From: Tom Lane <tgl@sss.pgh.pa.us>
To: Michael Paquier <michael@paquier.xyz>
cc: Robin Haberkorn <haberkorn@b1-systems.de>,
        Jim Jones <jim.jones@uni-muenster.de>,
 pgsql-bugs@lists.postgresql.org,
        maralist86@mail.ru
Subject: Re: BUG #18943: Return value of a function 'xmlBufferCreate' is
 dereferenced at xpath.c:177 without checking for NUL
In-reply-to: <aGz_ssvep-q7oM-M@paquier.xyz>
References: <aD53hVjiW-9C29VT@paquier.xyz> <861593.1748970933@sss.pgh.pa.us>
 <aEEingzOta_S_Nu7@paquier.xyz>
 <CAPLXN34Dr3Gbi+xJ6BgCeTyBJkMVe3cn7qxoADV72rC9ZHeBtQ@mail.gmail.com>
 <d2410ca0-c0dd-4f63-9e70-3d7a62a5d705@uni-muenster.de>
 <b35e2342-0f02-4365-94cf-55052ac9bda1@uni-muenster.de>
 <aEKCoNIfLxjyKY3r@paquier.xyz>
 <31f3480e-cd7d-4021-b392-87922572cc37@uni-muenster.de>
 <aETzMep2fGfB0AIp@paquier.xyz> <DB6KVQ60OJ8X.A8LWANY82NLG@b1-systems.de>
 <aGz_ssvep-q7oM-M@paquier.xyz>
Comments: In-reply-to Michael Paquier <michael@paquier.xyz>
	message dated "Tue, 08 Jul 2025 20:23:30 +0900"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <689494.1751981797.1@sss.pgh.pa.us>
Date: Tue, 08 Jul 2025 09:36:37 -0400
Message-ID: <689495.1751981797@sss.pgh.pa.us>
Archived-At: 
 <https://www.postgresql.org/message-id/689495.1751981797%40sss.pgh.pa.us>
Precedence: bulk

Michael Paquier <michael@paquier.xyz> writes:
> On Tue, Jul 08, 2025 at 09:49:20AM +0000, Robin Haberkorn wrote:
>> I know this has already been committed, but why are we still using
>> PG_XML_STRICTNESS_LEGACY in xpath.c? As we are always checking
>> pg_xml_error_occurred() this should no longer be necessary.

> Are you sure that you can do that?

The comment in xml_errorHandler() argues

     * Legacy error handling mode.  err_occurred is never set, we just add the
     * message to err_buf.  This mode exists because the xml2 contrib module
     * uses our error-handling infrastructure, but we don't want to change its
     * behaviour since it's deprecated anyway.  This is also why we don't
     * distinguish between notices, warnings and errors here --- the old-style
     * generic error handler wouldn't have done that either.

So switching to _ALL (or even _WELL_FORMED) mode would result in
nontrivial differences in the behavior of xpath.c's functions with
bad input.  Maybe that's a reasonable thing to do, but it's a
question of user-visible behavior not just code cleanliness.

			regards, tom lane