From: Antonin Houska <ah@cybertec.at>
To: SATYANARAYANA NARLAPURAM <satyanarlapuram@gmail.com>
cc: PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>,
    alvherre@kurilemu.de
Subject: Re: [PATCH] Compressed TOAST data corruption with REPACK CONCURRENTLY
In-reply-to: 
 <CAHg+QDeXb9HM2VGKXQedyCp52GzajJK5KOUdNi6oLjsS0nerQw@mail.gmail.com>
References: 
 <CAHg+QDeXb9HM2VGKXQedyCp52GzajJK5KOUdNi6oLjsS0nerQw@mail.gmail.com>
Comments: In-reply-to SATYANARAYANA NARLAPURAM <satyanarlapuram@gmail.com>
   message dated "Wed, 15 Apr 2026 23:13:42 -0700."
MIME-Version: 1.0
Content-Type: multipart/mixed; boundary="=-=-="
Date: Fri, 17 Apr 2026 17:45:52 +0200
Message-ID: <52301.1776440752@localhost>
Archived-At: 
 <https://www.postgresql.org/message-id/52301.1776440752%40localhost>
Precedence: bulk

--=-=-=
Content-Type: text/plain

SATYANARAYANA NARLAPURAM <satyanarlapuram@gmail.com> wrote:

> restore_tuple() in repack.c uses SET_VARSIZE() to reconstruct the varlena header when 
> reading back external attributes from the spill file. In this process, looks like the flag 
> SET_VARSIZE_COMPRESSED is silently lost. Because of this, when REPACK CONCURRENTLY
> run  any concurrently updated column whose value was TOAST-compressed ends up with raw 
> compressed bytes behind an "uncompressed" header returning garbled data on subsequent reads.
> It appears that existing tests are using random chars which are uncompressable.
> 
> Please find the attached 0001-Fix-restore_tuple-losing-varlena-compression-flag.patch to fix this. 
> Additionally I updated the existing repack_toast test to include the scenario I was talking about.

Good catch, thanks!

I'd slightly prefer to fix it w/o checking the varlena type, as
attached. However, your test fails to reproduce the issue here, so I'm not
able to verify the fix. I'll take a closer look early next week.

-- 
Antonin Houska
Web: https://www.cybertec-postgresql.com


--=-=-=
Content-Type: text/x-diff
Content-Disposition: attachment; filename=fix_restore_tuple.diff

diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index e2600a8888d..3c07e44d649 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -2736,7 +2736,7 @@ restore_tuple(BufFile *file, Relation relation, TupleTableSlot *slot)
 			varlensz = VARSIZE_ANY(&chunk_header);
 
 			value = palloc(varlensz);
-			SET_VARSIZE(value, VARSIZE_ANY(&chunk_header));
+			memcpy(value, &chunk_header, VARHDRSZ);
 			BufFileReadExact(file, (char *) value + VARHDRSZ, varlensz - VARHDRSZ);
 
 			slot->tts_values[i] = PointerGetDatum(value);

--=-=-=--