From mboxrd@z Thu Jan 1 00:00:00 1970 From: Boaz Harrosh Subject: Re: [0/4 ver2] Last 3 patches for bidi support Date: Thu, 08 Nov 2007 18:49:46 +0200 Message-ID: <47333E2A.7020307@panasas.com> References: <4730ACAA.2060007@panasas.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Return-path: Received: from sa9.bezeqint.net ([192.115.104.23]:44376 "EHLO sa9.bezeqint.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1759002AbXKHQt5 (ORCPT ); Thu, 8 Nov 2007 11:49:57 -0500 In-Reply-To: <4730ACAA.2060007@panasas.com> Sender: linux-scsi-owner@vger.kernel.org List-Id: linux-scsi@vger.kernel.org To: FUJITA Tomonori , James Bottomley , linux-scsi Cc: Pete Wyckoff , Benny Halevy , Andrew Morton On Tue, Nov 06 2007 at 20:04 +0200, Boaz Harrosh wrote: > [1] > I propose a small change to scsi_tgt_lib.c that will make > tgt completely neutral to the scsi_data_buffer patch. And will > make it all that more ready for bidi, too. TOMO is this OK? > > (Can you do without the GFP_KERNEL allocation flag? It could > make the code a bit more simple) > > [2] > scsi_data_buffer patch. > As requested by TOMO the all patch is squashed into one, 600 > and some lines. TOMO if it makes you happy, OK, here it is. > > There is one hunk from libsrp.c that I really hate. and from what > I understand of the code, only the "request_bufflen =" is really used, > All users of "request_buffer = addr" pass NULL, as far as I can see. > If you would like to make me happy, and it is at all possible, please > clean it. > > [3] > Once scsi_data_buffer is in, then scsi bidi support can be applied. > (Block bidi was already merged 3 kernels ago). It makes no changes > to scsi_cmnd and only adds some, not-at-all-dangerous, code to scsi_lib. > So I don't see any reason to wait. please all review. > > If ACKed by all parties and applied for inclusion for 2.6.25, > then they will have a long time to sit in MM and collect > compilation breakage reports from ARCHs we never compiled. > > I wish these make everybody happy this time > > Boaz Harrosh > Version 2, fixed as of Tomo's comments. Thanks Tomo. [0] Remove dead code from sd.c & sr.c. It was conceived by programmers in the passed that rq_data_dir() could perhaps be expanded to 2 bits and return READ/WRITE/BIDI. But we have decided long ago that a bidi request is when blk_bidi_rq() == true, and in that case rq_data_dir() == WRITE. So now we can be sure of what the compiler knew for a long time. (The future is already here) [1] [2] [3] See above I must insist again they should have time to sit in -mm. For posible compilation breakage of other ARCHs. Boaz