Date: Tue, 7 Jul 2026 03:41:57 +0500
From: Andrey Rachitskiy <pl0h0yp1@gmail.com>
Cc: PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>, Nikolay Shaplov
 <dhyan@nataraj.su>
Subject: [PATCH] Limit PL/Perl scalar copies to work_mem
Message-ID: <20260707034157.5e88bf34@pg-ThinkPad-T14-Gen-4>
MIME-Version: 1.0
Content-Type: multipart/mixed; boundary="MP_/sDTAhto+gkG/PMs8Alw6m8e"
Archived-At: 
 <https://www.postgresql.org/message-id/20260707034157.5e88bf34%40pg-ThinkPad-T14-Gen-4>
Precedence: bulk

--MP_/sDTAhto+gkG/PMs8Alw6m8e
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Content-Disposition: inline


Hi, Hackers!

When a PL/Perl function returns a large text value, sv2cstr() copies the
entire Perl string into backend memory with no size check.  The helper
is used on the path from Perl return values and SPI arguments to
PostgreSQL text datums; it simply palloc()s a copy after SvPVutf8().
A user who is allowed to create untrusted PL/Perl functions can
therefore force the backend to allocate strings far larger than any
session limit. On a memory-constrained host this can get the backend
process killed by the OOM killer (SIGKILL) rather than raising a
catchable PostgreSQL error.

Reproducer (unpatched master, plperl enabled):

CREATE FUNCTION perl_huge_text() RETURNS text LANGUAGE plperl
AS $$ return 'x' x (1024 * 1024 * 1024);
  $$;
  SELECT perl_huge_text();

On a container limited to about 768MB RAM, CREATE FUNCTION alone is
enough to lose the backend:

  LOG:  client backend (PID ...) was terminated by signal 9: Killed
  DETAIL:  Failed process was running: CREATE OR REPLACE FUNCTION ...

With plenty of free RAM the same code may succeed instead, which I think
shows missing enforcement rather than an intentional "no limit" design:
other PL/Perl paths already enforce bounds (MAXDIM, AV_SIZE_MAX for SPI
results, max_stack_depth in recursive conversion), but sv2cstr() had
none.

This patch rejects Perl strings larger than work_mem * 1024 bytes,
capped by MaxAllocSize, before copying them through sv2cstr().  That
follows the same work_mem-based pattern used elsewhere in the backend
for per-query working storage.  The check is done after SvPVutf8() has
reported the length but before utf_u2e() allocates the
database-encoding copy. A plperl regression test returns a 16MB string
with the default 4MB work_mem and expects:

  ERROR:  Perl value exceeds maximum allowed size (4194304 bytes)
  HINT:  Increase work_mem or reduce the result size.

Legitimate functions that need to move more data can raise work_mem for
the session, consistent with other operations bounded by that GUC.

Comments welcome.

-- 
Regards,
Andrey Rachitskiy


--MP_/sDTAhto+gkG/PMs8Alw6m8e
Content-Type: text/x-patch
Content-Transfer-Encoding: 7bit
Content-Disposition: attachment; filename=0001-plperl-limit-scalar-size.patch

From 50f5b04fd79548e65081db6a52478ee7faeb629b Mon Sep 17 00:00:00 2001
From: Andrey Rachitskiy <pl0h0yp1@gmail.com>
Date: Mon, 06 Jul 2026 22:13:53 +0000
Subject: [PATCH] Limit PL/Perl scalar copies to work_mem

When a PL/Perl function returns a very large text value, sv2cstr() copies
the entire Perl string into backend memory with no size check.  A user
with permission to create untrusted PL/Perl functions can return strings
far larger than work_mem and risk getting the backend killed by the OOM
killer.

Reject Perl strings larger than work_mem * 1024 bytes, capped by
MaxAllocSize, before copying them through sv2cstr().  This follows the
same work_mem-based limit pattern used elsewhere in the backend.

Add a plperl regression test that attempts to return a 16MB string with
the default 4MB work_mem setting.

Author: Andrey Rachitskiy <pl0h0yp1@gmail.com>
---
 src/pl/plperl/expected/plperl.out |  8 +++++++
 src/pl/plperl/plperl.h            | 34 ++++++++++++++++++++++++++++++
 src/pl/plperl/sql/plperl.sql      |  7 ++++++
 3 files changed, 49 insertions(+)
diff --git a/src/pl/plperl/expected/plperl.out b/src/pl/plperl/expected/plperl.out
index e3d7c88..50c788b 100644
--- a/src/pl/plperl/expected/plperl.out
+++ b/src/pl/plperl/expected/plperl.out
@@ -792,3 +792,11 @@ SELECT self_modify(42);
          126
 (1 row)
 
+-- oversized text results are rejected at the PL boundary
+CREATE OR REPLACE FUNCTION perl_oversized_text() RETURNS text AS $$
+	return 'x' x (16 * 1024 * 1024);
+$$ LANGUAGE plperl;
+SELECT perl_oversized_text();
+ERROR:  Perl value exceeds maximum allowed size (4194304 bytes)
+HINT:  Increase work_mem or reduce the result size.
+CONTEXT:  PL/Perl function "perl_oversized_text"
diff --git a/src/pl/plperl/plperl.h b/src/pl/plperl/plperl.h
index 4c03f9e..1ccce6d 100644
--- a/src/pl/plperl/plperl.h
+++ b/src/pl/plperl/plperl.h
@@ -17,6 +17,8 @@
 
 /* defines free() by way of system headers, so must be included before perl.h */
 #include "mb/pg_wchar.h"
+#include "miscadmin.h"
+#include "utils/memutils.h"
 
 /*
  * Pull in Perl headers via a wrapper header, to control the scope of
@@ -40,6 +42,36 @@ char	   *plperl_sv_to_literal(SV *, char *);
 void		plperl_util_elog(int level, SV *msg);
 
 
+/*
+ * Maximum byte size for a Perl scalar copied through sv2cstr().
+ *
+ * This follows the same work_mem * 1024 pattern used elsewhere in the
+ * backend (e.g. reorderbuffer.c, nodeHash.c) and is capped by MaxAllocSize.
+ */
+static inline Size
+plperl_max_scalar_bytes(void)
+{
+	Size		limit = (Size) work_mem * (Size) 1024;
+
+	return Min(limit, MaxAllocSize - 1);
+}
+
+/*
+ * Reject Perl strings that are too large to copy into backend memory.
+ */
+static inline void
+plperl_check_sv_length(STRLEN len)
+{
+	Size		max_len = plperl_max_scalar_bytes();
+
+	if ((Size) len > max_len)
+		erereport(ERROR,
+				(errcode(ERRCODE_PROGRAM_LIMIT_EXCEEDED),
+				 errmsg("Perl value exceeds maximum allowed size (%zu bytes)",
+						max_len),
+			 errhint("Increase work_mem or reduce the result size.")));
+}
+
 /* helper functions */
 
 /*
@@ -127,6 +159,8 @@ sv2cstr(SV *sv)
 	else
 		val = SvPVutf8(sv, len);
 
+	plperl_check_sv_length(len);
+
 	/*
 	 * Now convert to database encoding.  We use perl's length in the event we
 	 * had an embedded null byte to ensure we error out properly.
diff --git a/src/pl/plperl/sql/plperl.sql b/src/pl/plperl/sql/plperl.sql
index bb0b8ce..0470a2b 100644
--- a/src/pl/plperl/sql/plperl.sql
+++ b/src/pl/plperl/sql/plperl.sql
@@ -521,3 +521,10 @@ $$ LANGUAGE plperl;
 
 SELECT self_modify(42);
 SELECT self_modify(42);
+
+-- oversized text results are rejected at the PL boundary
+CREATE OR REPLACE FUNCTION perl_oversized_text() RETURNS text AS $$
+	return 'x' x (16 * 1024 * 1024);
+$$ LANGUAGE plperl;
+
+SELECT perl_oversized_text();

--MP_/sDTAhto+gkG/PMs8Alw6m8e--