Mailing-List: contact cygwin-help AT sourceware DOT cygnus DOT com; run by ezmlm List-Subscribe: List-Archive: List-Post: List-Help: , Sender: cygwin-owner AT sources DOT redhat DOT com Delivered-To: mailing list cygwin AT sources DOT redhat DOT com To: "Rui Carmo" Cc: "cygwin" Subject: Re: CGI apps crashing? (Was: Re: id.exe crashes when using ssh (since 1.1.6?)) References: <006901c071d3$ad46a460$012aa8c0 AT limbo> <20001229214427 DOT C4278 AT cygbert DOT vinschen DOT de> <006c01c072bb$104cbfd0$012aa8c0 AT limbo> From: ht AT cogsci DOT ed DOT ac DOT uk (Henry S. Thompson) Date: 03 Jan 2001 16:43:31 +0000 In-Reply-To: "Rui Carmo"'s message of "Sat, 30 Dec 2000 23:48:48 -0000" Message-ID: Lines: 73 User-Agent: Gnus/5.0805 (Gnus v5.8.5) XEmacs/21.2(beta13) (Demeter) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii "Rui Carmo" writes: > Well, I have new data on this issue. A number of CGI apps I am using (that > generate JPEG images based on all kinds of data) stopped working after my > update (I am using Apache for Win32 on my development box) > > Oddly enough, the CGI binaries work OK when run from a bash prompt, but > yield corrupted images when run under Apache. None of the binaries was > recompiled. > > The penny finally dropped when I tried compiling conicplanet > (http://conic.sourceforge.net/) under cygwin to figure out some > image-processing code. conicplanet.cgi yields perfect results when run from > a bash prompt, and multicolored garbage under Apache - exactly the same > problem my unmodified binaries had... > > So my guess is that we have something afoot inside cygwin1.dll... > Apparently, it has problems forking a subprocess under anything other than > the initial user context (hence the crashes on the remote id.exe when I ssh > to another box, and the misbehaving CGI binaries). > > Anyone here experiencing the same problemas with either ssh or CGI > executables created with cygwin? The id.exe crash has been reported several times (I just reproduced it having upgraded to 1.1.7), but none of us have been able to get enough info about it to Corinna to enable her to solve the problem, and she can't reproduce it. Maybe your additional input will help, but she's on vacation just now I believe. . . I tried running sshd -i -d from inetd and looking in the application event log, but not sure there's anything useful there. There appeared to be some issue with /var/log/lastlog, first not being there, then after I created it (empty), I see "lastlog_get_entry: Error reading from /var/log/lastlog: No such file or directory" but when I look at it it's now 200K of nulls with XSS:tty1...[name of host I tried to connect _from_] near the end. All the entries chronologically after that in the event log are debug1: tvp!=NULL kid 0 mili 10 The application log event messages before that, in reverse order, are: debug1: server_init_dispatch_15 debug1: server_init_dispatch_13 debug1: fd 8 setting O_NONBLOCK debug1: fd 7 setting O_NONBLOCK debug1: Entering interactive session and everything earlier than that is very similar to what I get from sshd -d from the command line (which, as you observe, works just fine). There are slight differences: from the command line debug I see at the equivalent point: debug1: tvp!=NULL kid 0 mili 10 debug1: server_init_dispatch_15 debug1: server_init_dispatch_13 debug1: fd 7 setting O_NONBLOCK debug1: fd 3 setting O_NONBLOCK debug1: Entering interactive session Is the difference between fd 8 and fd 3 significant? ht -- Henry S. Thompson, HCRC Language Technology Group, University of Edinburgh W3C Fellow 1999--2001, part-time member of W3C Team 2 Buccleuch Place, Edinburgh EH8 9LW, SCOTLAND -- (44) 131 650-4440 Fax: (44) 131 650-4587, e-mail: ht AT cogsci DOT ed DOT ac DOT uk URL: http://www.ltg.ed.ac.uk/~ht/ -- Want to unsubscribe from this list? Check out: http://cygwin.com/ml/#unsubscribe-simple