From owner-freebsd-hackers Sat Aug 16 21:30:08 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id VAA02996 for hackers-outgoing; Sat, 16 Aug 1997 21:30:08 -0700 (PDT) Received: from sendero-ppp.i-connect.net (sendero-ppp.i-Connect.Net [206.190.143.100]) by hub.freebsd.org (8.8.5/8.8.5) with SMTP id VAA02938 for ; Sat, 16 Aug 1997 21:29:55 -0700 (PDT) Received: (qmail 2291 invoked by uid 1000); 17 Aug 1997 04:29:51 -0000 Message-ID: X-Mailer: XFMail 1.2-alpha [p0] on FreeBSD Content-Type: text/plain; charset=iso-8859-8 Content-Transfer-Encoding: 8bit MIME-Version: 1.0 In-Reply-To: Date: Sat, 16 Aug 1997 21:29:51 -0700 (PDT) Organization: Atlas Telecom From: Simon Shapiro To: Jaye Mathisen Subject: RE: Besides time, what the heck does it take to get DPT driver i Cc: hackers@FreeBSD.ORG Sender: owner-freebsd-hackers@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi Jaye Mathisen; On 14-Aug-97 you wrote: > > > :) I'm getting tired repatching my source tree all the time. Me too, me too... I did not see the original posting, so please someone help me out here... > Now if the darn thing would make it into current so I could smp and DPT, > I could die a happy man. You are an exceptionally easy man to satisfy :-) I just posted a patch to current of the DPT driver 1.2.1. It even runs or I would not have been able to type this. > > I recall somebody saying it was going to happen a few weeks go, but > unless I"m doing something wrong, I don't see it. Here is a short version of my OPINION on what is happening (Justin is the gatekeeper of all SCSI code going into the kernel); The DPT driver suffers from certain problems, stopping it from being checked in. In no particular order: a. The coding style is (was) horrendously ugly. It is a mixture of K&R, System V, Oracle influence, bad habits (I used to write a lot of PL/I) and sheer madness. While very readable by yours truely, it is rightly not acceptable as a FreeBSD code. b. To get the performance levels I wanted and the load bearing ability I needed, the driver uses software interrupts (first SCSI HBA driver to actually do that on FreeBSD that I know of. This represents a modification to 2.2 interrupt handlers; Not acceptable by core. c. Version 3.0 of FreeBSD intends to use a whole new SCSI midlayer. One that complies with CAM3 (is this the correct way to describe it, Justin?). It was felt by the core team that the DPT driver should use this layer or pertinant pieces of it. Until this integration actually happens, the DPT driver will stay outside the source tree. d. There were several bugs in the driver that could show up as a crash or lost interrupts (I think). Although I could never actually cause some of them to surface, they are considered show stoppers. Much (if not all) of these should be gone now. I am not asking you to agree with this. Just telling you how I understand the situation. My employer intends to use FreeBSD and the DPT as key, (``strategic'' they call it) technology in some darn important products. As such we heavily review and test and review the code. As far as I can tell, the DPT driver works correctly and reliably. If not, let me know. Please. I am working closely with the core team to resolve these problems and bring the driver up to their excellent standards. We want the code we release (DPT is only the first), to be at least as good, stable and clean as the rest of the system. > (Now if I could just talk Simon into a Mylex driver, I'd really be > cookin'). Me? Mylex? Sure, send me a card, documentation and (if possible contact name). Simon