From owner-freebsd-bugs Sat Oct 17 11:48:17 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id LAA10160 for freebsd-bugs-outgoing; Sat, 17 Oct 1998 11:48:17 -0700 (PDT) (envelope-from owner-freebsd-bugs@FreeBSD.ORG) Received: from lariat.lariat.org (lariat.lariat.org [206.100.185.2]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id LAA10154 for ; Sat, 17 Oct 1998 11:48:13 -0700 (PDT) (envelope-from brett@lariat.org) Received: (from brett@localhost) by lariat.lariat.org (8.8.8/8.8.6) id MAA06165; Sat, 17 Oct 1998 12:47:49 -0600 (MDT) Message-Id: <4.1.19981017121426.06ec3ea0@mail.lariat.org> X-Sender: brett@mail.lariat.org X-Mailer: QUALCOMM Windows Eudora Pro Version 4.1 Date: Sat, 17 Oct 1998 12:46:57 -0600 To: bugs@FreeBSD.ORG From: Brett Glass Subject: Crash workaround: ATAPI and ATAPI_STATIC required in 2.2.7 wd driver Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Sender: owner-freebsd-bugs@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org I've finally taken an entire day and managed to get those 2.2.7 systems stable. As I've mentioned earlier, the machines were getting page faults in the kernel during disk-intensive operations and sometimes seemingly at random. I enabled the debugger and stimulated the systems to crash by running disk-intensive tasks. When I looked at the systems in the debugger after the crash, EVERY ONE was in an idle loop at the time. What's more, WHILE THE DEBUGGER WAS RUNNING, I received error messages from the wd driver. Since the wd driver enters an idle loop to wait for I/O completion, and the crashes occurred during periods of heavy disk I/O (fortunately, mostly reads), everything seemed to point toward the wd driver as the source of the problem. Experimentation was made difficult, however, by the fact that building a new kernel also generated sufficiently intense disk activity to crater the machine. I set up another machine with a SCSI drive just to build test kernels. Sure enough, after MUCH experimentation, I finally found that if one enables the ATAPI and ATAPI_STATIC kernel options -- even though there are no ATAPI CD-ROM drives or tapes present -- the problem vanishes. Once I installed a kernel with these options, we were even able to do a full backup without a crash -- something that hadn't worked in months. Has the code been tested without the ATAPI and ATAPI_STATIC options lately? (I imagine that a lot of work has gone into SCSI testing, so problems with IDE might have been missed even in 3.0.) Who is currently responsible for maintaining the wd driver? --Brett Glass To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-bugs" in the body of the message