From owner-freebsd-alpha Sat Oct 14 8: 7:24 2000 Delivered-To: freebsd-alpha@freebsd.org Received: from mail.rdc2.tx.home.com (ha1.rdc2.tx.home.com [24.14.77.20]) by hub.freebsd.org (Postfix) with ESMTP id 0A15837B66C for ; Sat, 14 Oct 2000 08:07:23 -0700 (PDT) Received: from attglobal.net ([24.10.140.87]) by mail.rdc2.tx.home.com (InterMail vM.4.01.03.00 201-229-121) with ESMTP id <20001014150722.TOKL75.mail.rdc2.tx.home.com@attglobal.net> for ; Sat, 14 Oct 2000 08:07:22 -0700 Message-ID: <39E87745.596865AC@attglobal.net> Date: Sat, 14 Oct 2000 09:09:57 -0600 From: "Andrew M. Miklic" X-Mailer: Mozilla 4.75 [en] (X11; U; Linux 2.2.12 i386) X-Accept-Language: en MIME-Version: 1.0 To: freebsd-alpha@freebsd.org Subject: TGA and PCI Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-freebsd-alpha@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org I'm trying to finish-up a port of NetBSD's TGA driver to FreeBSD, and I'm running into problems in the very early probe phase during pci_cfgread()... Basically, in (st550, st660, etc.)_cons_init(), it first calls (apecs, cia, etc.)_init() to enable PCI support. Shortly after that , it calls sccnattach where the probe for TGA occurs. To to the probe, I read cfg space to get the device id and vendor id for all slots to determine if what I find is a TGA card. During pci_cfgread(), however, the machine suffers an "invalid kernel stack" error and halts. I've tracked it down to a final call to alpha_mb()...does anyone know why this instruction should kill the machine at this point? Is PCI not really available to me this early? Andrew Miklic To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-alpha" in the body of the message