Mailing-List:	contact cygwin-apps-help AT sourceware DOT cygnus DOT com; run by ezmlm
list-help:	<mailto:cygwin-apps-help AT sources DOT redhat DOT com>
list-post:	<mailto:cygwin-apps AT sourceware DOT cygnus DOT com>
Sender:	cygwin-apps-owner AT sourceware DOT cygnus DOT com
Delivered-To:	mailing list cygwin-apps AT sources DOT redhat DOT com
content-class:	urn:content-classes:message
From:	"Robert Collins" <robert DOT collins AT itdomain DOT com DOT au>
To:	<cygwin-apps AT sourceware DOT cygnus DOT com>
Subject:	RE: stackdump revisited
Date:	Fri, 7 Jul 2000 10:11:45 +1000
Message-ID:	<8600BF007197944F8DD3906E40CB42808F6E@itdomain001.itdomain.net.au>
MIME-Version:	1.0
Thread-Topic:	stackdump revisited
Thread-Index:	Ab/nptMUoZBneigwRAiZeIFA60sN5AAAIb8w
X-MimeOLE:	Produced By Microsoft Exchange V6.0.4208.0
X-MIME-Autoconverted:	from quoted-printable to 8bit by delorie.com id UAA07182

Indeed they are,
I'm told that it's standard/common practice to do the close() loop
before spawning the main daemon though. 
0 is Std_in, 1 is std_out and 2 is std_err.

So squid on cygwin is stackdumping when close(2) is called. I have
worked around this in squid now, so it's certainly not urgent (for me),
but the behaviour is different than on most un*xes.

Rob

-----Original Message-----
From: Charles Wilson [mailto:cwilson AT ece DOT gatech DOT edu]
Sent: Friday, 7 July 2000 10:08 AM
To: Robert Collins
Cc: cygwin-apps AT sourceware DOT cygnus DOT com
Subject: Re: stackdump revisited


Call me crazy, but aren't fd's 0,1 and 2 == stdin, stdout, stderr (in
some permutation)?

--Chuck

> Robert Collins wrote:
> 
> Thanks all for your patience with my stackdumping squid!
> 
> this time I have made real progress...
> squid goes thru the following loop when before it starts up the very
> first worker instance.
> ===
>     for (i = 0; i < Squid_MaxFD; i++)
>         close(i);
> ====
> 
> when i=2 it stackdumps. Are fd's 0, 1 and 2 reserved? I am looking
> into whether squid _needs_ to close these, or if it's an old hack
> still present..
> 
> Thanks in advance,
> Rob

- Raw text -