From owner-freebsd-current@FreeBSD.ORG Mon Oct 11 17:24:55 2004 Return-Path: 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 82E8816A4CE for ; Mon, 11 Oct 2004 17:24:55 +0000 (GMT) Received: from philemon.caltech.edu (philemon.caltech.edu [131.215.158.74]) by mx1.FreeBSD.org (Postfix) with ESMTP id 68ED543D41 for ; Mon, 11 Oct 2004 17:24:55 +0000 (GMT) (envelope-from jd@philemon.caltech.edu) Received: from philemon.caltech.edu (localhost.caltech.edu [127.0.0.1]) by philemon.caltech.edu (8.12.9p1/8.12.9) with ESMTP id i9BHTHgX011679 for ; Mon, 11 Oct 2004 10:29:17 -0700 (PDT) (envelope-from jd@philemon.caltech.edu) Received: (from jd@localhost) by philemon.caltech.edu (8.12.9p1/8.12.9/Submit) id i9BHTHwx011678 for current@freebsd.org; Mon, 11 Oct 2004 10:29:17 -0700 (PDT) Date: Mon, 11 Oct 2004 10:29:17 -0700 From: Paul Allen To: current@freebsd.org Message-ID: <20041011172917.GC91892@philemon.caltech.edu> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline X-Mailman-Approved-At: Tue, 12 Oct 2004 14:36:20 +0000 Subject: 5.3-Beta7/RELENG_5, AGP/drm hard lock on xserver termination X-BeenThere: freebsd-current@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Discussions about the use of FreeBSD-current List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Mon, 11 Oct 2004 17:24:55 -0000 Since installing RELENG_5 (marked as 5.3-Beta7) I've experienced hard locks on xserver termination and more generally when switching between the xserver and the other text-mode ttys. This happens with GENERIC + a custom kernel with SMP removed. I built a kernel with the debugger, witness, and invariants only to discover that what had been very easy to reproduce simply went away. I suspect I've just made it much less probable... so I'll go at it some more tonight after work. AGP/drm locking suspected culprit? Can locking mistakes create problems on UP systems? Because of kernel thread preemption? I have a vague sense I've read about a similar issue on this list before, so if this is known problem, let me know so I don't waste my time. Thanks.