From owner-freebsd-atm Sun Aug 31 23:07:32 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id XAA08880 for atm-outgoing; Sun, 31 Aug 1997 23:07:32 -0700 (PDT) Received: from bells.cs.ucl.ac.uk (bells.cs.ucl.ac.uk [128.16.5.31]) by hub.freebsd.org (8.8.7/8.8.7) with SMTP id XAA08863 for ; Sun, 31 Aug 1997 23:07:15 -0700 (PDT) Received: from mickey.cs.ucl.ac.uk by bells.cs.ucl.ac.uk with local SMTP id ; Sun, 31 Aug 1997 18:51:08 +0100 X-Mailer: exmh version 1.6.6 3/24/96 To: salo@msc.edu (Tim Salo) cc: freebsd-atm@FreeBSD.ORG Subject: Re: Efficient ATM card. In-reply-to: Your message of "Fri, 29 Aug 1997 17:11:27 CDT." <199708292211.RAA27676@uh.msc.edu> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Sun, 31 Aug 1997 18:51:00 +0100 Message-ID: <16197.873049860@cs.ucl.ac.uk> From: Panos GEVROS Sender: owner-freebsd-atm@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk there was no problem report, i was trying to use HARP2 a couple of months ago, -in the first attempt it didnt work and i didnt pursue it any further, -i'm afraid I cannot help you with more details because it was some time ago and immediately afterwards i tried Cranor's driver and that did. We are simply looking for rate limiting on the PVCs , it would be very desirtable feature, because we are using the FreeBSD box as the endpoint of a wide area ATM connection which has a rate limit set, without some traffic shaping at the end host nothing gets through (except for very-very high rate limits) For our purposes, just configuring things from the command line or from a start up script is sufficient, an API though seems very interesting feature. Panos In message <199708292211.RAA27676@uh.msc.edu>, Tim Salo writes: |What problems did you have with HARP? (I couldn't find a problem report |from you, but I may have missed it.) You might want to try HARP 2.1, |http://www.msci.magic.net/ | |> Another thing is rate limiting currently it seems not to be present in any *of |> them, and it would be a very desirable feature for a driver included in |> FreeBSD. | |Could you provide a bit more detail about what you are looking for? | |Do you want an API which allows an application to set these parameters? | |Thanks, | -tjs From owner-freebsd-atm Mon Sep 1 00:56:45 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id AAA13261 for atm-outgoing; Mon, 1 Sep 1997 00:56:45 -0700 (PDT) Received: from inetfw.sonycsl.co.jp (inetfw.sonycsl.co.jp [203.137.129.4]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id AAA13256 for ; Mon, 1 Sep 1997 00:56:42 -0700 (PDT) Received: from hotaka.csl.sony.co.jp (hotaka.csl.sony.co.jp [43.27.98.57]) by inetfw.sonycsl.co.jp (8.8.5/3.5W) with ESMTP id QAA18134; Mon, 1 Sep 1997 16:56:34 +0900 (JST) Received: from localhost (localhost [127.0.0.1]) by hotaka.csl.sony.co.jp (8.8.4/3.3W3) with ESMTP id QAA15363; Mon, 1 Sep 1997 16:56:19 +0900 (JST) Message-Id: <199709010756.QAA15363@hotaka.csl.sony.co.jp> To: Panos GEVROS cc: salo@msc.edu (Tim Salo), freebsd-atm@FreeBSD.ORG Subject: Re: Efficient ATM card. In-reply-to: Your message of "Sun, 31 Aug 1997 18:51:00 +0100." <16197.873049860@cs.ucl.ac.uk> Date: Mon, 01 Sep 1997 16:56:18 +0900 From: Kenjiro Cho Sender: owner-freebsd-atm@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk >From my experience, I can say that ENI drivers choose chipset and motherboard. All ENI drivers make heavy use of PCI DMA, and the first generation pci chipsets don't seem to work well. So, use triton-2, natoma, or newer chipsets. Cranor's driver tries to find out the workable DMA combination (burst size and alignment) but still fails to work with older (broken?) chipsets. I don't know much about harp2 but cranor's driver seems to do more extensive testing. WRT adding a traffic shaping function, I'm planning to add a sysctl for it to cranor's driver but I won't have time to make a release until October. If you're desparate, drop me an e-mail. --kj From owner-freebsd-atm Mon Sep 1 02:21:36 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id CAA16157 for atm-outgoing; Mon, 1 Sep 1997 02:21:36 -0700 (PDT) Received: from jamaica.nal.ecl.net (jamaica.nal.ecl.net [163.138.70.52]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id CAA16146 for ; Mon, 1 Sep 1997 02:21:28 -0700 (PDT) Received: by jamaica.nal.ecl.net (8.8.2/eclnet*nal.mx4) with TCP; Mon, 1 Sep 1997 18:21:10 +0900 (JST) Message-Id: <199709010921.SAA01057@jamaica.nal.ecl.net> To: kjc@csl.sony.co.jp Cc: P.Gevros@cs.ucl.ac.uk, salo@msc.edu, freebsd-atm@freebsd.org Subject: Re: Efficient ATM card. From: Yoshikazu Oda In-Reply-To: Your message of "Mon, 01 Sep 1997 16:56:18 +0900" References: <199709010756.QAA15363@hotaka.csl.sony.co.jp> X-Mailer: Mew version 1.54 on Emacs 19.28.1, Mule 2.3 Mime-Version: 1.0 Content-Type: Text/Plain; charset=us-ascii Content-Transfer-Encoding: 7bit Date: Mon, 01 Sep 1997 18:21:03 +0900 Sender: owner-freebsd-atm@freebsd.org X-Loop: FreeBSD.org Precedence: bulk I'm using ENI ATM card with HARP. > >From my experience, I can say that ENI drivers choose chipset and > motherboard. All ENI drivers make heavy use of PCI DMA, and the > first generation pci chipsets don't seem to work well. So, use > triton-2, natoma, or newer chipsets. You can use only 1WORD-DMA for first generation pci chipsets. It is default setting on HARP version 2.1. (So, this version may work on old chipsets' machine.) Also, if you are using newer than triton-2 or else, you can use 4WORD-DMA, 8WORD-DMA for more performance by editting source code (eni_transmit.c). > WRT adding a traffic shaping function, I'm planning to add a sysctl > for it to cranor's driver but I won't have time to make a release > until October. If you're desparate, drop me an e-mail. > > --kj I'm also looking for traffic shaping function for ATM card too. I know ENI card supports traffic shaping on its hardware (but no other cards support it). There are 8 channels for PCR timer (only PCR, so it can't support VBR service), and setting two parameters (Pr,Rate_Res) to specify PCR for each channel. But I don't know this may work fine. I'd like to know more about WRT's work for traffic shaping. Is it hardware shaping on ENI card or shaping with software or else? Regards, -------------------------------- Yoshikazu Oda Integration Services Laboratory NTT Multimedia Networks Laboratories E-mail: oda.yoshikazu@na.tnl.ntt.co.jp oda@nal.ecl.net From owner-freebsd-atm Mon Sep 1 02:35:41 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id CAA16632 for atm-outgoing; Mon, 1 Sep 1997 02:35:41 -0700 (PDT) Received: from bells.cs.ucl.ac.uk (bells.cs.ucl.ac.uk [128.16.5.31]) by hub.freebsd.org (8.8.7/8.8.7) with SMTP id CAA16615 for ; Mon, 1 Sep 1997 02:35:34 -0700 (PDT) Received: from mickey.cs.ucl.ac.uk by bells.cs.ucl.ac.uk with local SMTP id ; Mon, 1 Sep 1997 10:35:00 +0100 X-Mailer: exmh version 1.6.6 3/24/96 To: Yoshikazu Oda cc: kjc@csl.sony.co.jp, P.Gevros@cs.ucl.ac.uk, salo@msc.edu, freebsd-atm@freebsd.org Subject: Re: Efficient ATM card. In-reply-to: Your message of "Mon, 01 Sep 1997 18:21:03 +0900." <199709010921.SAA01057@jamaica.nal.ecl.net> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Mon, 01 Sep 1997 10:34:57 +0100 Message-ID: <556.873106497@cs.ucl.ac.uk> From: Panos GEVROS Sender: owner-freebsd-atm@freebsd.org X-Loop: FreeBSD.org Precedence: bulk We 've been promised a release of Cranor's driver with rate-limit functionality within the next 2 or 3 weeks, i 'd expect it to be publicly available, My understanding was that the traffic shaping will be done all in s/w . When we'll try it i'll let you know. Panos Yoshikazu Oda writes: | |I'm using ENI ATM card with HARP. | | |You can use only 1WORD-DMA for first generation pci chipsets. |It is default setting on HARP version 2.1. (So, this version |may work on old chipsets' machine.) | |Also, if you are using newer than triton-2 or else, you can use |4WORD-DMA, 8WORD-DMA for more performance by editting source |code (eni_transmit.c). | | |I'm also looking for traffic shaping function for ATM card too. |I know ENI card supports traffic shaping on its hardware (but no |other cards support it). There are 8 channels for PCR timer (only |PCR, so it can't support VBR service), and setting two parameters |(Pr,Rate_Res) to specify PCR for each channel. But I don't know |this may work fine. | |I'd like to know more about WRT's work for traffic shaping. |Is it hardware shaping on ENI card or shaping with software or else? | |Regards, | | |-------------------------------- |Yoshikazu Oda | Integration Services Laboratory | NTT Multimedia Networks Laboratories | E-mail: oda.yoshikazu@na.tnl.ntt.co.jp | oda@nal.ecl.net | From owner-freebsd-atm Mon Sep 1 03:14:03 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id DAA18134 for atm-outgoing; Mon, 1 Sep 1997 03:14:03 -0700 (PDT) Received: from inetfw.sonycsl.co.jp (inetfw.sonycsl.co.jp [203.137.129.4]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id DAA18128 for ; Mon, 1 Sep 1997 03:13:56 -0700 (PDT) Received: from hotaka.csl.sony.co.jp (hotaka.csl.sony.co.jp [43.27.98.57]) by inetfw.sonycsl.co.jp (8.8.5/3.5W) with ESMTP id TAA21046; Mon, 1 Sep 1997 19:13:53 +0900 (JST) Received: from localhost (localhost [127.0.0.1]) by hotaka.csl.sony.co.jp (8.8.4/3.3W3) with ESMTP id TAA19588; Mon, 1 Sep 1997 19:13:37 +0900 (JST) Message-Id: <199709011013.TAA19588@hotaka.csl.sony.co.jp> To: Yoshikazu Oda cc: P.Gevros@cs.ucl.ac.uk, salo@msc.edu, freebsd-atm@freebsd.org Subject: Re: Efficient ATM card. In-reply-to: Your message of "Mon, 01 Sep 1997 18:21:03 JST." <199709010921.SAA01057@jamaica.nal.ecl.net> Date: Mon, 01 Sep 1997 19:13:37 +0900 From: Kenjiro Cho Sender: owner-freebsd-atm@freebsd.org X-Loop: FreeBSD.org Precedence: bulk >> You can use only 1WORD-DMA for first generation pci chipsets. >> It is default setting on HARP version 2.1. (So, this version >> may work on old chipsets' machine.) It's good to know they are really broken. It seems to me that using a broken old pci chipset for ATM doesn't make sense since people want to use ATM for performance and there's no way to get good performance with those chipsets. >> I'm also looking for traffic shaping function for ATM card too. >> I know ENI card supports traffic shaping on its hardware (but no >> other cards support it). There are 8 channels for PCR timer (only >> PCR, so it can't support VBR service), and setting two parameters >> (Pr,Rate_Res) to specify PCR for each channel. But I don't know >> this may work fine. >> I'd like to know more about WRT's work for traffic shaping. >> Is it hardware shaping on ENI card or shaping with software or else? You're right about ENI's hardware shaper. It works pretty well. You might want to take a look at Werner Almesberger's Linux driver to know how to set these parameters. (BTW, software shaper won't help to get through the UPC.) --kj From owner-freebsd-atm Fri Sep 5 08:11:22 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id IAA13684 for atm-outgoing; Fri, 5 Sep 1997 08:11:22 -0700 (PDT) Received: from plains.NoDak.edu (tinguely@plains.NoDak.edu [134.129.111.64]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id IAA13679 for ; Fri, 5 Sep 1997 08:11:16 -0700 (PDT) Received: (from tinguely@localhost) by plains.NoDak.edu (8.8.5/8.8.5) id KAA20004 for freebsd-atm@freebsd.org; Fri, 5 Sep 1997 10:11:11 -0500 (CDT) Date: Fri, 5 Sep 1997 10:11:11 -0500 (CDT) From: Mark Tinguely Message-Id: <199709051511.KAA20004@plains.NoDak.edu> To: freebsd-atm@freebsd.org Subject: Announcing IDT NICStAR device driver Sender: owner-freebsd-atm@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Announcing the initial release of the IDT NICStAR device driver for FreeBSD-2.2.2-RELEASE. This device driver supports the 77201/77221 ATM SAR from Integrated Device Technology Inc. (http://www.idt.com/) as documented in the "IDT77201 NICStAR User Manual" (http://www.idt.com/), and separate Errata documents received from IDT. IDT also sells an evaluation board that consists of the SAR, the 128KB of SRAM (used for the various tables), and ATM media controller. The 77201/77221 is different than most other SARs, in that the sequencing and Re-assembly of the PDU from cells is done in host memory. The NICStAR is priced much less that other commercial ATM cards. Since more host memory is used than with traditional ARM cards, make sure you have plenty of RAM in your system. This driver was developed on a machine with 16 MB of RAM. I would suggest a machine using this driver have at least 32 MB of RAM. This driver was written and tested with a C3 version of the 77201. The 77221 is the E release of the SAR. Because of various errors in the 77201 SAR releases, I highly recommend the purchase of the 77221. If you are already own a 77201 (C3 or D), it is possible/likely the card can/will hang under conditions of receiving while also doing heavy transmission. Driver Features: 1) Quality of Service: The NICStAR support Constant Bit Rate, Variable Bit Rate*, Available Bit Rate*, and Unspecified Bit Rate QOS. The CBR is strongly enforced by the scheduling of the connection transmit times. The driver attempts to spread the transmit slots equally in time. The VBR, ABR, and UBR are supported in the card as prioritized transmission queues. The driver specifies the maximum transmission rate for each VC in the priority queue. The driver cannot guarantee the minimum ABR rate. 2) HARP support: This driver has network hooks to support the Host ATM Research Platform (HARP) classical IP over ATM written by the Advanced Networking Group at Network Computing Services, Inc (formerly the Minnesota Supercomputer Center, Inc.). For more information on HARP see: http://www.msci.magic.net/ This driver has been tested using HARP version 2.0 and 2.1. The HARP stack uses only UBR QOS of this driver. 3) Berkeley Packet Filter: This driver can send transmitted/received PDUs through Berkeley Packet Filter (BPF) routines for ATM filtering/displaying. 4) Permanent MBUF: The NICStAR pre-loads receive buffers before receiving PDUs. To prevent a constant construction/destruction of a MBUF and its external buffer, I add a new flag in mbuf structure and a change the the MFREE macro. This change keeps the external buffer in a PERM MBUF. The FreeBSD NICStAR device driver maybe down-loaded from: ftp://ftp.cs.ndsu.NoDak.edu/pub/freebsd/atm/nicstar.tgz Mark Tinguely tinguely@plains.NoDak.edu Computer Science http://www.cs.ndsu.NoDak.edu/~tinguely/ 258 IACC (701) 231-7786 North Dakota State University Fargo, North Dakota 58105 From owner-freebsd-atm Fri Sep 5 13:38:24 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id NAA00379 for atm-outgoing; Fri, 5 Sep 1997 13:38:24 -0700 (PDT) Received: from plains.NoDak.edu (tinguely@plains.NoDak.edu [134.129.111.64]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id NAA00374 for ; Fri, 5 Sep 1997 13:38:21 -0700 (PDT) Received: (from tinguely@localhost) by plains.NoDak.edu (8.8.5/8.8.5) id PAA01824 for freebsd-atm@freebsd.org; Fri, 5 Sep 1997 15:38:13 -0500 (CDT) Date: Fri, 5 Sep 1997 15:38:13 -0500 (CDT) From: Mark Tinguely Message-Id: <199709052038.PAA01824@plains.NoDak.edu> To: freebsd-atm@freebsd.org Subject: Correction: Announcing IDT NICStAR device driver Sender: owner-freebsd-atm@freebsd.org X-Loop: FreeBSD.org Precedence: bulk It was pointed out to me that I have the wrong part number for the newer release of the NICStAR. I said it is "77221" when I should have said "77211". Sorry for the inconvenance. --mark.