Problem with setlocale (found in libecpg) [accessing a memory location after freeing it]

From: Christof Petig <christof(at)petig-baender(dot)de>
To: Michael Meskes <meskes(at)postgresql(dot)org>
Cc: pgsql-patches(at)postgresql(dot)org
Subject: Problem with setlocale (found in libecpg) [accessing a memory location after freeing it]
Date: 2001-09-24 07:18:42
Message-ID: 3BAEDE51.EB229159@petig-baender.de
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers pgsql-patches

Hello,

well at first I could not believe what I was seeing ...

Look at the following code (ecpg/lib/execute.c):

const char *locale=setlocale(LC_NUMERIC, NULL);
setlocale(LC_NUMERIC, "C");
[....]
setlocale(LC_NUMERIC, locale);

Well at least on glibc-2.2 it seems that setlocale retuns a pointer to
malloced memory, and frees this pointer on subsequent calls to
setlocale. This is standard conformant and has good reasons. But used as
above it is lethal (but not lethal enough to be easily recognized). So
the content locale points to is freed by the second call to setlocale.

The remedy is easy (given that _no other_ call to setlocale happens
inbetween ...)

const char *locale=setlocale(LC_NUMERIC, "C");
[...]
setlocale(LC_NUMERIC, locale);

So I would kindly ask you to take a second look at every invokation of
setlocale. And to apply the following patch.

Yours
Christof

Attachment Content-Type Size
B_setlocale.patch.gz application/x-gzip 343 bytes

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Dave Page 2001-09-24 07:37:20 Re: [ODBC] UTF-8 support
Previous Message Tatsuo Ishii 2001-09-24 07:12:59 Re: UTF-8 support

Browse pgsql-patches by date

  From Date Subject
Next Message Justin Clift 2001-09-24 15:02:52 Re: Makefiles for building with mingw32 on win32
Previous Message Gerhard Häring 2001-09-24 04:31:55 Re: [PATCHES] Makefiles for building with mingw32 on win32