From mboxrd@z Thu Jan 1 00:00:00 1970 From: James Bottomley Subject: Re: aacraid probe breaks in scsi-block-2.6.git Date: Wed, 10 Aug 2005 12:27:16 -0500 Message-ID: <1123694836.5134.14.camel@mulgrave> References: <1123694406.30596.10.camel@markh1.pdx.osdl.net> Mime-Version: 1.0 Content-Type: text/plain Content-Transfer-Encoding: 7bit Return-path: Received: from stat16.steeleye.com ([209.192.50.48]:41623 "EHLO hancock.sc.steeleye.com") by vger.kernel.org with ESMTP id S965224AbVHJR1Z (ORCPT ); Wed, 10 Aug 2005 13:27:25 -0400 In-Reply-To: <1123694406.30596.10.camel@markh1.pdx.osdl.net> Sender: linux-scsi-owner@vger.kernel.org List-Id: linux-scsi@vger.kernel.org To: Mark Haverkamp Cc: linux-scsi On Wed, 2005-08-10 at 10:20 -0700, Mark Haverkamp wrote: > While trying out a recent mm kernel I noticed that when I loaded the > aacraid driver I saw errors probing the disks: > > sdc : sector size 0 reported, assuming 512. > SCSI device sdc: 1 512-byte hdwr sectors (0 MB) Sigh, I guess this is because aacraid is making illegal assumptions about the format of certain commands it emulates? We had a similar issue with 3w-xxxx which assumed commands like INQUIRY, READ_CAPACITY and MODE_SENSE only came from the kernel, so would never have use_sg != 0. This is, of course, wrong, since such commands can and do come from SG_IO. The fix is probably similar to the 3w-xxxx one. James