Mailing-List: contact cygwin-help AT cygwin DOT com; run by ezmlm
List-Subscribe: <mailto:cygwin-subscribe AT cygwin DOT com>
List-Archive: <http://sourceware.org/ml/cygwin/>
List-Post: <mailto:cygwin AT cygwin DOT com>
List-Help: <mailto:cygwin-help AT cygwin DOT com>, <http://sourceware.org/ml/#faqs>
Sender: cygwin-owner AT cygwin DOT com
Mail-Followup-To: cygwin AT cygwin DOT com
Delivered-To: mailing list cygwin AT cygwin DOT com
Message-ID: <42F09150.9040702@opnet.com>
Date: Wed, 03 Aug 2005 11:41:36 +0200
From: Stein Somers <ssomers AT opnet DOT com>
User-Agent: Mozilla Thunderbird 1.0 (Windows/20041206)
MIME-Version: 1.0
To: cygwin AT cygwin DOT com
Subject: Re: bug report: abort in g++ 3.4.4 generated DLL & client
References: <42EFA9DC DOT 7030800 AT opnet DOT com> <42EFC5F3 DOT 9080708 AT familiehaase DOT de>
In-Reply-To: <42EFC5F3.9080708@familiehaase.de>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
X-IsSubscribed: yes

Gerrit P. Haase wrote:

> Do you mean the code compiled on the sane machine, where
> the binary runs ok when compiled with the older binutils, is crashing
> when simply copying the executable to another machine?

When copying the executable and dll, yes. That very same pair also 
crashes on the previously sane machine after it had its binutils 
upgraded. Unfortunately, I was too eager to try the upgrade to note the 
version it was upgraded from. The version was quite likely last upgraded 
at the end of June (but if gcc 3.4.4.1 wasn't out on cygwin yet then, 
that must be wrong).

> Isn't it a cygwin problem then and not an issue which version of gcc 
> or binutils
> is used?

It would seem like the/some C++ runtime library is part of binutils, the 
version of binutils doesn't (greatly) influence the code created by gcc, 
and that the latest binutils does strike back when invoked by the test 
program. However, Sysinternals' sublime Filemon reports that the only 
files accessed <http://www.sysinternals.com/Utilities/Filemon.html> 
(apart from the obvious executable, dll and stackdump) are:
C:\Programs\cygwin\bin\cygwin1.dll
C:\WINNT\system32\DNSAPI.dll
C:\WINNT\system32\EntApi.dll
C:\WINNT\system32\NETAPI32.dll
C:\WINNT\system32\NETRAP.dll
C:\WINNT\system32\NTDSAPI.dll
C:\WINNT\system32\PSAPI.DLL
C:\WINNT\system32\psapi.dll
C:\WINNT\system32\SAMLIB.dll
C:\WINNT\system32\SECUR32.DLL
C:\WINNT\system32\WS2_32.DLL
C:\WINNT\system32\WS2HELP.DLL
C:\WINNT\system32\WSOCK32.dll
C:\Documents and Settings\All Users\Application Data\Network 
Associates\BOPDATA\_Date-20050803_Time-103932000_EnterceptExceptions.dat
C:\Documents and Settings\All Users\Application Data\Network 
Associates\BOPDATA\_Date-20050803_Time-103932000_EnterceptRules.dat

the latter two stemming from McAfee VirusScan (even if On-Access scan is 
disabled!).

I guess that indeed the gcc and binutils versions are not at stake, that 
cygwin upgraded itself too on the sane machine, and that code created by 
gcc 3.4 exposes a problem in the latest cygwin release.

> When compiled on this machine, what is the cygwin version and what is
> the cygwin version on the box where it works ok?

Unfortunately, a new round of cygwin setup has crashed there and the 
cygwin environment appears messed up now. It's also very much in use so 
let's forget about that machine.

> I have not tried to build it with gcc-3.3.3, could someone with this
> version handy build an executable and send it to me, please?

I'll do so.

-- 
Stein


--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Problem reports:       http://cygwin.com/problems.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/