From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id ; Tue, 26 Jun 2001 14:02:36 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id ; Tue, 26 Jun 2001 14:02:27 -0400 Received: from gateway.sequent.com ([192.148.1.10]:6876 "EHLO gateway.sequent.com") by vger.kernel.org with ESMTP id ; Tue, 26 Jun 2001 14:02:13 -0400 Date: Tue, 26 Jun 2001 11:00:40 -0700 From: Patrick Mansfield To: linux-scsi@vger.kernel.org Cc: linux-kernel@vger.kernel.org Subject: multi-path IO in SCSI mid-layer Message-ID: <20010626110040.A28555@eng2.sequent.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Mailer: Mutt 1.0.1i Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Hi - I'm interested in multi-path IO in the linux scsi mid-layer. Are there developers working on changes to the scsi layers/interfaces? I've seen references about such work, but no details. Anyone else interested in or working on multi-path IO in the mid-layer? I've looked at the code as to what changes might be required, and did a simple prototype to issue IO requests using multiple paths for one device - configuring one device in with multiple paths using SCSI INQUIRY page 0x83, and selecting a path (just round-robin) at scsi_submit_cmd() time. But, a decent multi-path IO implementation requires significant changes to the current linux scsi interfaces/structures - especially where no functional interfaces exist, such as the direct references to Scsi_Device host, and Scsi_Host host_queue. -- Patrick Mansfield patman@sequent.com patman@us.ibm.com