From mboxrd@z Thu Jan 1 00:00:00 1970 From: James Bottomley Subject: RE: [PATCH] aic94xx: fix smartctl utility problem Date: Mon, 17 Sep 2007 22:31:56 -0500 Message-ID: <1190086316.3386.72.camel@localhost.localdomain> References: Mime-Version: 1.0 Content-Type: text/plain Content-Transfer-Encoding: 7bit Return-path: Received: from hancock.steeleye.com ([71.30.118.248]:41902 "EHLO hancock.sc.steeleye.com" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1751485AbXIRDcI (ORCPT ); Mon, 17 Sep 2007 23:32:08 -0400 In-Reply-To: Sender: linux-scsi-owner@vger.kernel.org List-Id: linux-scsi@vger.kernel.org To: "Wu, Gilbert" Cc: dougg@torque.net, Linux-scsi@vger.kernel.org On Mon, 2007-09-17 at 18:43 -0700, Wu, Gilbert wrote: > ata_task_return code=0 ====> ****** cmd success then the no ATA > output register return. > scsi 2:0:0:0: CD-ROM PLEXTOR DVDR PX-755A 1.02 PQ: 0 > ANSI: 5 > scsi 2:0:0:0: Attached scsi generic sg2 type 5 > ata command=a0 > ata_task_return code=c ======> ***** cmd failed then return ATA output > register > return reg= 34 40 51 54 0 0 0 a0 0 > ata command=a0 > ata_task_return code=1 That just says a packet command returned an illegal request ... I think the CD layer expects this, but without seeing the SCSI command it's hard to tell. > sas: sas_ata_phy_reset: Found ATAPI device. > ata command=a0 > ata_task_return code=c > return reg= 34 40 51 24 0 0 20 a0 0 > ata command=a0 > ata_task_return code=1 This again is a packet command returning not ready (probably no media present). > ata command=a0 > ata_task_return code=c > return reg= 34 40 51 24 0 0 20 a0 0 > ata command=a0 > ata_task_return code=1 > ata command=a0 > ata_task_return code=c > return reg= 34 40 51 24 0 0 20 a0 0 > ata command=a0 > ata_task_return code=1 > ata command=a0 > ata_task_return code=c > return reg= 34 40 51 24 0 0 20 a0 0 > ata command=a0 > ata_task_return code=1 These again look like normal CD probes returning not ready. > ata command=a0 > ------------[ cut here ]------------ > kernel BUG at drivers/ata/libata-core.c:1380! > invalid opcode: 0000 [#1] OK, this is the fascinating one ... I think it's a bug in libata itself. It seems to mean that the tag its trying to use (which is fixed by convention) is already outstanding. Since the current probe sequence is single threaded, I can't really see how this has happened. James