Mailing-List: contact cygwin-help AT cygwin DOT com; run by ezmlm
List-Subscribe: <mailto:cygwin-subscribe AT cygwin DOT com>
List-Archive: <http://sources.redhat.com/ml/cygwin/>
List-Post: <mailto:cygwin AT cygwin DOT com>
List-Help: <mailto:cygwin-help AT cygwin DOT com>, <http://sources.redhat.com/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
X-pair-Authenticated: 24.126.73.164
Message-ID: <40606A7E.1010409@kegel.com>
Date: Tue, 23 Mar 2004 08:49:02 -0800
From: Dan Kegel <dank AT kegel DOT com>
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.6) Gecko/20040113
MIME-Version: 1.0
To: cygwin AT cygwin DOT com
Subject: Re: Minimal test case for Make crash bug
References: <405C00BF DOT 3090507 AT kegel DOT com> <405C06C4 DOT 7030508 AT kegel DOT com> <405C6C4B DOT 90409 AT kegel DOT com> <405C939E DOT 5020909 AT kegel DOT com> <405D1ACD DOT 4060809 AT kegel DOT com> <405D3CEA DOT 3090006 AT kegel DOT com> <405DE73D DOT 4070100 AT kegel DOT com> <405E0B96 DOT 1010407 AT kegel DOT com> <405E8EF8 DOT 2070706 AT kegel DOT com>
In-Reply-To: <405E8EF8.2070706@kegel.com>
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit

Dan Kegel wrote:
>>> (gdb) p alloc_sz
>>> $3 = 0
>>
>> Bingo.  That's the problem.  I've checked in a fix for this and have 
>> uploaded a new snapshot.  Can you confirm that it fixes the problem?
> 
> Yep, fixes my little test case.
> I'll run more exhaustive tests tomorrow.

It handled my real-world case well, but when I increased the stress
level, I ran into a problem (the following is on my wife's beefy
winXP machine, same one as before):

Running two toolchain builds in parallel, both in the background,
and with tail -f in the foreground, I pressed ^C to kill the tail -f,
and one of the toolchain builds aborted with the message:

C:\cygwin\opt\crosstool\powerpc-860-linux-gnu\gcc-3.3.3-glibc-2.3.2\lib\gcc-lib\
powerpc-860-linux-gnu\3.3.3\collect2.exe (684): *** fork: can't reserve 
memory for stack 0x30000 - 0x230000, Win32 error 487
       5 [main] collect2 2932 sync_with_child: child 684(0x6EC) died 
before initialization with status code 0x1

Sounds rather like some stress testing is in order, but I don't know
if it should hold up your next release.

One further observation:
on winMe, you can only open a dozen or so /dev/null fds in a single
process before open starts returning -1.  Shouldn't matter, but it is
strange.

- Dan

--
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/