Date: Wed, 5 Aug 2009 09:37:22 -0400
From: Christopher Faylor <cgf-use-the-mailinglist-please AT cygwin DOT com>
To: cygwin AT cygwin DOT com
Subject: Re: Please test the latest snapshot!
Message-ID: <20090805133721.GA1105@ednor.casa.cgf.cx>
Reply-To: cygwin AT cygwin DOT com
Mail-Followup-To: cygwin AT cygwin DOT com
References: <20090804223616 DOT GA9686 AT ednor DOT casa DOT cgf DOT cx>  <2116948976 DOT 8684931249474318478 DOT JavaMail DOT root AT sz0059a DOT emeryville DOT ca DOT mail DOT comcast DOT net>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <2116948976.8684931249474318478.JavaMail.root@sz0059a.emeryville.ca.mail.comcast.net>
User-Agent: Mutt/1.5.20 (2009-06-14)
Mailing-List: contact cygwin-help AT cygwin DOT com; run by ezmlm
Precedence: bulk
Sender: cygwin-owner AT cygwin DOT com
Mail-Followup-To: cygwin AT cygwin DOT com

On Wed, Aug 05, 2009 at 12:11:58PM +0000, Eric Blake wrote:
>> >>> $ git clone git://git.sv.gnu.org/libsigsegv.git
>> >>> Initialized empty Git repository in /home/eblake/libsigsegv/.git/
>> >>> fatal: write error (Socket operation on non-socket)
>> >>
>> I tried it and it failed as you indicated.
>> 
>> The next snapshot should fix the problem.
>
>Nope - still seeing this in the strace:
>
>308283 3906650 [main] git 2280 cygwin_connect: 0 = connect (3, 0x7D90C8, 16)
>  572 3907222 [main] git 2280 dtable::dup2: dup2 (3, 4)
>  142 3907364 [main] git 2280 build_fh_pc: fh 0x612094FC
>  147 3907511 [main] git 2280 fhandler_socket::dup: here
>  137 3907648 [main] git 2280 fhandler_base::dup: in fhandler_base dup
>  133 3907781 [main] git 2280 dtable::dup_worker: duped '' old 0x678, new 0x650
>  124 3907905 [main] git 2280 dtable::dup2: newfh->io_handle 0x650, oldfh->io_ha
>ndle 0x678
>  122 3908027 [main] git 2280 dtable::dup2: 4 = dup2 (3, 4)
>  173 3908200 [main] git 2280 writev: writev (4, 0x22C864, 1)
>12504 3920704 [main] git 2280 __set_errno: void __set_winsock_errno(const char*,
> int):217 val 108
>  233 3920937 [main] git 2280 __set_winsock_errno: send_internal:1537 - winsock
>error 10038 -> errno 108
>  154 3921091 [main] git 2280 writev: -1 = write (4, 0x22C864, 1), errno 108

Unsolicited strace output is rarely useful.  In the surpring majority of
cases, people snip the wrong thing and this is no exception.

>
>$ uname -a
>CYGWIN_NT-5.1 eblake 1.7.0s(0.212/5/3) 20090804 23:19:31 i686 Cygwin

Well, all I can tell you is that I could duplicate your problem on two
different machines and, after the changes that I made last night, I no
longer could.  The problem was apparently introduced in early July so
this has nothing to do with the reported instability that people are
reporting in the newest releases.

Since I can no longer duplicate the problem, could you try setting the
size of rmem and wmem down in the fdsock function in net.cc and see if
that works around the problem?

cgf

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