From owner-freebsd-current Fri Jun 28 8:17: 5 2002 Delivered-To: freebsd-current@freebsd.org Received: from mx1.FreeBSD.org (mx1.FreeBSD.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id D414837B401 for ; Fri, 28 Jun 2002 08:16:52 -0700 (PDT) Received: from mxb.saturn-tech.com (mxb.saturn-tech.com [207.229.19.99]) by mx1.FreeBSD.org (Postfix) with ESMTP id 9593643E09 for ; Fri, 28 Jun 2002 08:16:49 -0700 (PDT) (envelope-from drussell@saturn-tech.com) Received: from mxb.saturn-tech.com (localhost [127.0.0.1]) by mxb.saturn-tech.com (8.12.3/8.12.3) with ESMTP id g5SFGp7E035444; Fri, 28 Jun 2002 09:16:51 -0600 (MDT) (envelope-from drussell@saturn-tech.com) Received: from localhost (drussell@localhost) by mxb.saturn-tech.com (8.12.3/8.12.3/Submit) with ESMTP id g5SFGoia035441; Fri, 28 Jun 2002 09:16:50 -0600 (MDT) (envelope-from drussell@saturn-tech.com) X-Authentication-Warning: mxb.saturn-tech.com: drussell owned process doing -bs Date: Fri, 28 Jun 2002 09:16:50 -0600 (MDT) From: Doug Russell To: Matthew Dillon Cc: current@FreeBSD.ORG Subject: Re: su gets SIGHUP randomly on startup with latest current In-Reply-To: <200206252327.g5PNR05U051967@apollo.backplane.com> Message-ID: <20020628090021.X13543-100000@mxa.saturn-tech.com> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-current@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.ORG On Tue, 25 Jun 2002, Matthew Dillon wrote: > This is really odd. Is anyone else getting this? I wasn't getting this > with a current of just a few days ago, now one in ten or so 'su' commands > does something really odd. Are there QLogic controllers in the machine? (Is the isp driver active?) Try turning off gdb if it is compiled into the kernel. I've been having random signal-induction problems with old QLogic 1041 adapters since about 4.0 or so. It was the changes to have more than 32 signals that seemed to start the problem. I get random SIGTERM, SIGPROFILE and other signals, usually 'profiling timer expired' if the system is lucky enough to actually boot all the way. Sometimes it is an fsck or other somewhat lengthy startup process that is the first to die. I finally found that removing gdb from the kernel made the problem dissapear, or at least less severe, but I also now have to statically compile in the ispfw device to load firmware early in the boot, otherwise it still often dies, even with no gdb, before loading the firmware, but that may be related more to the incredibly old firmware on the controllers. The problem seems significantly more severe on SMP-hardwared boxen. Let me know if you need more info, as I'd be glad to help. I have more than 10 1041 HBAs. Later...... To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-current" in the body of the message