Mailing-List: contact cygwin-help AT sourceware DOT cygnus DOT com; run by ezmlm
List-Subscribe: <mailto:cygwin-subscribe AT sources DOT redhat DOT com>
List-Archive: <http://sources.redhat.com/ml/cygwin/>
List-Post: <mailto:cygwin AT sources DOT redhat DOT com>
List-Help: <mailto:cygwin-help AT sources DOT redhat DOT com>, <http://sources.redhat.com/ml/#faqs>
Sender: cygwin-owner AT sources DOT redhat DOT com
Delivered-To: mailing list cygwin AT sources DOT redhat DOT com
Date: Tue, 9 Oct 2001 21:29:08 -0400
From: "'cgf AT redhat DOT com'" <cgf AT redhat DOT com>
To: cygwin AT cygwin DOT com
Subject: Re: Perl 5.7.2 (GDB breaks in 1.3.3-2 on certain Win2000 machines )
Message-ID: <20011009212908.A30223@redhat.com>
Reply-To: cygwin AT cygwin DOT com
Mail-Followup-To: cygwin AT cygwin DOT com
References: <30BDEE18E474D511BB0A00D0B7BB87AF09294A AT camel DOT mha DOT ca>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <30BDEE18E474D511BB0A00D0B7BB87AF09294A@camel.mha.ca>
User-Agent: Mutt/1.3.21i

On Tue, Oct 09, 2001 at 06:03:56PM -0700, Kris Erickson wrote:
>Since the error is occurring in one of the wrappers or libraries that
>I haven't yet figured out how to debug (_libkernel32_a_iname),

That would be a Windows DLL, possibly kernel32.dll.  Since there are
no debugging symbols in this DLL and since you don't, AFAIK,
have the sources to this DLL it's going to be very hard to debug it.

I would submit that it should not be dying there at all.  As I
speculated in another message, this sounds like a Windows bug to me.
Which is small consolation, of course...

>I have solved the solution by simply removing the call the
>NetUserGetInfo from uinfo.cc (which is effectively what was occuring in
>1.3.2-1).  I will look into this further, but I thought since there was
>such a mass of posts about this I would report what I had found...

One of the mass of posts that you apparently skipped is the one where I
already reported your same findings.

If we are all going to be reporting findings on this (and we all should
be doing this), it is important that we should be advancing our
knowledge of the problem rather than reporting the same thing.

FWIW, in case anyone is wondering, Corinna made the changes to uinfo.cc
that are causing this problem and she obviously did so for a reason.
Simply removing the lines isn't an option.  If Cygwin is debuggable without
those lines then that pinpoints the problem but it is not a solution.

This is probably obvious to most, but I wanted to mention this in case
someone thought that the problem was close to being solved.

What has happened is that after a couple of weeks of complaints, with
myself and others suggesting that cygwin needs to be debugged, we have
finally gotten to the point where some have built a debugging version of
the cygwin DLL and we have reached the initial steps of looking at the
problem -- we've identified the line in cygwin which calls a Windows
function that crashes.

I'm glad that we have gotten to this point but we will need some insight
into what is causing the crash in NetUserGetInfo.  I've searched both
google and MSDN but I haven't seen anything interesting on the subject.

So, I would suggest that the next step is to analyze the input to
this function and see if you can determine what is tickling this
apparent Windows bug.

I've made some changes to uinfo.cc which are in tonight's snapshot.  The
changes probably won't fix the problem but it will make it a little easier
to debug since it separates the two calls to NetUserGetInfo.  This will
at least allow us to determine which of the two calls is dying.

cgf

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/