RE: [xml] segfault with libxml2 and python extension

["William M. Brack" <wbrack mmm com hk>]

I can reproduce it on Linux; I thought I could quickly find and fix
it, so I didn't respond yesterday.  Unfortunately, it has proven to
be somewhat "challenging" - minor edits to the input file radically
change the symptoms of failure.

I'll advise as soon as I've succeeded in isolating it.  In the
meantime, if anyone else finds what's happening please keep us
posted.

Regards,

Bill

Stéphane Bidoul said:
> The problem does not show up on
> windows with python 2.3.3 and libxml2 2.6.9.
> So I can't debug it myself. Anyone else?
>
> -sbi
>
> -----Original Message-----
> From: xml-bounces gnome org [mailto:xml-bounces gnome org]On Behalf
> Of
> Robert Zimmermann
> Sent: jeudi 3 juin 2004 11:39
> To: 'xml gnome org'
> Subject: [xml] segfault with libxml2 and python extension
>
>
> Hi,
>
> While doing some tests with the python sax interface and
> drv_libxml2.py I
> ran in a problem with some of my test-xml files.
> I have attached a simple, minimal python script which triggers the
> problem
> together with two xml-files.
> "REC.xml" is the original one and "REC_test2.xml" is a bit
> shortened, as I
> thought some xml speciality would be the cause.
>
> The problem is an segfault after a long while. I ran an gdb session
> on it
> and it showed an infinite loop.
> Here's a shortened backtrace of gdb:
> """
> #0  0x4037e039 in chunk_alloc () from /lib/i686/libc.so.6
> #1  0x4037deb2 in malloc () from /lib/i686/libc.so.6
> #2  0x408d3c52 in xmlStrndup (cur=0x8113e2b "href", len=16) at
> xmlstring.c:45
> #3  0x408d3ce2 in xmlStrdup (cur=0x1 <Address 0x1 out of bounds>) at
> xmlstring.c:71
> #4  0x4088909b in xmlNewDocProp (doc=0x0, name=0x8113e2b "href",
> value=0x0)
> at tree.c:1946
> #5  0x4088ad33 in xmlCopyProp (target=0x8afef180, cur=0x80b3800) at
> tree.c:3576
> #6  0x4088afad in xmlCopyPropList (target=0x8afef180, cur=0x80b3800)
> at
> tree.c:3679
> #7  0x4088b252 in xmlStaticCopyNode (node=0x810f548, doc=0x0,
> parent=0x8afeeff8, extended=1) at tree.c:3829
> #8  0x4088b3c6 in xmlStaticCopyNodeList (node=0x810f548, doc=0x0,
> parent=0x8afeeff8) at tree.c:3879
> #9  0x4088ae5c in xmlCopyProp (target=0x8afeefa8, cur=0x80b3800) at
> tree.c:3635
> #10 0x4088afad in xmlCopyPropList (target=0x8afeefa8, cur=0x80b3800)
> at
> tree.c:3679
> #11 0x4088b252 in xmlStaticCopyNode (node=0x810f548, doc=0x0,
> parent=0x8afeee20, extended=1) at tree.c:3829
> #12 0x4088b3c6 in xmlStaticCopyNodeList (node=0x810f548, doc=0x0,
> parent=0x8afeee20) at tree.c:3879
> """
>
> libxml2.parseFile('REC_test2.xml') does not trigger this problem,
> nor does
> xmllint.
> So I assume the problem is somewhere in sax-parsing together with
> the
> python-extension.
>
> My testing environment is:
>  Linux with gcc 3.2
>  libxml2 2.6.9
>  libxml2-python 2.6.9
>  libxslt 1.1.7
>  python 2.3.3
>
> Could someone try to confirm this?
>
> Thanks,
>  Robert
>
>
>
> Robert Zimmermann
> Softwaredevelopment
> WEB.DE AG