From owner-freebsd-hackers Mon May 10 15:49:33 1999 Delivered-To: freebsd-hackers@freebsd.org Received: from herring.nlsystems.com (nlsys.demon.co.uk [158.152.125.33]) by hub.freebsd.org (Postfix) with ESMTP id A565515183 for ; Mon, 10 May 1999 15:49:09 -0700 (PDT) (envelope-from dfr@nlsystems.com) Received: from localhost (dfr@localhost) by herring.nlsystems.com (8.9.3/8.8.8) with ESMTP id XAA58966; Mon, 10 May 1999 23:49:05 +0100 (BST) (envelope-from dfr@nlsystems.com) Date: Mon, 10 May 1999 23:49:05 +0100 (BST) From: Doug Rabson To: Dennis Cc: Wilko Bulte , hackers@freebsd.org Subject: Re: What is a "transmit underflow"? In-Reply-To: <199905102218.SAA11495@etinc.com> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-hackers@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG On Mon, 10 May 1999, Dennis wrote: > In case some of you care, here is what it is. The proper term is > "transmitter underrun", but that is not really important. > > Once a frame has begun transmission (and this goes for HDLC and other > continuous packet protocols as well), all of the bytes of the frame/packet > must be in contiguous time slots until the complete end of frame has been > sent. With a bus master, the board must be able to fetch the data from > system memory in time to fill all the slots. If the device cannot get the > bus (usually because some other device has it) in time to fill the slot, > then the frame must be aborted and an underrun occurs. If you get a lot of > these it basically means that you are out of bus bandwidth or that some > other device is hogging the bus. This is typically why you will start to > lose packets, no matter how fast your CPU is, at some point where > contention for the bus is too high. Since most devices give priority to the > receiver, problems usually manifest themselves as transmitter underruns first. I don't think this is the case with the tulip. I have one alpha machine which exhibits the problem and it *always* happens on the first large nfs write the machine makes. Nothing else in the machine is using the bus for anything significant at the time. -- Doug Rabson Mail: dfr@nlsystems.com Nonlinear Systems Ltd. Phone: +44 181 442 9037 To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hackers" in the body of the message