Many thanks to Yuri Zamyatin for reporting and helping to debug this
issue.
Bug: #19078
Reported-by: Yuri Zamyatin <yuri@yrz.am>
Author: David Rowley <dgrowleyml@gmail.com>
Reviewed-by: Jeff Davis <pgsql@j-davis.com>
Discussion: https://postgr.es/m/19078-dfd62f840a2c0766@postgresql.org
Backpatch-through: 18
Commit a1b4f289be improved the hashjoin sizing to also consider the
memory used by BufFiles for batches. The code however had multiple
issues, making it ineffective or not working as expected in some cases.
* The amount of memory needed by buffers was calculated using uint32,
so it would overflow for nbatch >= 262144. If this happened the loop
would exit prematurely and the memory usage would not be reduced.
The nbatch overflow is fixed by reworking the condition to not use a
multiplication at all, so there's no risk of overflow. An explicit
cast was added to a similar calculation in ExecHashIncreaseBatchSize.
* The loop adjusting the nbatch value used hash_table_bytes to calculate
the old/new size, but then updated only space_allowed. The consequence
is the total memory usage was not reduced, but all the memory saved by
reducing the number of batches was used for the internal hash table.
This was fixed by using only space_allowed. This is also more correct,
because hash_table_bytes does not account for skew buckets.
* The code was also doubling multiple parameters (e.g. the number of
buckets for hash table), but was missing overflow protections.
The loop now checks for overflow, and terminates if needed. It'd be
possible to cap the value and continue the loop, but it's not worth
the complexity. And the overflow implies the in-memory hash table is
already very large anyway.
While at it, rework the comment explaining how the memory balancing
works, to make it more concise and easier to understand.
The initial nbatch overflow issue was reported by Vaibhav Jain. The
other issues were noticed by me and Melanie Plageman. Fix by me, with a
lot of review and feedback by Melanie.
Backpatch to 18, where the hashjoin memory balancing was introduced.
Reported-by: Vaibhav Jain <jainva@google.com>
Reviewed-by: Melanie Plageman <melanieplageman@gmail.com>
Backpatch-through: 18
Discussion: https://postgr.es/m/CABa-Az174YvfFq7rLS+VNKaQyg7inA2exvPWmPWqnEn6Ditr_Q@mail.gmail.com
EvalPlanQualStart() failed to propagate es_partition_directory into
the child EState used for EPQ rechecks. When execution time partition
pruning ran during the EPQ scan, executor code dereferenced a NULL
partition directory and crashed.
Previously, propagating es_partition_directory into the EPQ EState was
unnecessary because CreatePartitionPruneState(), which sets it on
demand, also initialized the exec-pruning context. After commit
d47cbf474, CreatePartitionPruneState() now initializes only the init-
time pruning context, leaving exec-pruning context initialization to
ExecInitNode(). Since EvalPlanQualStart() runs only ExecInitNode() and
not CreatePartitionPruneState(), it can encounter a NULL
es_partition_directory. Other executor fields initialized during
CreatePartitionPruneState() are already copied into the child EState
thanks to commit 8741e48e5d, but es_partition_directory was missed.
Fix by borrowing the parent estate's es_partition_directory in
EvalPlanQualStart(), and by clearing that field in EvalPlanQualEnd()
so the parent remains responsible for freeing the directory.
Add an isolation test permutation that triggers EPQ with execution-
time partition pruning, the case that reproduces this crash.
Bug: #19078
Reported-by: Yuri Zamyatin <yuri@yrz.am>
Diagnosed-by: David Rowley <dgrowleyml@gmail.com>
Author: David Rowley <dgrowleyml@gmail.com>
Co-authored-by: Amit Langote <amitlangote09@gmail.com>
Discussion: https://postgr.es/m/19078-dfd62f840a2c0766@postgresql.org
Backpatch-through: 18
Presently, these functions look up the relation's OID, lock it, and
then check privileges. Not only does this approach provide no
guarantee that the locked relation matches the arguments of the
lookup, but it also allows users to briefly lock relations for
which they do not have privileges, which might enable
denial-of-service attacks. This commit adjusts these functions to
use RangeVarGetRelidExtended(), which is purpose-built to avoid
both of these issues. The new RangeVarGetRelidCallback function is
somewhat complicated because it must handle both tables and
indexes, and for indexes, we must check privileges on the parent
table and lock it first. Also, it needs to handle a couple of
extremely unlikely race conditions involving concurrent OID reuse.
A downside of this change is that the coding doesn't allow for
locking indexes in AccessShare mode anymore; everything is locked
in ShareUpdateExclusive mode. Per discussion, the original choice
of lock levels was intended for a now defunct implementation that
used in-place updates, so we believe this change is okay.
Reviewed-by: Jeff Davis <pgsql@j-davis.com>
Discussion: https://postgr.es/m/Z8zwVmGzXyDdkAXj%40nathan
Backpatch-through: 18
log_error() would probably fail completely if used, and would
certainly print garbage for anything that needed to be interpolated
into the message, because it was failing to use the correct printing
subroutine for a va_list argument.
This bug likely went undetected because the error cases this code
is used for are rarely exercised - they only occur when Windows
security API calls fail catastrophically (out of memory, security
subsystem corruption, etc).
The FRONTEND variant can be fixed just by calling vfprintf()
instead of fprintf(). However, there was no va_list variant
of write_stderr(), so create one by refactoring that function.
Following the usual naming convention for such things, call
it vwrite_stderr().
Author: Bryan Green <dbryan.green@gmail.com>
Reviewed-by: Tom Lane <tgl@sss.pgh.pa.us>
Discussion: https://postgr.es/m/CAF+pBj8goe4fRmZ0V3Cs6eyWzYLvK+HvFLYEYWG=TzaM+tWPnw@mail.gmail.com
Backpatch-through: 13
Remove a variable that is no longer in use following commit 9a2e2a28.
It's not immediately clear why there were no compiler warnings about
this oversight.
Author: Peter Geoghegan <pg@bowt.ie>
Backpatch-through: 18
