Content-Type: text/plain;
	charset=us-ascii
Mime-Version: 1.0 (Mac OS X Mail 16.0 \(3864.500.181\))
Subject: Re: [BUG] false positive in bt_index_check in case of short 4B
 varlena datum
From: Andrey Borodin <x4mmm@yandex-team.ru>
In-Reply-To: <7ckc7oka4bvafkf5bwlqs6ygrhlsbhz25ppozfch7zbuxcx3rf@e4pr4oqenalc>
Date: Fri, 1 May 2026 23:06:49 +0500
Cc: Alexander Korotkov <aekorotkov@gmail.com>,
 Alexander Lakhin <exclusion@gmail.com>,
 Michael Zhilin <m.zhilin@postgrespro.ru>,
 pgsql-bugs@postgresql.org,
 Yura Sokolov <y.sokolov@postgrespro.ru>
Content-Transfer-Encoding: quoted-printable
Message-Id: <DA0695AB-4A40-4272-BE64-D22F81F7713B@yandex-team.ru>
References: <76bc0dc9-4e43-4cd8-8eec-249b254ed1c9@postgrespro.ru>
 <8C83FCCA-2548-499A-8B1C-96C3D8ADB787@yandex-team.ru>
 <c6f376bf-9896-111f-ccb1-a19dc1381ad0@gmail.com>
 <C8DD7CB5-CFE8-4AF5-88D7-B9F441BB5A8C@yandex-team.ru>
 <0b535249-a00c-a38a-85f6-d5a38c62dd55@gmail.com>
 <CAPpHfdsAs-B7O2_=jGbF+BQzuW3kGSboY9CcxzEOWUgxxZCr5Q@mail.gmail.com>
 <e3a78e44-ec6c-5924-7832-63de3aad3e9b@gmail.com>
 <CAPpHfdtZxhgjZTpaLd--dCxzUOL6tN+aAZPBeTsGKHLkMQAZpw@mail.gmail.com>
 <CAPpHfdu412Z+jh5Oyc1yzWFM0+52h0kSGQKV4=xQ0aT0UeY2BA@mail.gmail.com>
 <CAPpHfdspqdzVdEU2qHMibKn8OwVuLoZZYvKm-m5Ffqy2aMXgGQ@mail.gmail.com>
 <7ckc7oka4bvafkf5bwlqs6ygrhlsbhz25ppozfch7zbuxcx3rf@e4pr4oqenalc>
To: Andres Freund <andres@anarazel.de>
Archived-At: <https://www.postgresql.org/message-id/DA0695AB-4A40-4272-BE64-D22F81F7713B%40yandex-team.ru>
Precedence: bulk


> On 1 May 2026, at 22:11, Andres Freund <andres@anarazel.de> wrote:
>=20
> While hacking on something, I added an assertion to VARSIZE() that the
> argument is actually a VARATT_4B (which it assumes). Worked =
everywhere, except
> for this caller: amcheck/regress fails, because sometimes the varlena =
is
> actually a short/1B varlena.
>=20
> Note that VARSIZE_4B on a short datum will give you completely bogus
> answers. E.g. in the case that failed the assertion, VARSIZE_1B() is =
2, but
> VARSIZE_4B(PTR) is 7681.

I remember the original code was taken from somewhere else because there
was already some instances like this:

/*
 * If value is above size target, and is of a compressible datatype,
 * try to compress it in-line.
 */
if (!VARATT_IS_EXTENDED(DatumGetPointer(untoasted_values[i])) &&
VARSIZE(DatumGetPointer(untoasted_values[i])) > TOAST_INDEX_TARGET &&
(att->attstorage =3D=3D TYPSTORAGE_EXTENDED ||
att->attstorage =3D=3D TYPSTORAGE_MAIN))
{

I don't have VARATT_IS_EXTENDED vs VARATT_IS_COMPRESSED vs =
VARATT_IS_SHORT
business in my warm cache right away, but I'll try to remember what it =
means soon.


Best regards, Andrey Borodin.=