From owner-freebsd-ports Thu Jun 13 8:36:48 2002 Delivered-To: freebsd-ports@freebsd.org Received: from hda.hda.com (host65.hda.com [63.104.68.65]) by hub.freebsd.org (Postfix) with ESMTP id DC88D37B414; Thu, 13 Jun 2002 08:36:34 -0700 (PDT) Received: (from dufault@localhost) by hda.hda.com (8.11.6/8.11.6) id g5DFafd09711; Thu, 13 Jun 2002 11:36:41 -0400 (EDT) (envelope-from dufault@hda.com) Date: Thu, 13 Jun 2002 11:36:21 -0400 From: Peter Dufault To: Thomas Gellekum Cc: ports@FreeBSD.ORG, rhh@FreeBSD.ORG Subject: Re: Opendx core dumps for me Message-ID: <20020613113621.A9483@hda.hda.com> References: <20020607072528.A20647@hda.hda.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline User-Agent: Mutt/1.2.5i In-Reply-To: ; from tg@melaten.rwth-aachen.de on Thu, Jun 13, 2002 at 01:30:16PM +0200 Sender: owner-freebsd-ports@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.org On Thu, Jun 13, 2002 at 01:30:16PM +0200, Thomas Gellekum wrote: > I've noticed that both libc_r and libc are linked into dxui, probably > because something along the dependency chain used libtool. My > home-built binaries seem to work fine, though. > > Does ktrace give any clues? I tried truss first: > stat("/usr/local/bin/dx",0xbfbfdc80) = 0 (0x0) > pipe() = 9 (0x9) > fcntl(0x9,0x3,0x0) = 2 (0x2) > fcntl(0x9,0x4,0x6) = 0 (0x0) > fcntl(0xa,0x3,0x0) = 2 (0x2) > fcntl(0xa,0x4,0x6) = 0 (0x0) > pipe() = 11 (0xb) > fcntl(0xb,0x3,0x0) = 2 (0x2) > fcntl(0xb,0x4,0x6) = 0 (0x0) > fcntl(0xc,0x3,0x0) = 2 (0x2) > fcntl(0xc,0x4,0x6) = 0 (0x0) > pipe() = 13 (0xd) > fcntl(0xd,0x3,0x0) = 2 (0x2) > fcntl(0xd,0x4,0x6) = 0 (0x0) > fcntl(0xe,0x3,0x0) = 2 (0x2) > fcntl(0xe,0x4,0x6) = 0 (0x0) > fork() = 3065 (0xbf9) > fstat(9,0xbfbfd67c) = 0 (0x0) > close(9) = 0 (0x0) > fstat(12,0xbfbfd66c) = 0 (0x0) > close(12) = 0 (0x0) > fstat(14,0xbfbfd67c) = 0 (0x0) > close(14) = 0 (0x0) > getdtablesize() = 3722 (0xe8a) > gettimeofday(0xbfbfe2a4,0x0) = 0 (0x0) > poll(0x8118000,0x3ca,0x0) = 969 (0x3c9) > write(2,0xbfbfdbc4,37) = 37 (0x25) "timeout waiting for response from DX" > SIGNAL 11 > SIGNAL 11 > > Process stopped because of: 16 > process exit, rval = 139 I put a break point in "fork()", get the "starting data explorer window" and break at conn.c:867. If I set a break point at 868 and continue I wind up with the child startupui already a zombie, so it exits without printing anything even if I set the startupui "verbose" variable to 1. I tried to set the gdb "follow-fork-mode" to child to trace into the child to see what is going on but I still wind up in the parent; is follow-fork-mode broken? At any rate, the sig 11 of startupui is probably mishandled error handling when the parent startupui gets the timeout. I tried "ktrace -i" to see what was up, we definitely do start the child, I see it return from the fork and eventually exec the /usr/local/bin/dx shell, yet I don't get the debug output I expect from having verbose set to 1. I'll try more later, maybe this will give someone an idea. Peter -- Peter Dufault (dufault@hda.com) Realtime development, Machine control, HD Associates, Inc. Fail-Safe systems, Agency approval To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-ports" in the body of the message