From: Tom Lane <tgl@sss.pgh.pa.us>
To: Jim Jones <jim.jones@uni-muenster.de>
cc: Peter Smith <smithpb2250@gmail.com>,
        Peter Eisentraut <peter.eisentraut@enterprisedb.com>,
        Nikolay Samokhvalov <samokhvalov@gmail.com>,
        PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>,
        Andrey Borodin <amborodin86@gmail.com>
Subject: Re: [PATCH] Add pretty-printed XML output option
In-reply-to: <1c7cf9f0-14c3-0341-6894-8ca577bf325e@uni-muenster.de>
References: <2f5df461-dad8-6d7d-4568-08e10608a69b@uni-muenster.de>
 <CAHut+PtMHeBBZ2zv0Bk7-4KMkSrrxZ9PZxUCzhpwTGVk+8G3cw@mail.gmail.com>
 <2eac4d78-2b20-e8ec-54cc-22a88653a1cc@uni-muenster.de>
 <CAHut+PsgXxbzzsG+VqHUa4Kz3G4B-anNmF3-GENwuUdfMJtowA@mail.gmail.com>
 <fb2a543b-e615-d9ba-8402-909631f9b212@uni-muenster.de>
 <CAHut+Pvryt_+wA7TwKVaaF0e6Pm5LaPZQLKatp=78oFd8vMh3A@mail.gmail.com>
 <6cb296fe-4b49-3691-633e-2e2366997b04@uni-muenster.de>
 <3baf8f25-3bd7-7db5-0b31-e4c92b6269ba@enterprisedb.com>
 <1cbe1087-6acb-2b0a-dc9b-0a2d5b99e466@uni-muenster.de>
 <CAHut+PvBUTf9p+Y6+mpWpz59=3eL_EKQ+7Ysxpa9hYGdncDMdg@mail.gmail.com>
 <532259.1678393296@sss.pgh.pa.us> <906cfd1d-63a0-3cf5-7291-2dadb2e!
 149ff@uni-muenster.de> <803304.1678458776@sss.pgh.pa.us> <efe0b19b-!
 d41c-f8ab-f3b8-afb0108f3706@uni-muenster.de>
 <2752578.1678815625@sss.pgh.pa.us>
 <abd25443-ef6d-7b8a-c593-a2a991d3e5ce@uni-! ! muenster.de>
 <3057762.1678836321@sss.pgh.pa.us> <3622266.1678914816@sss.pgh.pa.us>
 <1c7cf9f0-14c3-0341-6894!
 -8ca577bf325e@uni-muenster.de>
Comments: In-reply-to Jim Jones <jim.jones@uni-muenster.de>
	message dated "Thu, 16 Mar 2023 20:07:06 +0100"
MIME-Version: 1.0
Content-Type: multipart/mixed; boundary="----- =_aaaaaaaaaa0"
Content-ID: <1358881.1747858770.0@sss.pgh.pa.us>
Date: Wed, 21 May 2025 16:20:17 -0400
Message-ID: <1358967.1747858817@sss.pgh.pa.us>
Archived-At: 
 <https://www.postgresql.org/message-id/1358967.1747858817%40sss.pgh.pa.us>
Precedence: bulk

------- =_aaaaaaaaaa0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <1358881.1747858770.1@sss.pgh.pa.us>

Jim Jones <jim.jones@uni-muenster.de> writes:
> In my environment (libxml2 v2.9.10 and Ubuntu 22.04) I couldn't 
> reproduce this memory leak.

Just when you thought it was safe to go back in the water ...

Experimenting with the improved valgrind leak detection code at [1],
I discovered that XMLSERIALIZE(... INDENT) has yet a different memory
leak problem.  It turns out that xmlDocSetRootElement() doesn't
merely install the given root node: it unlinks the document's old
root node and returns it to you.  If you don't free it, it's leaked
(for the session, since this is a malloc not palloc).  The amount of
leakage isn't that large, seems to be a couple hundred bytes per
iteration, which may explain why this escaped our notice in the
previous testing.  Still, it could add up under extensive usage.
So I think we need to apply the attached, back to PG 16.

			regards, tom lane

[1] https://www.postgresql.org/message-id/1295385.1747847681%40sss.pgh.pa.us


------- =_aaaaaaaaaa0
Content-Type: text/x-diff; name="fix-leakage-in-xmlserialize-indent.patch";
	charset="us-ascii"
Content-ID: <1358881.1747858770.2@sss.pgh.pa.us>
Content-Description: fix-leakage-in-xmlserialize-indent.patch
Content-Transfer-Encoding: quoted-printable

diff --git a/src/backend/utils/adt/xml.c b/src/backend/utils/adt/xml.c
index db8d0d6a7e8..73fd4fa090c 100644
--- a/src/backend/utils/adt/xml.c
+++ b/src/backend/utils/adt/xml.c
@@ -754,6 +754,7 @@ xmltotext_with_options(xmltype *data, XmlOptionType xm=
loption_arg, bool indent)
 			 * content nodes, and then iterate over the nodes.
 			 */
 			xmlNodePtr	root;
+			xmlNodePtr	oldroot;
 			xmlNodePtr	newline;
 =

 			root =3D xmlNewNode(NULL, (const xmlChar *) "content-root");
@@ -761,8 +762,14 @@ xmltotext_with_options(xmltype *data, XmlOptionType x=
mloption_arg, bool indent)
 				xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY,
 							"could not allocate xml node");
 =

-			/* This attaches root to doc, so we need not free it separately. */
-			xmlDocSetRootElement(doc, root);
+			/*
+			 * This attaches root to doc, so we need not free it separately...
+			 * but instead, we have to free the old root if there was one.
+			 */
+			oldroot =3D xmlDocSetRootElement(doc, root);
+			if (oldroot !=3D NULL)
+				xmlFreeNode(oldroot);
+
 			xmlAddChildList(root, content_nodes);
 =

 			/*

------- =_aaaaaaaaaa0--