Re: recent --with-libxml support - Mailing list pgsql-hackers

From Tom Lane
Subject Re: recent --with-libxml support
Date
Msg-id 21235.1166923360@sss.pgh.pa.us
Whole thread Raw
In response to Re: recent --with-libxml support  (Jeremy Drake <pgsql@jdrake.com>)
Responses Re: recent --with-libxml support  (Jeremy Drake <pgsql@jdrake.com>)
List pgsql-hackers
Jeremy Drake <pgsql@jdrake.com> writes:
> On Fri, 22 Dec 2006, Tom Lane wrote:
>> Jeremy Drake <pgsql@jdrake.com> writes:
>>> As seen, I needed to add an include dir for configure to pass.  However,
>>> make check fails now with the backend crashing.  This can be seen in the
>>> buildfarm results for mongoose.
>> 
>> Can you provide a stack trace for that crash?

> #0  0xb7c4dc85 in memcpy () from /lib/tls/libc.so.6
> #1  0x08190f59 in appendBinaryStringInfo (str=0xbfd87f90,
>     data=0x841ffc0 "qux", datalen=138543040) at stringinfo.c:192
> #2  0x0828377f in map_sql_identifier_to_xml_name (ident=0x841ffc0 "qux",
>     fully_escaped=0 '\0') at xml.c:933
> #3  0x0811ce83 in transformXmlExpr (pstate=0x84202b8, x=0x8420034)
>     at parse_expr.c:1426
> #4  0x0811ac91 in transformExpr (pstate=0x84202b8, expr=0x8420034)
>     at parse_expr.c:238

Ah-hah, I've sussed it.  sqlchar_to_unicode() calls the
mb2wchar_with_len converters, which are defined to return a *null
terminated* pg_wchar string.  So even if you only ask for the conversion
of a single character, you need a 2-pg_wchar array to hold the result,
and that wasn't being allocated.  So the code was clobbering a word of
stack space that evidently is critical on some machines and not others.

Fix committed, please see if it works for you now.
        regards, tom lane


pgsql-hackers by date:

Previous
From: "Andrew Dunstan"
Date:
Subject: Re: libxml on unixware
Next
From: Tom Lane
Date:
Subject: Loose ends in PG XML patch