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
In-Reply-To:	<GCEBKEBLOHKOGJEBHNNECELOEBAB.vbandke@bsp-gmbh.com>
To:	"Volker Bandke" <vbandke AT bsp-gmbh DOT com>
Cc:	cygwin AT cygwin DOT com
Subject:	Re: Cygwin1.dll problem with Hyperthreaded machines (reprise)
MIME-Version:	1.0
Message-ID:	<OF9F4AC402.FAE935A6-ON85256EE1.005586B2-85256EE1.00563631@abinitio.com>
From:	Chuck McDevitt <cmcdevitt AT ABINITIO DOT COM>
Date:	Fri, 30 Jul 2004 11:41:39 -0400
X-IsSubscribed:	yes

"Volker Bandke" <vbandke AT bsp-gmbh DOT com> 
Sent by: cygwin-owner AT cygwin DOT com
07/30/2004 09:30 AM

To
<cygwin AT cygwin DOT com>
cc

Subject
Cygwin1.dll problem with Hyperthreaded machines (reprise)






 
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi everyone,


This msg relates to an earlier one at 


http://sources.redhat.com/ml/cygwin/2004-04/msg00223.html

 
with title "Cygwin1.dll problem with Hyperthreaded machines"


I have a ASUS P4-P800E DELUXE Motherboard in my PC, with an Intel 3.2
GHz HT capable CPU.  The same symptoms as described for Cygwin 1.5.9
still show up for Cygwin 1.5.10-3, IOW:

When the machine is somewhat loaded, the simple loop like


#!/bin/bash
declare -ix ITER;
while :; do
  `let ITER=0`
  mypwd=`pwd`
  if test -z "$mypwd"; then
     echo "Test failed, mypwd is empty"
     exit 1
  else
     ITER=ITER+1
     echo "Next iteration is $ITER"
  fi
done

fails under Windows XP professional, Service Pack 1a.  Sometimes
there are about a hundred iterations, sometimes 15000, but fail it
will.


I have looked through the archives, but haven't found the
(Re-)solution.   Why is this a problem?  Large ./configure && make &&
make install jobs fail at indeterminable points in time, with the
same symptoms 
 
 
 
 -------------------------------------------------------------------       
 
We've been trying to track down this problem, but so far have not 
succeeded.  The problem is not isolated to hyperthreaded machines,
but can be reproduced on multi-CPU machines as well.

All recent version of Cygwin have this bug.  From what I can tell,
1.3.x versions of Cygwin1.dll did not have this bug (probably 
that was before cygwin1.dll was multithreaded).

Loops like the one you show will fail saying "Test failed, mypwd is empty"
other times the test will simply hang (making it not obvious that 
the test failed.

I've also been trying to convince my company to donate a
hyperthreaded machine to someone on the cygwin project, but so far,
management's reaction is "How do we know anyone will work on 
the problem, even if we give them a machine?"
And I'm not even sure who we would send the machine to.

Charles McDevitt
Ab Initio Software







--
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/

- Raw text -