From: Tom Lane <tgl@sss.pgh.pa.us>
To: Peter Eisentraut <peter@eisentraut.org>
cc: Tatsuo Ishii <ishii@postgresql.org>, pgsql-hackers@postgresql.org
Subject: Re: warning: dereferencing type-punned pointer
In-reply-to: <1254018.1721844589@sss.pgh.pa.us>
References: <20240724.155525.366150353176322967.ishii@postgresql.org> <1218071.1721829945@sss.pgh.pa.us> <1fe7031f-3a25-40c3-b244-530ea42fc133@eisentraut.org> <1254018.1721844589@sss.pgh.pa.us>
Comments: In-reply-to Tom Lane <tgl@sss.pgh.pa.us>
	message dated "Wed, 24 Jul 2024 14:09:49 -0400"
MIME-Version: 1.0
Content-Type: text/plain; charset="UTF-8"
Content-ID: <1256313.1721845785.1@sss.pgh.pa.us>
Content-Transfer-Encoding: quoted-printable
Date: Wed, 24 Jul 2024 14:29:45 -0400
Message-ID: <1256314.1721845785@sss.pgh.pa.us>
Archived-At: <https://www.postgresql.org/message-id/1256314.1721845785%40sss.pgh.pa.us>
Precedence: bulk

BTW, I tried the same experiment of building without
-fno-strict-aliasing using gcc 11.4.1 (from RHEL9).
I see one more warning than Tatsuo-san did:

In file included from verify_heapam.c:18:
verify_heapam.c: In function =E2=80=98check_tuple_attribute=E2=80=99:
../../src/include/access/toast_internals.h:37:11: warning: dereferencing t=
ype-punned pointer will break strict-aliasing rules [-Wstrict-aliasing]
   37 |         (((toast_compress_header *) (ptr))->tcinfo >> VARLENA_EXTS=
IZE_BITS)
      |          ~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
verify_heapam.c:1693:24: note: in expansion of macro =E2=80=98TOAST_COMPRE=
SS_METHOD=E2=80=99
 1693 |                 cmid =3D TOAST_COMPRESS_METHOD(&toast_pointer);
      |                        ^~~~~~~~~~~~~~~~~~~~~

This looks a bit messy to fix: we surely don't want to pierce
the abstraction TOAST_COMPRESS_METHOD provides.  Perhaps
the toast_pointer local variable could be turned into a union
of struct varatt_external and toast_compress_header, but that
would impose a good deal of notational overhead on the rest
of this function.

The good news is that we get through check-world (although
I didn't try very many build options).

			regards, tom lane