Michael Paquier [Wed, 15 Jul 2026 01:03:46 +0000 (10:03 +0900)]
Include check on polpermissive relcache for policies
equalPolicy() is used in the relation cache to check if two policy
definitions are equivalent, but missed to check for polpermissive.
ALTER POLICY cannot switch a policy to be PERMISSIVE or RESTRICTIVE, so
this would need a dropped and then re-created policy, which would
trigger a relcache invalidation. Anyway, there is no harm in being
consistent in the check, and if one decides to add an ALTER POLICY to
switch PERMISSIVE or RESTRICTIVE, we would be silently in trouble.
Tomas Vondra [Sat, 11 Jul 2026 13:14:50 +0000 (15:14 +0200)]
Shorten pg_attribute_always_inline to pg_always_inline
The pg_attribute_always_inline macro name is so long it forces pgindent
to format the code in strange ways. Which may incentivize patch authors
to either structure the code in strange ways (e.g. reorder prototypes),
use shorter names, etc. Neither is very desirable for code readability.
This shortens the name by removing the _attribute_ part. It also makes
it more consistent with pg_noinline, which does not have the _attribute_
part either.
Backpatched to all supported branches, to prevent conflicts when
backpatching other fixes. The backbranches however keep both the old and
new macro name, so that existing code keeps working.
Author: Andres Freund <andres@anarazel.de> Reviewed-by: Peter Geoghegan <pg@bowt.ie> Reviewed-by: Tomas Vondra <tomas@vondra.me>
Discussion: https://postgr.es/m/bqqdehahpoa36igpictuqyn2s2mexk3t3ehidh2ffd2slb35e5@rzgksuiszgbg
Backpatch-through: 14
libpq: Make error checks in the new buffer draining code more robust
Check explicitly for pqsecure_read() returning an error. It shouldn't
fail, and we would've caught it in the check for a short read, but
better to be explicit so that the error message is more informative.
We also shouldn't update 'inEnd' when the read fails, although that
too is just pro forma as we will bail out and close the connection on
error.
Reported-by: Peter Eisentraut <peter@eisentraut.org>
Discussion: https://www.postgresql.org/message-id/34844e8c-267c-4daf-b1e0-f26059a4a7d3@eisentraut.org
Backpatch-through: 14
ssl: Include limits.h to get INT_MAX when using LibreSSL
When compiling against OpenSSL, the <limits.h> header is indirectly
included via openssl/ossl_typ.h from openssl/conf.h, but the LibreSSL
version of ossl_typ.h does not include <limits.h> which cause compiler
failure due to missing symbol (since ffd080d94fe). Fix by explicitly
including <limits.h>.
Author: Daniel Gustafsson <dgustafsson@postgresql.org>
Discussion: https://www.postgresql.org/message-id/6A9E7815-BD5A-4C31-A515-48159823406B@yesql.se
Backpatch-through: 14
doc: Clarify COPY FROM WHERE expression restrictions
Commit aa606b9316a disallowed generated columns in COPY FROM WHERE
expressions, and commit 21c69dc73f9 disallowed system columns.
However, the COPY reference page still mentions only the restriction
on subqueries.
Update the documentation to also list generated columns and system
columns as unsupported in COPY FROM WHERE expressions.
Backpatch the generated-column documentation change to all supported
versions. Backpatch the system-column documentation change to v19,
where that restriction was introduced.
Commit dcb00495236 accidentally changed the final expanded query's
condition to > 2 while rewriting the example into SQL operator notation.
The original query and the preceding rewritten forms all use >= 2,
and view expansion should preserve that qualification. This commit
changes the final condition from > 2 to >= 2.
libpq: Drain all pending bytes from SSL/GSS during pqReadData()
The previous commit strengthened a workaround for a hang when large
messages are split across TLS records/GSS tokens. Because that
workaround is implemented in libpq internals, it can only help us when
libpq itself is polling on the socket. In nonblocking situations,
where the client above libpq is expected to poll, the same bugs can
show up.
As a contrived example, consider a large protocol-2.0 error coming
back from a server during PQconnectPoll(), split in an odd way across
two records:
-- TLS record (8192-byte payload) --
EEEE[...repeated a total of 8192 times]
-- TLS record (8193-byte payload) --
EEEE[...repeated a total of 8192 times]\0
The first record will fill the first half of the libpq receive buffer,
which is 16k long by default. The second record completely fills the
last half with its first 8192 bytes, leaving the terminating NULL in
the OpenSSL buffer. Since we still haven't seen the terminator at our
level, PQconnectPoll() will return PGRES_POLLING_READING, expecting to
come back when the server has sent "the rest" of the data. But there
is nothing left to read from the socket; OpenSSL had to pull all of
the data in the 8193-byte record off of the wire to decrypt it.
A real server would probably not split up the records this way, nor
keep the connection open after sending a fatal connection error. But
servers that regularly use larger TLS records can get the libpq
receive buffer into the same state if DataRows are big enough, as
reported on the list. While the PostgreSQL server doesn't use larger
TLS records like that, other non-PostgreSQL servers that implement the
wire protocol are known to do that, as well as proxies that sit
between the server and the client
This is a layering violation. libpq makes decisions based on data in
the application buffer, above the transport buffer (whether SSL or
GSS), but clients are polling the socket below the transport buffer.
One way to fix this in a backportable way, without changing APIs too
much, is to ensure data never stays in the transport buffer. Then
pqReadData's postconditions will look similar for both raw sockets and
SSL/GSS: any available data is either in the application buffer, or
still on the socket.
Building on the prior commit, make pqReadData() to drain all pending
data from the transport layer into conn->inBuffer, expanding the
buffer as necessary. This is not particularly efficient from an
architectural perspective (the pqsecure_read() implementations take
care to fit their packets into the current buffer, and that effort is
now completely discarded), but it's hopefully easier to reason about
than a full rewrite would be for the back branches.
Author: Jacob Champion <jacob.champion@enterprisedb.com> Reviewed-by: Mark Dilger <mark.dilger@enterprisedb.com> Reviewed-by: solai v <solai.cdac@gmail.com> Reported-by: Lars Kanis <lars@greiz-reinsdorf.de>
Discussion: https://postgr.es/m/2039ac58-d3e0-434b-ac1a-2a987f3b4cb1%40greiz-reinsdorf.de
Backpatch-through: 14
libpq: Extend "read pending" check from SSL to GSS
An extra check for pending bytes in the SSL layer has been part of
pqReadReady() for a very long time (79ff2e96d). But when GSS transport
encryption was added, it didn't receive the same treatment. (As 79ff2e96d notes, "The bug that I fixed in this patch is exceptionally
hard to reproduce reliably.")
Without that check, it's possible to hit a hang in gssencmode, if the
server splits a large libpq message such that the final message in a
streamed response is part of the same wrapped token as the split
message:
If the split message takes up enough memory to nearly fill libpq's
receive buffer, libpq may return from pqReadData() before the later
messages are pulled out of the PqGSSRecvBuffer. Without additional
socket activity from the server, pqReadReady() (via pqSocketCheck())
will never again return true, hanging the connection.
Pull the pending-bytes check into the pqsecure API layer, where both
SSL and GSS now implement it.
Note that this does not fix the root problem! Third party clients of
libpq have no way to call pqsecure_read_is_pending() in their own
polling. This just brings the GSS implementation up to par with the
existing SSL workaround; a broader fix is left to a subsequent commit.
In preparation for the broader fix, this patch already changes the
*_read_pending() functions to return the number of bytes in the buffer
rather than just a boolean. The current callers don't need that, but
the subsequent fix will.
Author: Jacob Champion <jacob.champion@enterprisedb.com>
Discussion: https://postgr.es/m/CAOYmi%2BmpymrgZ76Jre2dx_PwRniS9YZojwH0rZnTuiGHCsj0rA%40mail.gmail.com
Backpatch-through: 14
pg_dump: check for _beginthreadex() failure in parallel dump
ParallelBackupStart() stored _beginthreadex()'s return value as the
worker's thread handle without checking it. On failure that value is 0,
which would later reach WaitForMultipleObjects() as a null handle, caught
only by an Assert. The fork() path already calls pg_fatal() when it
fails; do the same for _beginthreadex(), as pgbench does.
Author: Bryan Green <dbryan.green@gmail.com>
Discussion: https://www.postgresql.org/message-id/8c712d76-ecf7-4749-a6d8-dddc01f298ec@gmail.com
Backpatch-through: 14
Robert Haas [Mon, 6 Jul 2026 16:12:41 +0000 (12:12 -0400)]
Prevent satisfies_hash_partition from crashing with VARIADIC NULL.
Commit f3b0897a1213f46b4d3a99a7f8ef3a4b32e03572 fixed some
related problems, but overlooked this one. That commit first
appeared in PostgreSQL 11, so back-patch to all supported branches.
Backpatch-through: 14
Discussion: http://postgr.es/m/CA+TgmobsvQw3F+KRYT83=N3teh8D2t-oPR=U06QDZJE3viCJRg@mail.gmail.com Reviewed-by: Tender Wang <tndrwang@gmail.com> Reviewed-by: Ewan Young <kdbase.hack@gmail.com>
Tom Lane [Sat, 4 Jul 2026 15:34:26 +0000 (11:34 -0400)]
Disallow renaming a rule to "_RETURN".
ON SELECT rules must be named "_RETURN", while other kinds of rules
must not be; this ancient restriction is depended on by various client
code. We successfully enforced this convention in most places, but
ALTER RULE allowed renaming a non-SELECT rule to "_RETURN". Notably,
that would break dump/restore, since the eventual CREATE RULE command
would reject the name.
While at it, remove DefineQueryRewrite's hack to substitute "_RETURN"
for the convention that was used before 7.3. We dropped other
server-side code that supported restoring pre-7.3 dumps some time ago
(notably in e58a59975 and nearby commits), but this bit was missed.
Bug: #19543 Reported-by: Adam Pickering <adamkpickering@gmail.com>
Author: Tom Lane <tgl@sss.pgh.pa.us>
Discussion: https://postgr.es/m/19543-461228e77f3b32fc@postgresql.org
Backpatch-through: 14
Tom Lane [Fri, 3 Jul 2026 17:50:14 +0000 (13:50 -0400)]
Fix btree_gist's NotEqual strategy on internal index pages.
gbt_var_consistent() handled the <> (BtreeGistNotEqual) strategy without
distinguishing leaf from internal pages, unlike every other strategy.
In particular, it tried to apply the datatype-specific f_eq method,
which is completely wrong since internal keys might not have the same
representation as leaf keys. This led to OOB reads and potentially
crashes, and most likely to wrong query results as well.
On leaf pages we can apply the inverse of what the Equal strategy does.
On internal pages, use a correct implementation of what the previous
code intended: we can descend if the query value equals both bounds,
*so long as the bounds aren't truncated*. With truncated bounds we
don't quite know the range of what's below, so we must always descend.
Adjust the code in gbt_num_consistent() to look similar, too. This
fixes a performance buglet in that there's no need to do two comparisons
on a leaf entry, but the main point is just to keep code consistency.
Reported-by: 王跃林 <violin0613@tju.edu.cn>
Author: Ayush Tiwari <ayushtiwari.slg01@gmail.com> Reviewed-by: Tom Lane <tgl@sss.pgh.pa.us>
Discussion: https://postgr.es/m/AH*AvQCYKhQGVvPWi1GiU4oY.8.1781609375063.Hmail.3020001251@tju.edu.cn
Backpatch-through: 14
Remove replication slot advice from MultiXact wraparound hints
Previously, MultiXactId wraparound hints suggested dropping stale
replication slots. While that advice is appropriate for transaction ID
wraparound, where replication slots can hold back XID horizons,
it was misleading for MultiXactId wraparound. Following it could lead
users to drop replication slots unnecessarily without helping resolve
the MultiXactId wraparound condition.
MultiXact cleanup is not directly delayed by replication slots.
Instead, it depends on whether old MultiXactIds can still be seen
as live by running transactions.
This commit removes the replication slot advice from MultiXactId
wraparound hints, and documents that stale replication slots are
normally not relevant to resolving MultiXactId wraparound problems.
Tom Lane [Wed, 1 Jul 2026 17:27:22 +0000 (13:27 -0400)]
btree_gist: fix NaN handling in float4/float8 opclasses.
The float4 and float8 btree_gist opclasses compared keys with raw C
operators (==, <, >). IEEE 754 makes every comparison involving NaN
false, so GiST disagreed with the regular float comparison operators
and with the btree opclass, which uses float[4|8]_cmp_internal()
(so that all NaNs are equal and NaN sorts after every non-NaN value).
In addition, the penalty and distance functions were not careful
about NaNs, and the penalty functions could also misbehave for IEEE
infinities. Wrong answers from the penalty functions would probably
do no more than make the index non-optimal, but the distance mistakes
were visible from SQL.
To fix, make the comparison functions rely on the same NaN-aware
comparison functions the core code uses, and rewrite the penalty
and distance functions to follow the rules that NaNs are equal
but maximally far away from non-NaNs. The penalty_num() code was
formerly shared between integral and float cases, but I chose to make
two copies so that the integral cases are not saddled with the extra
logic for NaNs and infinities/overflows. I also rewrote it as static
inline functions instead of an unreadable and uncommented macro.
The float penalty functions were previously unreached by the
regression tests, so add new test cases to exercise them.
There's no on-disk format change, but users who have NaN entries
in a btree_gist index would be well advised to reindex it.
Bug: #19501
Bug: #19524 Reported-by: Man Zeng <zengman@halodbtech.com> Reported-by: Yuelin Wang <3020001251@tju.edu.cn>
Author: Bill Kim <billkimjh@gmail.com> Co-authored-by: Tom Lane <tgl@sss.pgh.pa.us>
Discussion: https://postgr.es/m/19501-3bff3bbc97f1e7c9@postgresql.org
Discussion: https://postgr.es/m/19524-9559d302c8455664@postgresql.org
Discussion: https://postgr.es/m/CAMQXxcgbtD2LXfX0tpgvOizxP-XxrCHV2ZDy4By_TZnJMsxXWQ@mail.gmail.com
Backpatch-through: 14
Richard Guo [Mon, 29 Jun 2026 05:21:15 +0000 (14:21 +0900)]
Use plpythonu in plpython tests added by commit 0b7719f74
Commit 0b7719f74 added regression tests that spell the language name
as plpython3u. That works on the master branch, but on v14 the
plpython tests must use the unversioned name plpythonu: the plpython3u
extension is built only in Python 3 builds, and for those builds
regress-python3-mangle.mk rewrites plpythonu to plpython3u in the test
files. In a Python 2 build the new tests instead failed with
"language \"plpython3u\" does not exist".
Spell the language as plpythonu in the new tests, matching every other
plpython test in this branch, so they work in both Python 2 and Python
3 builds. This is needed only in v14; later branches no longer
support Python 2 and have dropped the mangling step.
Richard Guo [Mon, 29 Jun 2026 02:44:35 +0000 (11:44 +0900)]
plpython: Fix NULL pointer dereferences for broken sequence and mapping objects
PL/Python and its hstore and jsonb transforms build SQL values from
Python containers by calling Python C API functions that can return
NULL, and in several places the result was used without first checking
it.
On the sequence side, PySequence_GetItem() is used when converting a
returned sequence into a SQL array or composite value, when reading
the argument list passed to plpy.execute() or plpy.cursor(), and when
reading the list of type names given to plpy.prepare(). On the
mapping side, the hstore and jsonb transforms call PyMapping_Size()
and PyMapping_Items() and then index the result with PyList_GetItem()
and PyTuple_GetItem().
All of these return NULL (or -1), with a Python exception set, for a
broken object: for example one whose __getitem__() or items() raises,
or which reports a length that disagrees with what it actually yields.
The unchecked result was then dereferenced, crashing the backend.
Fix this by checking the result of each call and reporting a regular
error if it failed, so that the underlying Python exception is
surfaced instead of taking down the session.
Richard Guo [Wed, 24 Jun 2026 00:09:48 +0000 (09:09 +0900)]
plperl: Fix NULL pointer dereference for forged array object
In get_perl_array_ref(), for a PostgreSQL::InServer::ARRAY object, we
look up its "array" key with hv_fetch_string() and then inspect the
returned SV. However, hv_fetch_string() returns a NULL pointer when
the key is absent, and the code dereferenced that result without first
checking whether the pointer itself was NULL. As a result, a plperl
function returning a forged PostgreSQL::InServer::ARRAY object that
lacks the "array" key would crash the backend with a segmentation
fault.
Fix this by checking the pointer returned by hv_fetch_string() before
dereferencing it, matching how other callers in this file already
guard the result. With the check in place, such an object falls
through to the existing error report instead of crashing.
Michael Paquier [Tue, 23 Jun 2026 07:52:21 +0000 (16:52 +0900)]
doc: Describe better handling of indexes in ALTER TABLE ATTACH PARTITION
When ALTER TABLE ... ATTACH PARTITION matches partition indexes to the
parent table's indexes, invalid indexes are skipped. This commit
improves the documentation to describe what e90e9275f56 has changed:
invalid indexes are skipped, and only valid indexes are considered for a
match.
Author: Mohamed Ali <moali.pg@gmail.com> Reviewed-by: Sami Imseih <samimseih@gmail.com>
Discussion: https://postgr.es/m/CAGnOmWpAMaE-BOkpwM6mJnHcpS2QZ8yLSSaqmz+vryEsbCWWWA@mail.gmail.com
Backpatch-through: 14
Tom Lane [Fri, 19 Jun 2026 16:52:00 +0000 (12:52 -0400)]
Make pg_mkdir_p() tolerant of a concurrent directory creation.
pg_mkdir_p creates each missing path component with a stat() followed
by mkdir(). If the stat() reports the component as absent but another
process creates it in the window before this process's mkdir(), mkdir()
fails with EEXIST and pg_mkdir_p treated that as a hard error -- unlike
"mkdir -p", which is meant to be idempotent and race-tolerant.
This shows up when several processes concurrently create paths that
share an ancestor directory: for example, parallel initdb runs whose
data directories live under a common temporary directory. One process
wins the race to create the shared ancestor and the others fail with
could not create directory "...": File exists
Fix this race condition by first trying mkdir() and only attempting
stat() if it fails with EEXIST.
On Windows, there's an additional problem: stat() opens a file handle
and participates in share-mode locking, which means it can transiently
fail on a directory another process is concurrently creating. Use
GetFileAttributes() instead: it requests only FILE_READ_ATTRIBUTES
and is exempt from share-mode denial, so it reliably sees a
concurrently-created directory.
I (tgl) also chose to back-patch 039f7ee0f's effects on this function,
so that pgmkdirp.c remains identical in all live branches.
Author: Andrew Dunstan <andrew@dunslane.net> Co-authored-by: Tom Lane <tgl@sss.pgh.pa.us>
Discussion: https://postgr.es/m/3ca004de-e49b-4471-b8aa-fd656e70f68c@dunslane.net
Backpatch-through: 14
Nathan Bossart [Thu, 18 Jun 2026 16:29:49 +0000 (11:29 -0500)]
Silence "may be used uninitialized" compiler warning.
Newer gcc warns that this "actual_arg_types" variable may be used
uninitialized, but visual inspection indicates there's no bug. To
silence the warning, initialize the variable to zeros.
Bug: #19485 Reported-by: Hans Buschmann <buschmann@nidsa.net> Tested-by: Erik Rijkers <er@xs4all.nl> Tested-by: Hans Buschmann <buschmann@nidsa.net> Reviewed-by: Tristan Partin <tristan@partin.io> Reviewed-by: Álvaro Herrera <alvherre@kurilemu.de>
Discussion: https://postgr.es/m/19485-2b03231a775756f1%40postgresql.org
Discussion: https://postgr.es/m/6c52a1a6612948519468d46cb224a8c4%40nidsa.net
Tom Lane [Thu, 18 Jun 2026 16:22:55 +0000 (12:22 -0400)]
hstore_plperl: Add CHECK_FOR_INTERRUPTS() in reference-unwinding loop.
Add CHECK_FOR_INTERRUPTS() to the while loop in plperl_to_hstore()
that dereferences chains of Perl references, so that a circular
reference (e.g. $x = \$x) can be cancelled by the user instead of
spinning indefinitely. (We looked at detecting such circular
references, but it seems more trouble than it's worth.)
This is a follow-up to da82fbb8f, which fixed the same issue in
SV_to_JsonbValue() in jsonb_plperl.
Author: Aleksander Alekseev <aleksander@tigerdata.com> Reviewed-by: Tom Lane <tgl@sss.pgh.pa.us>
Discussion: https://postgr.es/m/CAJ7c6TPbjkzUk4qJ5dHvDNEz0hBuFue3A-XWz_=897z+BC+z8A@mail.gmail.com
Backpatch-through: 14
Nathan Bossart [Thu, 18 Jun 2026 14:31:27 +0000 (09:31 -0500)]
doc: Fix "Prev" link, take 2.
Commit 6678b58d78 fixed a wrong "Prev" link by changing the link
generation code to use [position()=last()] instead of [last()] in
the predicate on the union of reverse axes. Unfortunately, that
caused documentation builds to take much longer. To fix, combine
the "preceding" and "ancestor" steps into one "preceding" step and
one "ancestor" step, and revert the predicate back to [last()].
The smaller union evades the libxml2 bug while avoiding the build
time regression.
Reported-by: Tom Lane <tgl@sss.pgh.pa.us> Tested-by: Tom Lane <tgl@sss.pgh.pa.us>
Discussion: https://postgr.es/m/1132496.1781718007%40sss.pgh.pa.us
Backpatch-through: 14
Amit Langote [Thu, 18 Jun 2026 06:51:38 +0000 (15:51 +0900)]
Report undefined jsonpath variable when no variables are supplied
The two-argument jsonb @? and @@ operators invoke the jsonpath executor
with no variable set. In that case getJsonPathVariable() treated any
"$name" reference as JSON null and continued evaluating, instead of
reporting the variable as undefined.
This produced incorrect results -- for example '42'::jsonb @? '$"x"'
returned true -- and, for some malformed or hostile jsonpath expressions
with deeply nested predicates, allowed essentially unbounded memory
consumption that could get the backend killed by the OOM killer.
Report the undefined variable as an error in this case as well, reusing
the message already emitted when a variable is not found among supplied
variables. This matches the behavior of v17 and later, where the
jsonpath executor was reorganized. Stopping at the first undefined
variable reference also resolves the reported memory-growth case.
Note this is a user-visible change in the back branches: a jsonpath
expression that references a variable while no variables are supplied now
raises an error rather than silently evaluating it as NULL. The previous
behavior was incorrect, so the change is judged worthwhile.
Tom Lane [Wed, 17 Jun 2026 15:04:41 +0000 (11:04 -0400)]
jsonb_plperl, jsonb_plpython: Fix unguarded recursion and loops.
Add check_stack_depth() to Jsonb_to_SV, SV_to_JsonbValue,
PLyObject_FromJsonbContainer, and PLyObject_ToJsonbValue. Without
this, deeply nested JSONB values can crash the backend with SIGSEGV
instead of raising a proper error.
Also add CHECK_FOR_INTERRUPTS() to the while loop in SV_to_JsonbValue
that dereferences chains of Perl references, so that a circular
reference (e.g. $x = \$x) can be cancelled by the user instead of
spinning indefinitely. (We looked at detecting such circular
references, but it seems more trouble than it's worth.)
Author: Aleksander Alekseev <aleksander@tigerdata.com> Reviewed-by: Tom Lane <tgl@sss.pgh.pa.us>
Discussion: https://postgr.es/m/CAJ7c6TPbjkzUk4qJ5dHvDNEz0hBuFue3A-XWz_=897z+BC+z8A@mail.gmail.com
Backpatch-through: 14
Michael Paquier [Tue, 16 Jun 2026 23:42:13 +0000 (08:42 +0900)]
Fix another instability in recovery TAP test 004_timeline_switch
The test did not wait for the standby to be connected to the primary.
This breaks one assumption at the beginning of the test, where the
primary is stopped to ensure that all its records are flushed to both
standbys before moving on with its next steps.
If standby_1 finishes ahead of standby_2, the test would be able work
fine as the former waits for the latter. The opposite is not true,
standby_2 getting ahead of standby_1 would cause the test to fail on
timeout when standby_1 attempts to connect to standby_2.
This commit adds an additional polling query after the two standbys are
started, checking that both standbys are connected to the primary before
processing with the initial steps of the test.
The error path in ReorderBufferProcessTXN was not freeing
(reorderbuffer.c's representation of) a speculative insertion record
correctly. In assert-enabled builds, this leads to an assertion
failure. In production builds, I see no effect; there may be a small
transient leak, but in an improbable code path such as this, such a leak
is not of any significance. For users running with assertions enabled,
the crash is annoying.
Fix by having ReorderBufferProcessTXN() free the speculative insert
ahead of freeing the rest of the transaction, and no longer try to
handle that insert as a separate argument to ReorderBufferResetTXN().
This code came in with commit 7259736a6e5b (14-era). Backpatch all the
way back.
In branches 14-16, also backpatch the assertion that originally fails in
the problem scenario, which was added by dbed2e36625d (originally
backpatched to 17), that at the end of ReorderBufferReturnTXN() the
in-memory size of the transaction is zero.
The expression (len_diff * 10 * (an + 1)) used as the return value of
ltree_compare() is computed at int32 width. With LTREE_MAX_LEVELS =
65535, the product can exceed INT32_MAX once an ltree has more than
~14,653 levels, which causes the result to wrap and invert its sign.
That corrupts btree ordering as well as the "magnitude" consumed by
ltree_penalty() for GiST page splits.
To fix, split ltree_compare() into two functions. The new
ltree_compare_distance() function returns a float, which won't
overflow. It's used by the ltree_penalty() caller. All the other
callers only care about the sign of the return value, i.e. which of
the arguments is greater, so change ltree_compare() to not multiply
the result with (10 * (an + 1)), which avoids the overflow for those
callers.
Existing btree or GiST indexes on ltree columns containing values with
more than ~14,653 levels may be corrupt and should be REINDEXed.
Add a regression test based on the reporter's PoC.
Tom Lane [Mon, 15 Jun 2026 19:35:37 +0000 (15:35 -0400)]
Clean up quoting of variable strings within replication commands.
Our handling of quoting within replication commands was pretty
sloppy, typically looking like
appendStringInfo(&cmd, " SLOT \"%s\"", options->slotname);
This is fine as long as options->slotname doesn't contain a double
quote mark, but what if it does? In principle this'd allow injection
of harmful options into replication commands, in the probably-unlikely
case that a slot name comes from untrustworthy input. We ought to
clean that up.
Moreover, even the places that were trying to be more careful
generally got it wrong, because they used quoting subroutines
intended for SQL commands rather than something that will work
with the replication-command scanner repl_scanner.l. For example,
several places naively use PQescapeLiteral() to quote option values
for replication commands. If the string contains a backslash,
PQescapeLiteral() will produce E'...' literal syntax, which
repl_scanner.l doesn't recognize. Another near miss was to use
quote_identifier() to quote identifiers. That function won't quote
valid lowercase identifiers unless they match SQL keywords ... but in
this context, replication keywords are what matter. Neither of these
errors seem to risk string injection, but they definitely can cause
syntax errors in replication commands that ought to be valid.
We can clean all this up by using simple quoting logic that just
doubles single or double quotes respectively.
Or at least, we could if repl_scanner.l handled doubled double quotes
in identifiers, but for some reason it doesn't! So the first step in
this fix has to be to fix that. (The fact that we'll later reject
slot names containing double quotes is very far short of justifying
this omission.)
Having done that, this patch runs around and applies correct
quoting in all places that generate replication commands containing
strings coming from outside the immediate context. Probably some
of these places are safe because of restrictions elsewhere, but it
seems best to just quote all the time.
This was originally reported as a security bug, which it could be
if replication slot names or parameters were to originate from
untrustworthy sources. But the security team concluded that that
was a very improbable situation, so we're just going to fix this
as a regular bug.
Reported-by: Team Dhiutsa
Author: Tom Lane <tgl@sss.pgh.pa.us> Reviewed-by: Ayush Tiwari <ayushtiwari.slg01@gmail.com>
Discussion: https://postgr.es/m/1648659.1781287310@sss.pgh.pa.us
Backpatch-through: 14
Nathan Bossart [Mon, 15 Jun 2026 17:16:38 +0000 (12:16 -0500)]
doc: Fix "Prev" link.
Presently, the "Prev" link on the page for background workers sends
you to the middle of the previous chapter instead of the actual
previous page. This appears to be caused by a libxml2 bug, but
regardless, a minimal fix is to change the link generation code to
use [position()=last()] instead of [last()] in the predicate on the
union of reverse axes.
If a query has more than 7498 params, the ParameterDescription message
exceeds the 30000 byte limit on messages that are not specifically
marked as possibly being longer than that (VALID_LONG_MESSAGE_TYPE).
To fix, add ParameterDescription to the list.
Author: Ning Sun <classicning@gmail.com>
Discussion: https://www.postgresql.org/message-id/dbfb4b65-0aa8-470a-8b87-b6496160b28a@gmail.com
Backpatch-through: 14
Michael Paquier [Mon, 15 Jun 2026 02:38:02 +0000 (11:38 +0900)]
Trim regression test expected output for xml
This commit reduces the number of expected output files for the "xml"
test from three to two (well, mostly one, see below for details).
xml_2.out existed to handle some differences in output due to libxml2
2.9.3, due to some error context missing (085423e3e326). This file is
removed, by tweaking the XML inputs to trigger the same error patterns
for the problematic 2.9.3 and other libxml2 versions. This part is
authored by Tom Lane.
xml_1.out (no libxml2 support) is reduced in size by adding an \if query
that exits the test early. This still checks NO_XML_SUPPORT() through
xmlin(). The rest of the test is skipped if XML input cannot be
handled by the backend. This part has been written by me.
Author: Tom Lane <tgl@sss.pgh.pa.us>
Author: Michael Paquier <michael@paquier.xyz> Reviewed-by: Daniel Gustafsson <daniel@yesql.se>
Discussion: https://postgr.es/m/aiu6CXO67q-s70n5@paquier.xyz
Backpatch-through: 14
Tom Lane [Sun, 14 Jun 2026 15:01:48 +0000 (11:01 -0400)]
Doc: remove stale entry for removed aclitem[] ~ aclitem operator.
Commit 2f70fdb06 removed the deprecated containment operator
~(aclitem[],aclitem) from the catalogs, but missed removing its entry
from the documentation. (Arguably the blame should fall on c62dd80cd,
which added this entry in contravention of the longstanding policy
that we don't document deprecated aliases in the first place.)
Author: Shinya Kato <shinya11.kato@gmail.com> Reviewed-by: Tom Lane <tgl@sss.pgh.pa.us>
Discussion: https://postgr.es/m/CAOzEurQSyR5psWukyhUz1LtxyO55C2Vfp0Fmt8w2jGKxhszQmQ@mail.gmail.com
Backpatch-through: 14
amcheck: Use correct varlena size accessor in bt_normalize_tuple()
bt_normalize_tuple() uses VARSIZE() to get the size of varlena, even though
it's not yet known, that it has a 4-byte header. Fix this by replacing a
accessor with a universal VARSIZE_ANY().
Backpatch to all supported versions.
Reported-by: Andres Freund <andres@anarazel.de>
Discussion: https://postgr.es/m/7ckc7oka4bvafkf5bwlqs6ygrhlsbhz25ppozfch7zbuxcx3rf%40e4pr4oqenalc
Author: Andrey Borodin <x4mmm@yandex-team.ru> Reviewed-by: Alexander Korotkov <aekorotkov@gmail.com>
Backpatch-through: 14
Andrew Dunstan [Fri, 12 Jun 2026 22:05:25 +0000 (18:05 -0400)]
Adjust cross-version upgrade tests for seg_out() fix
Commit 0e1f1ed157e taught seg_out() to print the certainty indicator
on an interval's upper boundary, but it was back-patched only as far
as v14. When upgrading from an older release, the old server prints
the one test_seg row exercising that case ('4.6 .. ~7.0') without the
indicator, so the pre- and post-upgrade dumps do not match. Make
AdjustUpgrade.pm delete just that row; seg's comparison function does
distinguish the certainty indicators, so the otherwise identical row
'4.6 .. 7.0' is unaffected.
Andrew Dunstan [Fri, 12 Jun 2026 14:20:34 +0000 (10:20 -0400)]
Don't try to import a non-exported object in vcregress.pl
Commit ca9e9b08e453 wrongly tried to import devnull from File::Spec, but
it's not exported, you just call the method via the class. This was
harmless until modern perls complained, so stop doing that.
OpenSSL 4.0.0 changed some parameters and returnvalues to const, so
we need to update our declarations and subsequently cast away const-
ness from a few callsites to make libpq build without warnings. This
is tested with OpenSSL 1.1.1 through 4.0.0 as well as with LibreSSL.
No functional change is introduced, this commit only allows postgres
to be compiled against OpenSSL 4.0.0 without warnings.
There is also an errormessage change in OpenSSL 4.0.0 which needed
to be covered by our testharness.
This will be backpatched to all supported branches since they are
all equally likely to be built against OpenSSL 4.0.0 as it becomes
available in distributions. Backpatching will be done once it has
been in master for a few days without issues.
Author: Daniel Gustafsson <daniel@yesql.se> Reviewed-by: Michael Paquier <michael@paquier.xyz>
Discussion: https://postgr.es/m/066B07BB-85FA-487C-BE8C-40F791CFC3C4@yesql.se
Backpatch-through: 14
Fujii Masao [Fri, 12 Jun 2026 02:08:33 +0000 (11:08 +0900)]
doc: fix reference for finding replication slots to drop
Commit a70bce43fb added instructions on how to recover if PostgreSQL
refuses to issue new transaction IDs because of imminent wraparound,
but when describing how to find replication slots that should be dropped,
it referred to pg_stat_replication where it should have referenced
pg_replication_slots.
In passing, decorate references to views with <structname> tags.
Michael Paquier [Fri, 12 Jun 2026 01:25:59 +0000 (10:25 +0900)]
Fix handling of namespace nodes in xpath() (xml)
xpath() attempted to call xmlCopyNode() and xmlNodeDump() on a
XML_NAMESPACE_DECL, finishing with a confusing error:
=# SELECT xpath('//namespace::foo', '<root xmlns:foo="http://127.0.0.1"/>');
ERROR: 53200: could not copy node
CONTEXT: SQL function "xpath" statement 1
xpath() is changed so as it goes through xmlXPathCastNodeToString()
instead, that is able to handle namespace nodes. xml2 uses the same
solution. This issue has been discovered while digging into 9d33a5a804db.
Author: Michael Paquier <michael@paquier.xyz>
Discussion: https://postgr.es/m/aioT7ui_ZJ9RMlfM@paquier.xyz
Backpatch-through: 14
seg: Fix seg_out() to preserve the upper boundary's certainty indicator
When printing the upper boundary of a seg interval, seg_out() decided
whether to emit the certainty indicator ('<', '>' or '~') by testing the
upper indicator (u_ext) for '<' and '>', but mistakenly tested the lower
indicator (l_ext) for '~'. This is a copy-and-paste slip from the
symmetric code that prints the lower boundary a few lines above.
The consequences for valid input were:
* A '~' on the upper boundary was dropped on output, e.g.
'1.5 .. ~2.5'::seg printed as '1.5 .. 2.5'.
* When the lower boundary carried '~' but the upper boundary had no
indicator, the wrong test matched and sprintf(p, "%c", seg->u_ext)
wrote a NUL byte (u_ext == '\0'), which truncated the result string
and silently lost the entire upper boundary, e.g.
'~6.5 .. 8.5'::seg printed as '~6.5 .. '.
Certainty indicators are documented to be preserved on output (they are
ignored by the operators, but kept as comments), so this broke the
input/output round-trip for the affected values.
The bug has existed since seg was added. It went unnoticed because the
existing regression tests only exercised certainty indicators on
single-point segs, which are printed by a different branch of seg_out().
Add tests that place indicators on both boundaries of an interval.
Author: Ewan Young <kdbase.hack@gmail.com>
Discussion: https://www.postgresql.org/message-id/CAON2xHPYeRRCEVAv8XfE18KsEsEHCiYcJ5fOsoxFuMEfpxF1=g@mail.gmail.com
Backpatch-through: 14
Michael Paquier [Thu, 11 Jun 2026 05:29:29 +0000 (14:29 +0900)]
xml2: Fix crash with namespace nodes in xpath_nodeset()
pgxmlNodeSetToText() passed nodeTab[i]->doc to xmlNodeDump() without
checking the node type, which could cause a crash as a
XML_NAMESPACE_DECL maps to a xmlNs struct. The passed-in code would
then be dereferenced in xmlNodeDump().
This commit switches the code to render XML_NAMESPACE_DECL nodes with
xmlXPathCastNodeToString(), like xpath_table(). Some tests are added,
written by me.
Jeff Davis [Mon, 8 Jun 2026 18:47:40 +0000 (11:47 -0700)]
dict_synonym.c: remove incorrect outlen.
Previously, outlen was miscalculated if case_sensitive was false and
str_tolower() changed the byte length of the string. If outlen was too
large, pnstrdup() would stop at the NUL terminator, preventing
overrun. But if outlen was too small, it would cause truncation.
Fix by just removing outlen. It was only used in a single site, which
could just as well use pstrdup().
Tom Lane [Mon, 8 Jun 2026 15:48:07 +0000 (11:48 -0400)]
Fix missed checks for hashability of container-type equality.
The operators for array_eq, record_eq, range_eq, and multirange_eq
are all marked oprcanhash, but there's a pitfall: their hash functions
can fail at runtime if the contained type(s) are not hashable.
Therefore, the planner has to check hashability of the contained types
before deciding it can use hashing in these cases. Not every place
had gotten this memo, and noplace at all had considered the issue
for ranges or multiranges. In particular we could attempt to use
hashing for a ScalarArrayOpExpr on a container type when it won't
actually work, leading to "could not identify a hash function ..."
runtime failures.
For the most part we should fix this in the lookup functions provided
by lsyscache.c, to wit get_op_hash_functions and op_hashjoinable.
But there's a problem: get_op_hash_functions is not passed the input
data type it would need to check. We mustn't change the API of that
exported function in a back-patched fix, and even if we wanted to,
its call sites in the executor mostly don't have easy access to the
required data type OID. Fortunately, the executor call sites don't
actually need fixing, because it's expected that the planner verified
hashability before building a plan that requires it. Therefore,
leave get_op_hash_functions as-is and invent a wrapper function
get_op_hash_functions_ext that does the additional checking needed
in the planner's uses.
We also need to fix hash_ok_operator (extending the fix in 647889667).
While at it, neaten up a couple of places in lookup_type_cache where
relevant code for multirange cases was written differently from the
code for other container types.
Note: while this touches pg_operator.dat, it's only to add oid_symbol
macros. So there's no on-disk data change and no need for a
catversion bump.
Reported-by: Andrei Lepikhov <lepihov@gmail.com>
Author: Andrei Lepikhov <lepihov@gmail.com> Co-authored-by: Tom Lane <tgl@sss.pgh.pa.us>
Discussion: https://postgr.es/m/ed221f95-f09b-4a9c-b05b-e1fed621ec87@gmail.com
Backpatch-through: 14
Nathan Bossart [Mon, 8 Jun 2026 15:33:52 +0000 (10:33 -0500)]
doc: Expand on proper use of refint.
The security team has received a couple of reports about potential
SQL injection via refint's trigger arguments. We discussed this
while preparing CVE-2026-6637 and concluded that forcibly quoting
these arguments is more likely to break working code than to
prevent exploits. Unlike data values, the table/column names come
from trigger arguments, and there is little reason for a trigger
author to put hostile inputs into those arguments. So, let's
document it accordingly.
Reported-by: Nikolay Samokhvalov <nik@postgres.ai> Reported-by: Alex Young <alex000young@gmail.com> Reported-by: Satyanarayana Narlapuram <satyanarlapuram@gmail.com> Suggested-by: Noah Misch <noah@leadboat.com> Reviewed-by: Noah Misch <noah@leadboat.com> Reviewed-by: Fujii Masao <masao.fujii@oss.nttdata.com> Reviewed-by: Christoph Berg <myon@debian.org> Reviewed-by: Satyanarayana Narlapuram <satyanarlapuram@gmail.com>
Discussion: https://postgr.es/m/ahXP7z7nsfGPOZ3T%40nathan
Backpatch-through: 14
Fujii Masao [Mon, 8 Jun 2026 08:11:12 +0000 (17:11 +0900)]
ecpg: Reject multiple header items in GET/SET DESCRIPTOR
Previously, ecpg accepted multiple descriptor header items in GET DESCRIPTOR
and SET DESCRIPTOR, but generated broken C code when they were used.
Although the grammar allowed this syntax, the implementation did not actually
support it.
This commit tightens the ecpg grammar so the header form of GET/SET DESCRIPTOR
accepts only a single header item, matching the implementation and preventing
generation of broken C code.
Also update the documentation synopsis accordingly.
Michael Paquier [Mon, 8 Jun 2026 05:38:01 +0000 (14:38 +0900)]
psql: Fix expanded aligned output
When a table's columns are narrower than the record header line, the
expanded aligned format produced misaligned output because the data
column width was not adjusted to match the record header width, leading
to output like:
+-[ RECORD 1 ]-+
| a | 10 |
| b | 20 |
+---+----+
This commit adjusts the output so as the column width match with the
header line, giving:
+-[ RECORD 1 ]-+
| a | 10 |
| b | 20 |
+---+----------+
Author: Pavel Stehule <pavel.stehule@gmail.com> Reviewed-by: Chao Li <li.evan.chao@gmail.com>
Discussion: https://postgr.es/m/CAFj8pRCzGpsr9zTHbtTd4mGh2YPJqOEgLgt8JLiopuYA9_1xGw@mail.gmail.com
Backpatch-through: 14
Michael Paquier [Fri, 5 Jun 2026 23:16:46 +0000 (08:16 +0900)]
pg_surgery: Fix off-by-one bug with heap offset
heap_force_common() declared a boolean array indexed with an
OffsetNumber for a size of MaxHeapTuplesPerPage. OffsetNumbers are
1-based, so an input TID whose offset number equals MaxHeapTuplesPerPage
wrote one byte past the end of the stack array, crashing the server.
Like heapam_handler.c, this commit changes the array so as it uses a
0-based index, substracting one from the OffsetNumbers.
Nathan Bossart [Fri, 5 Jun 2026 17:08:05 +0000 (12:08 -0500)]
refint: Remove plan cache.
Presently, refint stores plans in a per-backend cache to avoid
re-preparing in each call. This has a few problems. For one,
check_foreign_key() embeds the new key values in its cascade-UPDATE
queries, so a cached plan reuses the values from preparation.
Also, the cache is never invalidated, so it can return stale
entries that cause other problems. There may very well be more
bugs lurking.
We could spend a lot of time trying to address all these problems,
but this module is primarily intended as sample code, and by all
indications, it sees minimal use. Furthermore, there is a growing
consensus for removing refint in v20. However, since we'll need to
support it on the back-branches for a while longer, it probably
still makes sense to fix some of the more egregious bugs.
Therefore, let's just remove refint's plan cache entirely. That
means we'll re-prepare on every call, but that seems quite unlikely
to bother anyone. On v17 and older versions, the regression test
for triggers fails after this change, so I've borrowed pieces of
commit 8cfbdf8f4d to fix it.
Michael Paquier [Thu, 4 Jun 2026 22:50:18 +0000 (07:50 +0900)]
Fix off-by-one with NFC recomposition for Hangul U+11A7 (TBASE)
The NFC recomposition incorrectly included TBASE as a valid T syllable,
which is incorrect based on the Unicode specification (TBASE is one
below the start of the range, range beginning at U+11A8).
This would cause the TBASE to be silently swallowed in the
normalization, leading to an incorrect result.
A couple of regression tests are added to check more patterns with
Hangul recomposition and decomposition, on top of a test to check the
problem with TBASE. Diego has submitted the code fix, and I have
written the tests.
Author: Diego Frias <mail@dzfrias.dev> Co-authored-by: Michael Paquier <michael@paquier.xyz>
Discussion: https://postgr.es/m/B92ED640-7D4A-4505-B09F-3548F58CBB16@dzfrias.dev
Backpatch-through: 14
Tom Lane [Thu, 4 Jun 2026 16:24:51 +0000 (12:24 -0400)]
Improve reporting of invalid weight symbols in setweight() et al.
This commit addresses two related issues:
tsvector_filter() assumed it could print an incorrect weight value
with %c. This could result in an invalidly-encoded error message
if the database encoding is multibyte and the char value has its
high bit set. Weight values that are ASCII control characters
could render illegibly too. Fix by printing such values in octal
(\ooo), similarly to how charout() would render them.
tsvector_setweight() and tsvector_setweight_by_filter() reported
the same unrecognized-weight error condition with elog(), as though
it were an internal error. That'd not translate, would produce an
unwanted XX000 SQLSTATE code, and also reported the bad value as a
decimal integer which seems unhelpful. Fix by refactoring so that
all three functions share one copy of the code that interprets a
weight argument.
The invalid-encoding aspect seems to me (tgl) to justify
back-patching.
Author: Ewan Young <kdbase.hack@gmail.com> Reviewed-by: Tom Lane <tgl@sss.pgh.pa.us>
Discussion: https://postgr.es/m/CAON2xHNaeLAUzRCXL5AmXLcXaSE_gWAVjWQRmLzc_oZ=1_Vf4Q@mail.gmail.com
Backpatch-through: 14
Tom Lane [Thu, 4 Jun 2026 15:37:43 +0000 (11:37 -0400)]
Fix another case of indirectly casting away const.
Like 8f1791c61, this fixes a case of implicitly casting away
const by not treating the result of strrchr() on a const pointer
as const. This was missed at the time because the machines
reporting those warnings weren't building with --with-llvm.
While here, clean up another infelicity: in the probably-
impossible case that the input string contains only one dot,
this function would call pnstrdup() with a length of -1
and thereby emit a module name equal to the function name.
It seems to me we should emit modname = NULL instead.
Also remove a useless Assert and two redundant assignments.
Back-patch, as 8f1791c61 was, so that users of back branches
don't see this warning when building with late-model gcc.
Reported-by: hubert depesz lubaczewski <depesz@depesz.com>
Author: Tom Lane <tgl@sss.pgh.pa.us>
Discussion: https://postgr.es/m/aiGNJ89PBqvq2Yyz@depesz.com
Backpatch-through: 14
Fujii Masao [Wed, 3 Jun 2026 09:47:31 +0000 (18:47 +0900)]
Fix race in ReplicationSlotRelease() for ephemeral slots
When releasing an ephemeral replication slot, ReplicationSlotRelease()
drops the slot via ReplicationSlotDropAcquired().
However, after dropping the slot, ReplicationSlotRelease() continued
to use its local "slot" pointer, which still referenced the dropped
slot's former shared-memory entry. It could then update fields such as
effective_xmin in that entry.
Once an ephemeral slot has been dropped (via ReplicationSlotDropAcquired()),
its slot array entry can be reused immediately by another backend
creating a new slot. As a result, those updates could corrupt
the state of an unrelated replication slot.
Fix by skipping those shared-memory updates for phemeral slots and
performing them only for non-ephemeral slots, whose shared-memory
entries remain valid after release.
Michael Paquier [Wed, 3 Jun 2026 03:47:34 +0000 (12:47 +0900)]
Fix copy-paste error in hash_record_extended()
The code failed to initialize the second isnull argument passed to
FunctionCallInvoke(). This is harmless for existing in-core extended
hash support functions, since FunctionCallInvoke() does not use the
value (note that all the in-core extended hash functions are strict),
examining only the argument values. However, extension-provided
extended hash functions could be affected if they inspect
PG_ARGISNULL(1).
Andres Freund [Thu, 28 May 2026 15:34:14 +0000 (11:34 -0400)]
Make stack depth check work with asan's use-after-return
With address sanitizer's stack-use-after-return check, stack variables are
moved to heap allocations, to allow to detect references to the memory at a
later time. That broke our stack-depth check, which is why we had to disable
detect_stack_use_after_return in CI. Luckily __builtin_frame_address() works
correctly, even under asan, so use that.
We started using __builtin_frame_address() with de447bb8e6fb, however as of
that commit we just used it for the stack base address, not for the value to
compare to the base address. Now we use it for both.
When building without __builtin_frame_address() support, we continue to use
stack variables for the stack depth determination.
Reviewed-by: Tom Lane <tgl@sss.pgh.pa.us>
Discussion: https://postgr.es/m/2kk4z4odvuyrg7qlwjd7ft4eron4cle4btb33v4qatgsdkayir@gj6e62rgsel4
Backpatch-through: 14
Concurrent DDL can leave behind objects referencing other objects that
no longer exist. This can happen if an object is dropped, while a new
object that depends on it is created concurrently. For example:
session 1: BEGIN; CREATE FUNCTION myschema.myfunc() ...;
session 2: DROP SCHEMA myschema;
session 1: COMMIT;
DROP SCHEMA does check that there are no objects dependending on the
schema being dropped, but it does not see objects being concurrently
created by other sessions. Even if it did, this scenario would still
fail:
session 1: BEGIN: DROP SCHEMA myschema;
session 2: CREATE FUNCTION myschema.myfunc() ...;
session 1: COMMIT;
When the DROP SCHEMA runs, the schema was empty, but the new function
is created in it before the dropping transaction completes. The CREATE
FUNCTION does not see that the schema is concurrently being dropped.
In both of these scenarios, the function is left behind in the schema
that no longer exists.
To fix, acquire AccessShareLock on all referenced objects when
recording dependencies. This conflicts with the AccessExclusiveLock
taken by DROP, preventing the race. After acquiring the lock, verify
that the object still exists, and if it was dropped concurrently,
report an error. We already had such a mechanism for shared
dependencies, but for some reason we didn't do it for in-database
dependendies.
Ideally the locks would be acquired much earlier when creating a new
object, but that will require modifying a lot of callers. This check
while recording the dependency is a nice wholesale protection, and
even if we change all the CREATE commands to acquire locks earlier,
it's still good to have this as a backstop to catch any cases where we
forgot to do so.
The patch adds a few tests for some cases that left behind orphaned
objects before this. It also adds a test for roles, which already had
such protection, although that test is partially disabled because the
error message includes an OID which is not predictable.
Don't try to record dependency on a dropped column's datatype
When creating a relation with a dropped column, we called
recordDependencyOn() also on the datatype of the dropped column, which
is always InvalidOid. In versions 15 and above, that was harmless
because recordDependencyOn() considers InvalidOid as a pinned object,
and skips over it. On version 14, isPinnedObject() does not consider
InvalidOid as pinned, so we created a bogus pg_depend entry with
refobjectid == 0.
As far as I can tell, the only case when AddNewAttributeTuples() is
called with dropped columns is when performing a table-rewriting ALTER
TABLE command. That temporarily creates a new relation with the same
columns, including dropped ones, then swaps the relations, and drops
the newly created table again. So even on version 14, the bogus
pg_depend entry was only on the transient relation that was dropped at
the end of the ALTER TABLE command, which was harmless.
Even though this is harmless, let's be tidy, similar to commit 713bce9484. The reason I noticed this now and why I backported this,
is because the next commit will add code to acquire locks on the
referenced objects, and we don't want to acquire a lock on InvalidOid.
Fix self-deadlock when replaying WAL generated by older minor version
Commit 77dff5d937 introduced a SimpleLruWriteAll() call when replaying
multixact WAL records generated by older minor versions. However,
SimpleLruWriteAll() acquires the SLRU lock and on v16 and below, it's
called while already holding the lock, leading to self-deadlock.
Version 17 and 18 did not have that problem, because in those versions
the lock is acquired later in the function.
To fix, acquire MultiXactOffsetSLRULock later in RecordNewMultiXact(),
at the same place where it's acquired on version 17 and 18.
Michael Paquier [Wed, 27 May 2026 08:20:00 +0000 (17:20 +0900)]
Fix procLatch ownership race in ProcKill()
DisownLatch() was executed after the PGPROC entry of the process
terminated is pushed back into a freelist. A newly-forked backend that
recycles the slot could call OwnLatch() and PANIC with a "latch already
owned by PID", taking down the server.
There were two scenarios related to lock groups where this issue could
be reached:
* A follower pushes the leader's PGPROC back to the freelist while the
leader has not yet called DisownLatch() in its own ProcKill().
* A leader outliving all its followers pushes its own PGPROC onto the
freelist before reaching DisownLatch(), which would be the most common
scenario.
This issue is fixed by calling SwitchBackToLocalLatch() and
DisownLatch() at an earlier phase of ProcKill(), before any freelist
manipulation happens, so that the slot of the backend terminated is
never exposed as owning a latch.
Note that pgstat_reset_wait_event_storage() is kept at a later stage.
An upcoming commit will take advantage of that by introducing a test
able to check the original PANIC scenario.
Michael Paquier [Wed, 27 May 2026 05:52:31 +0000 (14:52 +0900)]
Fix race conditions in ProcKill()'s lock-group freelist handling
This commit fixes two bugs in ProcKill()'s lock-group teardown freelist
publication:
* a double push of the leader's PGPROC that corrupts the freelist.
* a leak of the last follower's PGPROC slot.
ProcKill()'s lock-group teardown had two PGPROC freelist updates
scattered through the function, done under two separate freeProcsLock
acquisitions:
* A follower's push of the leader's PGPROC, done when a follower is the
last group member exiting.
* Every backend's self-push at the bottom of the function.
The two freelist updates were coordinated only by inspecting
proc->lockGroupLeader, which a follower could clear as a side effect of
pushing the leader. This coordination was broken. For example, with
two concurrent backends:
* The follower clears leader->lockGroupLeader and pushes the leader's
PGPROC under leader_lwlock.
* The follower does not clear its own proc->lockGroupLeader, being
skipped.
* When the leader reaches the bottom of ProcKill(), it sees a NULL
proc->lockGroupLeader (the follower cleared it) and pushes itself,
causing a second dlist_push_tail() of the same node onto the same
freelist.
* The follower at the bottom sees its own proc->lockGroupLeader being
not NULL (never cleared) and skips its own push, causing its own slot
to leak.
This commit refactors the freelist manipulation to be done in two
distinct phases, each step using its own lock acquisition to ensure that
each freelist operation happens in an isolated manner for each backend
(follower or leader):
- First, under a single leader_lwlock acquisition, check the state of
the lock-group. Depending on if we are dealing with a follower and/or a
leader, and if the leader has exited before a follower, then set some
state booleans that define which actions should be taken with the
freelist.
- Second, under a single freeProcsLock acquisition, perform the cleanup
actions, self-push of a backend and/or push of the leader back to the
freelist.
This is an old issue, dating back to 9.6 where parallel workers and lock
grouping has been added.
Tom Lane [Mon, 25 May 2026 22:15:49 +0000 (18:15 -0400)]
Fix missed ReleaseVariableStats() in intarray's _int_matchsel().
Given a WHERE clause like "int[] @@ query_int" or "query_int ~~ int[]"
where the query_int side is a table column having statistics,
_int_matchsel() exited without remembering to free the statistics
tuple. This would typically lead to warnings about cache refcount
leakage, like
WARNING: resource was not closed: cache pg_statistic (73), tuple 42/12 has count 1
It's been wrong since this code was added, in commit c6fbe6d6f.
Bug: #19492 Reported-by: Man Zeng <zengman@halodbtech.com>
Author: Man Zeng <zengman@halodbtech.com> Reviewed-by: Tom Lane <tgl@sss.pgh.pa.us>
Discussion: https://postgr.es/m/19492-ddcd0e22399ef85a@postgresql.org
Backpatch-through: 14
Michael Paquier [Mon, 25 May 2026 05:39:07 +0000 (14:39 +0900)]
Fix size check in statext_dependencies_deserialize()
The check for the minimum expected bytea size of a MVDependencies object
was using SizeOfItem() for its calculation. This macro uses the number
of attributes in a single dependency.
This minimum size calculation should be based on MinSizeOfItems(), that
computes the minimum expected size as the header plus the
minimally-sized number of dependency items.
Michael Paquier [Fri, 22 May 2026 23:10:18 +0000 (08:10 +0900)]
Avoid exposing WAL receiver raw conninfo during timeline jumps
When reusing an existing WAL receiver after it has reached
WALRCV_WAITING for new instructions, RequestXLogStreaming() copied
PrimaryConnInfo into WalRcv->conninfo before switching the state to
WALRCV_RESTARTING. At that point ready_to_display could still be true,
so pg_stat_wal_receiver could expose the raw connection string,
including sensitive fields, but it should only show the user-displayable
version of the connection string.
WALRCV_RESTARTING does not establish a new connection. The waiting WAL
receiver reuses its existing connection and only needs a new startpoint
and timeline, so there is no need to copy the raw connection string into
shared memory again. Let's only copy conninfo when launching a new WAL
receiver after WALRCV_STOPPED, not while waiting for instructions.
This commit adds coverage for the case fixed by this commit to the
timeline-switch test by verifying that the WAL receiver conninfo remains
consistent across the jump.
Backpatch all the way down, as this issue is possible since
pg_stat_wal_receiver has been introduced.
Author: Chao Li <li.evan.chao@gmail.com> Reviewed-by: Michael Paquier <michael@paquier.xyz>
Discussion: https://postgr.es/m/EF91FF76-1E2B-4F3B-9162-290B4DC517FF@gmail.com
Backpatch-through: 14
Fujii Masao [Wed, 20 May 2026 06:54:13 +0000 (15:54 +0900)]
pg_recvlogical: Honor source cluster file permissions for output files
Commit c37b3d08ca6 attempted to preserve group permissions on pg_recvlogical
output files when group access was enabled on the source cluster. However,
the output files were still created with a fixed S_IRUSR | S_IWUSR mode,
preventing group-read permissions from being applied.
This commit fixes the issue by creating output files with pg_file_create_mode
instead of a hard-coded mode. This allows pg_recvlogical to correctly preserve
group permissions from the source cluster.
Noah Misch [Sun, 17 May 2026 01:01:35 +0000 (18:01 -0700)]
Use ereport(ERROR), not Assert(), for publisher tuples missing columns.
Three locations use Assert() to guard against a mismatch between the
number of columns advertised in the RELATION message and the number
actually received in the subsequent INSERT/UPDATE tuple message. Since
these values originate from the publisher, the check must survive into
production builds.
A malicious or buggy publisher can send a RELATION claiming N columns
and an INSERT claiming M < N columns. The subscriber's apply worker
indexes into colvalues[]/colstatus[] using column indices from the
RELATION message's attribute map, causing a heap out-of-bounds read when
the tuple's column array is smaller than expected. We've looked, without
success, for a scenario in which the publisher holds sufficient control
over these out-of-bounds bytes to exploit this or even to reach a
SIGSEGV. Despite not finding one, the code has been fragile. Back-patch
to v14 (all supported versions).
Tom Lane [Fri, 15 May 2026 22:32:33 +0000 (18:32 -0400)]
Doc: fix release-note typo.
This mention of memcpy() should of course have said memcmp().
Reported-by: chris@chrullrich.net
Author: Tom Lane <tgl@sss.pgh.pa.us>
Discussion: https://postgr.es/m/177883653690.764749.14038057906859461991@wrigleys.postgresql.org
Backpatch-through: 14
Michael Paquier [Fri, 15 May 2026 09:02:54 +0000 (18:02 +0900)]
Re-add regression tests for ltree and intarray
These tests have been removed by 906ea101d0d5, due to some of them being
unstable in the buildfarm with low max_stack_depth values. They are now
reworked so as they should be more portable.
The tests to cover the findoprnd() overflows use a balanced tree to
avoid using too much stack, per a suggestion and an investigation by Tom
Lane.
Note: This is initially applied only on HEAD; a backpatch will follow
should the buildfarm be fine with the situation.
Nathan Bossart [Thu, 14 May 2026 18:11:49 +0000 (13:11 -0500)]
refint: Fix segfault in check_foreign_key().
When an UPDATE statement triggers check_foreign_key() with the
action set to "cascade", it generates more UPDATE statements to
modify the key values in referencing relations. If a new key value
is NULL, SPI_getvalue() returns a NULL pointer, which is
subsequently passed to quote_literal_cstr(), causing a segfault.
To fix, skip quoting when a new key value is NULL and insert an
unquoted NULL keyword instead.
Oversight in commit 260e97733b. While the refint documentation
recommends marking primary key columns NOT NULL, the aforementioned
scenario accidentally worked on platforms where snprintf()
substitutes "(null)" for NULL pointers. Note that for
character-type columns, the old code quoted "(null)" as a string
literal, so this didn't always produce correct results. But it
still seems better to fix this than to reject cases that previously
worked.
Michael Paquier [Wed, 13 May 2026 05:43:52 +0000 (14:43 +0900)]
Add more tests for corrupted data with pglz_decompress()
Two cases fixed by 2b5ba2a0a141 were not covered, to emulate the
handling of corrupted data, for:
- set control bit with a valid 2-byte match tag where offset is 0.
- set control bit with a valid 2-byte match tag where offset exceeds
output written.
Fujii Masao [Wed, 13 May 2026 02:44:31 +0000 (11:44 +0900)]
Fix stale COPY progress during logical replication table sync
Previously, pg_stat_progress_copy in the subscriber could continue to show
the initial COPY operation for logical replication table synchronization as
active even after the data copy had finished. The stale progress entry
remained visible until synchronization caught up with the publisher.
This happened because the table synchronization code called BeginCopyFrom()
and CopyFrom(), but failed to call EndCopyFrom() afterward.
This commit fixes the issue by adding the missing EndCopyFrom() call so that
the COPY progress state in the subscriber is cleared as soon as the initial
data copy completes.
Tom Lane [Mon, 11 May 2026 16:12:03 +0000 (12:12 -0400)]
Remove test cases for field overflows in intarray and ltree.
These checks are failing in the buildfarm, reporting stack overflows
rather than the expected errors, though seemingly only on ppc64 and
s390x platforms. Perhaps there is something off about our tests
for stack depth on those architectures? But there's no time to
debug that right now, and surely these tests aren't too essential.
Revert for now and plan to revisit after the release dust settles.
Nathan Bossart [Mon, 11 May 2026 12:13:52 +0000 (05:13 -0700)]
refint: Fix SQL injection and buffer overruns.
Maliciously crafted key value updates could achieve SQL injection
within check_foreign_key(). To fix, ensure new key values are
properly quoted and escaped in the internally generated SQL
statements. While at it, avoid potential buffer overruns by
replacing the stack buffers for internally generated SQL statements
with StringInfo.
Reported-by: Nikolay Samokhvalov <nik@postgres.ai>
Author: Nathan Bossart <nathandbossart@gmail.com> Reviewed-by: Noah Misch <noah@leadboat.com> Reviewed-by: Tom Lane <tgl@sss.pgh.pa.us> Reviewed-by: Fujii Masao <masao.fujii@gmail.com>
Security: CVE-2026-6637
Backpatch-through: 14
Nathan Bossart [Mon, 11 May 2026 12:13:51 +0000 (05:13 -0700)]
Mark PQfn() unsafe and fix overrun in frontend LO interface.
When result_is_int is set to 0, PQfn() cannot validate that the
result fits in result_buf, so it will write data beyond the end of
the buffer when the server returns more data than requested. Since
this function is insecurable and obsolete, add a warning to the top
of the pertinent documentation advising against its use.
The only in-tree caller of PQfn() is the frontend large object
interface. To fix that, add a buf_size parameter to
pqFunctionCall3() that is used to protect against overruns, and use
it in a private version of PQfn() that also accepts a buf_size
parameter.
Reported-by: Yu Kunpeng <yu443940816@live.com> Reported-by: Martin Heistermann <martin.heistermann@unibe.ch>
Author: Nathan Bossart <nathandbossart@gmail.com> Reviewed-by: Noah Misch <noah@leadboat.com> Reviewed-by: Tom Lane <tgl@sss.pgh.pa.us> Reviewed-by: Etsuro Fujita <etsuro.fujita@gmail.com>
Security: CVE-2026-6477
Backpatch-through: 14
Fix integer overflow in array_agg(), when the array grows too large
If you accumulate many arrays full of NULLs, you could overflow
'nitems', before reaching the MaxAllocSize limit on the allocations.
Add an explicit check that the number of items doesn't grow too large.
With more than MaxArraySize items, getting the final result with
makeArrayResultArr() would fail anyway, so better to error out early.
Reported-by: Xint Code
Author: Heikki Linnakangas <heikki.linnakangas@iki.fi> Reviewed-by: Tom Lane <tgl@sss.pgh.pa.us>
Backpatch-through: 14
Security: CVE-2026-6473
Tom Lane [Mon, 11 May 2026 12:13:51 +0000 (05:13 -0700)]
Fix integer-overflow and alignment hazards in locale-related code.
pg_locale_icu.c was full of places where a very long input string
could cause integer overflow while calculating a buffer size,
leading to buffer overruns.
It also was cavalier about using char-type local arrays as buffers
holding arrays of UChar. The alignment of a char[] variable isn't
guaranteed, so that this risked failure on alignment-picky platforms.
The lack of complaints suggests that such platforms are very rare
nowadays; but it's likely that we are paying a performance price on
rather more platforms. Declare those arrays as UChar[] instead,
keeping their physical size the same.
pg_locale_libc.c's strncoll_libc_win32_utf8() also had the
disease of assuming it could double or quadruple the input
string length without concern for overflow.
Reported-by: Xint Code Reported-by: Pavel Kohout <pavel.kohout@aisle.com>
Author: Tom Lane <tgl@sss.pgh.pa.us>
Backpatch-through: 14
Security: CVE-2026-6473
Michael Paquier [Mon, 11 May 2026 12:13:51 +0000 (05:13 -0700)]
Prevent path traversal in pg_basebackup and pg_rewind
pg_rewind and pg_basebackup could be fed paths from rogue endpoints that
could overwrite the contents of the client when received, achieving path
traversal.
There were two areas in the tree that were sensitive to this problem:
- pg_basebackup, through the astreamer code, where no validation was
performed before building an output path when streaming tar data. This
is an issue in v15 and newer versions.
- pg_rewind file operations for paths received through libpq, for all
the stable branches supported.
In order to address this problem, this commit adds a helper function in
path.c, that reuses path_is_relative_and_below_cwd() after applying
canonicalize_path(). This can be used to validate the paths received
from a connection point. A path is considered invalid if any of the two
following conditions is satisfied:
- The path is absolute.
- The path includes a direct parent-directory reference.
Reported-by: XlabAI Team of Tencent Xuanwu Lab Reported-by: Valery Gubanov <valerygubanov95@gmail.com>
Author: Michael Paquier <michael@paquier.xyz> Reviewed-by: Amit Kapila <amit.kapila16@gmail.com>
Backpatch-through: 14
Security: CVE-2026-6475
Nathan Bossart [Mon, 11 May 2026 12:13:51 +0000 (05:13 -0700)]
Avoid overflow in size calculations in formatting.c.
A few functions in this file were incautious about multiplying a
possibly large integer by a factor more than 1 and then using it as
an allocation size. This is harmless on 64-bit systems where we'd
compute a size exceeding MaxAllocSize and then fail, but on 32-bit
systems we could overflow size_t, leading to an undersized
allocation and buffer overrun. To fix, use palloc_array() or
mul_size() instead of handwritten multiplication.
Reported-by: Sven Klemm <sven@tigerdata.com> Reported-by: Xint Code
Author: Nathan Bossart <nathandbossart@gmail.com> Reviewed-by: Tom Lane <tgl@sss.pgh.pa.us> Reviewed-by: Tatsuo Ishii <ishii@postgresql.org>
Security: CVE-2026-6473
Backpatch-through: 14
Nathan Bossart [Mon, 11 May 2026 12:13:51 +0000 (05:13 -0700)]
Check CREATE privilege on multirange type schema in CREATE TYPE.
This omission allowed roles to create multirange types in any
schema, potentially leading to privilege escalations. Note that
when a multirange type name is not specified in CREATE TYPE, it is
automatically placed in the range type's schema, which is checked
at the beginning of DefineRange().
Tom Lane [Mon, 11 May 2026 12:13:51 +0000 (05:13 -0700)]
Guard against unsafe conditions in usage of pg_strftime().
Although pg_strftime() has defined error conditions, no callers bother
to check for errors. This is problematic because the output string is
very likely not null-terminated if an error occurs, so that blindly
using it is unsafe. Rather than trusting that we can find and fix all
the callers, let's alter the function's API spec slightly: make it
guarantee a null-terminated result so long as maxsize > 0.
Furthermore, if we do get an error, let's make that null-terminated
result be an empty string. We could instead truncate at the buffer
length, but that risks producing mis-encoded output if the tz_name
string contains multibyte characters. It doesn't seem reasonable for
src/timezone/ to make use of our encoding-aware truncation logic.
Also, the only really likely source of a failure is a user-supplied
timezone name that is intentionally trying to overrun our buffers.
I don't feel a need to be particularly friendly about that case.
Author: Tom Lane <tgl@sss.pgh.pa.us> Reviewed-by: John Naylor <johncnaylorls@gmail.com>
Backpatch-through: 14
Security: CVE-2026-6474
Tom Lane [Mon, 11 May 2026 12:13:51 +0000 (05:13 -0700)]
Avoid passing unintended format codes to snprintf().
timeofday() assumed that the output of pg_strftime() could not contain
% signs, other than the one it explicitly asks for with %%. However,
we don't have that guarantee with respect to the time zone name (%Z).
A crafted time zone setting could abuse the subsequent snprintf()
call, resulting in crashes or disclosure of server memory.
To fix, split the pg_strftime() call into two and then treat the
outputs as literal strings, not a snprintf format string. The
extra pg_strftime() call doesn't really cost anything, since the
bulk of the conversion work was done by pg_localtime().
Also, adjust buffer widths so that we're not risking string truncation
during the snprintf() step, as that would create a hazard of producing
mis-encoded output.
This also fixes a latent portability issue: the format string expects
an int, but tp.tv_usec is long int on many platforms.
Reported-by: Xint Code
Author: Tom Lane <tgl@sss.pgh.pa.us> Reviewed-by: John Naylor <johncnaylorls@gmail.com>
Backpatch-through: 14
Security: CVE-2026-6474
Michael Paquier [Mon, 11 May 2026 12:13:51 +0000 (05:13 -0700)]
Apply timingsafe_bcmp() in authentication paths
This commit applies timingsafe_bcmp() to authentication paths that
handle attributes or data previously compared with memcpy() or strcmp(),
which are sensitive to timing attacks.
The following data is concerned by this change, some being in the
backend and some in the frontend:
- For a SCRAM or MD5 password, the computed key or the MD5 hash compared
with a password during a plain authentication.
- For a SCRAM exchange, the stored key, the client's final nonce and the
server nonce.
- RADIUS (up to v18), the encrypted password.
- For MD5 authentication, the MD5(MD5()) hash.
Reported-by: Joe Conway <mail@joeconway.com>
Security: CVE-2026-6478
Author: Michael Paquier <michael@paquier.xyz> Reviewed-by: John Naylor <johncnaylorls@gmail.com>
Backpatch-through: 14
Add timingsafe_bcmp(), for constant-time memory comparison
timingsafe_bcmp() should be used instead of memcmp() or a naive
for-loop, when comparing passwords or secret tokens, to avoid leaking
information about the secret token by timing. This commit just
introduces the function but does not change any existing code to use
it yet.
This has been initially applied as of 09be39112654 in v18 and newer
versions, and will be used in all the stable branches for an upcoming
fix.
Tom Lane [Mon, 11 May 2026 12:13:51 +0000 (05:13 -0700)]
Guard against overflow in "left" fields of query_int and ltxtquery.
contrib/intarray's query_int type uses an int16 field to hold the
offset from a binary operator node to its left operand. However, it
allows the number of nodes to be as much as will fit in MaxAllocSize,
so there is a risk of overflowing int16 depending on the precise shape
of the tree. Simple right-associative cases like "a | b | c | ..."
work fine, so we should not solve this by restricting the overall
number of nodes. Instead add a direct test of whether each individual
offset is too large.
contrib/ltree's ltxtquery type uses essentially the same logic and
has the same 16-bit restriction.
(The core backend's tsquery.c has a variant of this logic too, but
in that case the target field is 32 bits, so it is okay so long
as varlena datums are restricted to 1GB.)
In v16 and up, these types support soft error reporting, so we have
to complicate the recursive findoprnd function's API a bit to allow
the complaint to be reported softly. v14/v15 don't need that.
Undocumented and overcomplicated code like this makes my head hurt,
so add some comments and simplify while at it.
Reported-by: Xint Code
Author: Tom Lane <tgl@sss.pgh.pa.us> Reviewed-by: Michael Paquier <michael@paquier.xyz>
Backpatch-through: 14
Security: CVE-2026-6473
Tom Lane [Mon, 11 May 2026 12:13:51 +0000 (05:13 -0700)]
Unify src/common/'s definitions of MaxAllocSize.
Define MaxAllocSize in src/include/common/fe_memutils.h rather
than having several copies of it in different src/common/*.c files.
This also provides an opportunity to document it better.
Back-patch of commit 11b7de4a7, needed now because assorted security
fixes are adding additional references to MaxAllocSize in frontend
code.
Michael Paquier [Mon, 11 May 2026 12:13:51 +0000 (05:13 -0700)]
Fix unbounded recursive handling of SSL/GSS in ProcessStartupPacket()
The handling of SSL and GSS negotiation messages in
ProcessStartupPacket() could cause a recursion of the backend,
ultimately crashing the server as the negotiation attempts were not
tracked across multiple calls processing startup packets.
A malicious client could therefore alternate rejected SSL and GSS
requests indefinitely, each adding a stack frame, until the backend
crashed with a stack overflow, taking down a server.
This commit addresses this issue by modifying ProcessStartupPacket() so
as processed negotiation attempts are tracked, preventing infinite
recursive attempts. A TAP test is added to check this problem, where
multiple SSL and GSS negotiated attempts are stacked.
Reported-by: Calif.io in collaboration with Claude and Anthropic
Research
Author: Michael Paquier <michael@paquier.xyz> Reviewed-by: Daniel Gustafsson <daniel@yesql.se>
Security: CVE-2026-6479
Backpatch-through: 14
Michael Paquier [Mon, 11 May 2026 12:13:51 +0000 (05:13 -0700)]
Add raw_connect and raw_connect_works to Cluster.pm
These two routines will be used in a test of an upcoming fix. This
commit affects the v14~v17 range. v18 and newer versions already
include them, thanks to 85ec945b7880.
Tom Lane [Mon, 11 May 2026 12:13:51 +0000 (05:13 -0700)]
Fix assorted places that need to use palloc_array().
multirange_recv and BlockRefTableReaderNextRelation were incautious
about multiplying a possibly-large integer by a factor more than 1
and then using it as an allocation size. This is harmless on 64-bit
systems where we'd compute a size exceeding MaxAllocSize and then
fail, but on 32-bit systems we could overflow size_t leading to an
undersized allocation and buffer overrun.
Fix these places by using palloc_array() instead of a handwritten
multiplication. (In HEAD, some of them were fixed already, but
none of that work got back-patched at the time.)
In addition, BlockRefTableReaderNextRelation passes the same value
to BlockRefTableRead's "int length" parameter. If built for
64-bit frontend code, palloc_array() allows a larger array size
than it otherwise would, potentially allowing that parameter to
overflow. Add an explicit check to forestall that and keep the
behavior the same cross-platform.
Reported-by: Xint Code
Author: Tom Lane <tgl@sss.pgh.pa.us>
Backpatch-through: 14
Security: CVE-2026-6473
Tom Lane [Mon, 11 May 2026 12:13:51 +0000 (05:13 -0700)]
Prevent buffer overrun in unicode_normalize().
Some UTF8 characters decompose to more than a dozen codepoints.
It is possible for an input string that fits into well under
1GB to produce more than 4G decomposed codepoints, causing
unicode_normalize()'s decomp_size variable to wrap around to a
small positive value. This results in a small output buffer
allocation and subsequent buffer overrun.
To fix, test after each addition to see if we've overrun MaxAllocSize,
and break out of the loop early if so. In frontend code we want to
just return NULL for this failure (treating it like OOM). In the
backend, we can rely on the following palloc() call to throw error.
I also tightened things up in the calling functions in varlena.c,
using size_t rather than int and allocating the input workspace
with palloc_array(). These changes are probably unnecessary
given the knowledge that the original input and the normalized
output_chars array must fit into 1GB, but it's a lot easier to
believe the code is safe with these changes.
Reported-by: Xint Code Reported-by: Bruce Dang <bruce@calif.io>
Author: Tom Lane <tgl@sss.pgh.pa.us> Co-authored-by: Heikki Linnakangas <hlinnaka@iki.fi>
Backpatch-through: 14
Security: CVE-2026-6473
Tom Lane [Mon, 11 May 2026 12:13:51 +0000 (05:13 -0700)]
Harden our regex engine against integer overflow in size calculations.
The number of NFA states, number of NFA arcs, and number of colors
are all bounded to reasonably small values. However, there are
places where we try to allocate arrays sized by products of those
quantities, and those calculations could overflow, enabling
buffer-overrun attacks. In practice there's no problem on 64-bit
machines, but there are some live scenarios on 32-bit machines.
A related problem is that citerdissect() and creviterdissect()
allocate arrays based on the length of the input string, which
potentially could overflow.
To fix, invent MALLOC_ARRAY and REALLOC_ARRAY macros that rely on
palloc_array_extended and repalloc_array_extended with the NO_OOM
option, similarly to the existing MALLOC and REALLOC macros.
(Like those, they'll throw an error not return a NULL result for
oversize requests. This doesn't really fit into the regex code's
view of error handling, but it'll do for now. We can consider
whether to change that behavior in a non-security follow-up patch.)
I installed similar defenses in the colormap construction code.
It's not entirely clear whether integer overflow is possible
there, but analyzing the behavior in detail seems not worth
the trouble, as the risky spots are not in hot code paths.
I left a bunch of calls as-is after verifying that they can't
overflow given reasonable limits on nstates and narcs. Those
limits were enforced already via REG_MAX_COMPILE_SPACE, but
add commentary to document the interactions.
In passing, also fix a related edge case, which is that the
special color numbers used in LACON carcs could overflow the
"color" data type, if ncolors is close to MAX_COLOR.
In v14 and v15, the regex engine calls malloc() directly instead
of using palloc(), so MALLOC_ARRAY and REALLOC_ARRAY do likewise.
Reported-by: Xint Code
Author: Tom Lane <tgl@sss.pgh.pa.us> Reviewed-by: Masahiko Sawada <sawada.mshk@gmail.com>
Backpatch-through: 14
Security: CVE-2026-6473
Tom Lane [Mon, 11 May 2026 12:13:51 +0000 (05:13 -0700)]
Make palloc_array() and friends safe against integer overflow.
Sufficiently large "count" arguments could result in undetected
overflow, causing the allocated memory chunk to be much smaller
than what the caller will subsequently write into it. This is
unlikely to be a hazard with 64-bit size_t but can sometimes
happen on 32-bit builds, primarily where a function allocates
workspace that's significantly larger than its input data.
Rather than trying to patch the at-risk callers piecemeal,
let's just redefine these macros so that they always check.
To do that, move the longstanding add_size() and mul_size() functions
into palloc.h and mcxt.c, and adjust them to not be specific to
shared-memory allocation. Then invent palloc_mul(), palloc0_mul(),
palloc_mul_extended() to use these functions. Actually, the latter
use inlined copies to save one function call. repalloc_array() gets
similar treatment. I didn't bother trying to inline the calls for
repalloc0_array() though.
In v14 and v15, this also adds repalloc_extended(), which previously
was only available in v16 and up.
We need copies of all this in fe_memutils.[hc] as well, since that
module also provides palloc_array() etc.
Reported-by: Xint Code
Author: Tom Lane <tgl@sss.pgh.pa.us> Reviewed-by: Masahiko Sawada <sawada.mshk@gmail.com>
Backpatch-through: 14
Security: CVE-2026-6473
Tom Lane [Mon, 11 May 2026 12:13:51 +0000 (05:13 -0700)]
Add pg_add_size_overflow() and friends
Commit 600086f47 added (several bespoke copies of) size_t addition with
overflow checks to libpq. Move this to common/int.h, along with
its subtraction and multiplication counterparts.
pg_neg_size_overflow() is intentionally omitted; I'm not sure we should
add SSIZE_MAX to win32_port.h for the sake of a function with no
callers.
Back-patch of commit 8934f2136, done now because pg_add_size_overflow()
and friends are needed more widely for security fixes.
Author: Jacob Champion <jacob.champion@enterprisedb.com> Reviewed-by: Chao Li <li.evan.chao@gmail.com> Reviewed-by: Michael Paquier <michael@paquier.xyz>
Discussion: https://postgr.es/m/CAOYmi%2B%3D%2BpqUd2MUitvgW1pAJuXgG_TKCVc3_Ek7pe8z9nkf%2BAg%40mail.gmail.com
Backpatch-through: 14-18
Security: CVE-2026-6473
Michael Paquier [Mon, 11 May 2026 12:13:51 +0000 (05:13 -0700)]
Fix overflows with ts_headline()
The options "StartSel", "StopSel" and "FragmentDelimiter" given by a
caller of the SQL function ts_headline() have their lengths stored as
int16. When providing values larger than PG_INT16_MAX, it was possible
to overflow the length values stored, leading to incorrect behaviors in
generateHeadline(), in most cases translating to a crash.
Attempting to use values for these options larger than PG_INT16_MAX is
now blocked. Some test cases are added to cover our tracks.
Michael Paquier [Mon, 11 May 2026 12:13:51 +0000 (05:13 -0700)]
ltree: Fix overflows with lquery parsing
The lquery parser in contrib/ltree/ had two overflow problems:
- A single lquery level with many OR-separated variants (e.g.,
'label1|label2|...'), could cause an overflow of totallen, this being
stored as a uint16, meaning a maximum value of UINT16_MAX or 65k. Each
variant contributes MAXALIGN(LVAR_HDRSIZE + len) bytes. With enough
long variants, the value would wraparound. This would corrupt the data
written by LQL_NEXT(), leading to a stack corruption, most likely
translating into a crash, but it would allow incorrect memory access.
- numvar, labelled as a uint16, counts the number of OR-variants in a
single level, and it is incremented without bounds checking. With more
than PG_UINT16_MAX (65k) variants in a single level, and a minimum of
131kB of input data, it would wrap to 0. When a (wildcard) '*' is
used, this would change the query results silently.
For both issues, a set of overflows checks are added to guard against
these problematic patterns.
The first issue has been reported by the three people listed below,
affecting v16 and newer versions due to b1665bf01e5f. Its coding was
still unsafe in v14 and v15. The second issue affects all the stable
branches; I have bumped into while reviewing the code of the module.
Reported-by: Vergissmeinnicht <vergissmeinnichtzh@gmail.com> Reported-by: A1ex <alex000young@gmail.com> Reported-by: Jihe Wang <wangjihe.mail@gmail.com>
Author: Michael Paquier <michael@paquier.xyz>
Security: CVE-2026-6473
Backpatch-through: 14
projects / thirdparty / postgresql.git / log
