From mboxrd@z Thu Jan 1 00:00:00 1970 From: James Bottomley Subject: Re: [PATCH] tmscsim: INQUIRY always only untagged Date: 06 Sep 2004 10:34:23 -0400 Sender: linux-scsi-owner@vger.kernel.org Message-ID: <1094481264.2149.14.camel@mulgrave> References: <20040904140102.A15247@infradead.org> <20040904233835.A18659@infradead.org> <1094398979.10977.9.camel@mulgrave> <20040905173134.A31521@infradead.org> Mime-Version: 1.0 Content-Type: text/plain Content-Transfer-Encoding: 7bit Return-path: Received: from stat16.steeleye.com ([209.192.50.48]:29370 "EHLO hancock.sc.steeleye.com") by vger.kernel.org with ESMTP id S267393AbUIFOe3 (ORCPT ); Mon, 6 Sep 2004 10:34:29 -0400 In-Reply-To: List-Id: linux-scsi@vger.kernel.org To: Guennadi Liakhovetski Cc: Christoph Hellwig , SCSI Mailing List On Sun, 2004-09-05 at 16:19, Guennadi Liakhovetski wrote: > >> 1) remove the inquiry snooping and believe the flags in struct > >> scsi_device instead > > ...and use tags set up by the block layer... Well, yes, that too ... it depends how much time anyone has to spend on this driver. > >> 2) Don't turn on TCQ until slave_configure() > > Well, this alone wouldn't help with this specific problem, exactly because > INQUIRY can be sent later too. Or would the mid-layer completely disable > tagging for such a disk? The bug report claims 2.4 works, so probably there the device itself will still screw up if another inquiry were sent as part of the tagged stream, I think (perhaps it doesn't, someone needs to check). However, turning on TCQ in slave_configure() will mean that the two initial inquiries are sent untagged, so it will only manifest in 2.6 in the same circumstances as 2.4. James