From: Chao Li <li.evan.chao@gmail.com>
Content-Type: multipart/mixed;
	boundary="Apple-Mail=_2B383A20-221B-4189-9E95-17E53660CD4D"
Mime-Version: 1.0 (Mac OS X Mail 16.0 \(3864.600.51.1.1\))
Subject: Fix NO_OOM handling in MemoryContextAllocAligned
Message-Id: <07DAC4C3-120D-4F3C-8FEE-BA236F7E9C1D@gmail.com>
Date: Mon, 3 Aug 2026 16:49:48 +0800
To: PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>
Archived-At: 
 <https://www.postgresql.org/message-id/07DAC4C3-120D-4F3C-8FEE-BA236F7E9C1D%40gmail.com>
Precedence: bulk


--Apple-Mail=_2B383A20-221B-4189-9E95-17E53660CD4D
Content-Transfer-Encoding: quoted-printable
Content-Type: text/plain;
	charset=utf-8

Hi,

This has been in my notes for a while. MemoryContextAllocAligned() calls =
MemoryContextAllocExtended() and passes =E2=80=9Cflags" with only =
MCXT_ALLOC_ZERO removed. When =E2=80=9Cflags" includes =
MCXT_ALLOC_NO_OOM, MemoryContextAllocExtended() may return NULL, but =
MemoryContextAllocAligned() doesn't handle the returned NULL, which may
lead to a crash.

AlignedAllocRealloc() wraps MemoryContextAllocAligned() and handles a =
NULL return from it, which I think further supports that this is a bug.

I didn't find an in-tree call site that passes MCXT_ALLOC_NO_OOM to =
MemoryContextAllocAligned() or palloc_aligned(), so this cannot be =
triggered from current core code. However, a third-party extension could =
hit the bug.

The attached patch is a quick fix that adds a NULL check.

Best regards,
--
Chao Li (Evan)
HighGo Software Co., Ltd.
https://www.highgo.com/


--Apple-Mail=_2B383A20-221B-4189-9E95-17E53660CD4D
Content-Disposition: attachment;
	filename=v1-0001-Fix-NO_OOM-handling-in-MemoryContextAllocAligned.patch
Content-Type: application/octet-stream;
	x-unix-mode=0644;
	name="v1-0001-Fix-NO_OOM-handling-in-MemoryContextAllocAligned.patch"
Content-Transfer-Encoding: quoted-printable

=46rom=20e9e7bf36bf8122277cee7b6b74a9cba8bc10b17f=20Mon=20Sep=2017=20=
00:00:00=202001=0AFrom:=20"Chao=20Li=20(Evan)"=20<lic@highgo.com>=0A=
Date:=20Mon,=203=20Aug=202026=2016:21:23=20+0800=0ASubject:=20[PATCH=20=
v1]=20Fix=20NO_OOM=20handling=20in=20MemoryContextAllocAligned=0A=0A=
MemoryContextAllocAligned()=20passes=20its=20flags=20through=20to=0A=
MemoryContextAllocExtended(),=20except=20for=20MCXT_ALLOC_ZERO.=20=
Therefore,=20when=0AMCXT_ALLOC_NO_OOM=20is=20specified,=20the=20=
underlying=20allocation=20can=20return=20NULL.=0A=0AThe=20=
aligned-allocation=20path=20did=20not=20check=20for=20that=20case=20=
before=20computing=20the=0Aaligned=20pointer=20and=20setting=20up=20the=20=
redirection=20MemoryChunk.=20That=20could=20perform=0Apointer=20=
arithmetic=20from=20NULL=20and=20then=20access=20an=20invalid=20chunk=20=
header,=20instead=0Aof=20returning=20NULL=20to=20the=20caller.=0A=0A=
Return=20NULL=20immediately=20when=20the=20underlying=20allocation=20=
fails.=20This=20also=20lets=0AAlignedAllocRealloc()=20handle=20the=20=
failure=20as=20intended=20when=20repalloc_extended()=0Ais=20used=20on=20=
an=20aligned=20allocation=20with=20MCXT_ALLOC_NO_OOM.=0A=0AAuthor:=20=
Chao=20Li=20<lic@highgo.com>=0A---=0A=20src/backend/utils/mmgr/mcxt.c=20=
|=202=20++=0A=201=20file=20changed,=202=20insertions(+)=0A=0Adiff=20=
--git=20a/src/backend/utils/mmgr/mcxt.c=20=
b/src/backend/utils/mmgr/mcxt.c=0Aindex=20930fc457328..fe5c2d8e01b=20=
100644=0A---=20a/src/backend/utils/mmgr/mcxt.c=0A+++=20=
b/src/backend/utils/mmgr/mcxt.c=0A@@=20-1540,6=20+1540,8=20@@=20=
MemoryContextAllocAligned(MemoryContext=20context,=0A=20=09=20*/=0A=20=09=
unaligned=20=3D=20MemoryContextAllocExtended(context,=20alloc_size,=0A=20=
=09=09=09=09=09=09=09=09=09=09=20=20=20flags=20&=20~MCXT_ALLOC_ZERO);=0A=
+=09if=20(unlikely(unaligned=20=3D=3D=20NULL))=0A+=09=09return=20NULL;=0A=
=20=0A=20=09/*=20compute=20the=20aligned=20pointer=20*/=0A=20=09aligned=20=
=3D=20(void=20*)=20TYPEALIGN(alignto,=20(char=20*)=20unaligned=20+=0A--=20=
=0A2.50.1=20(Apple=20Git-155)=0A=0A=

--Apple-Mail=_2B383A20-221B-4189-9E95-17E53660CD4D--