From owner-freebsd-hackers Wed Jan 22 13:52:36 1997 Return-Path: Received: (from root@localhost) by freefall.freebsd.org (8.8.5/8.8.5) id NAA29835 for hackers-outgoing; Wed, 22 Jan 1997 13:52:36 -0800 (PST) Received: from dan.emsphone.com (-@quake.emsphone.com [199.67.51.101]) by freefall.freebsd.org (8.8.5/8.8.5) with ESMTP id NAA29830 for ; Wed, 22 Jan 1997 13:52:29 -0800 (PST) Received: (from dan@localhost) by dan.emsphone.com (8.8.2/8.8.2) id PAA28119; Wed, 22 Jan 1997 15:51:24 -0600 (CST) Message-ID: Date: Wed, 22 Jan 1997 15:51:23 -0600 From: dnelson@emsphone.com (Dan Nelson) To: dwhite@resnet.uoregon.edu (Doug White) Cc: hackers@FreeBSD.ORG Subject: Re: Erroneous Ierrs from vx0 References: X-Mailer: Mutt 0.56 Mime-Version: 1.0 In-Reply-To: ; from "Doug White" on Jan 22, 1997 12:45:25 -0800 Sender: owner-hackers@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk In the last episode (Jan 22), Doug White said: > Hello! > > I just purchased a 3com Etherlink XL PCI ethenet card and it's working > great on 2.2-BETA, except for one thing: > > gdi,ttyp1,~,12>netstat -i > Name Mtu Network Address Ipkts Ierrs Opkts OerrsColl > vx0 1500 00.a0.24.ba.13.f8 76188 44915 23649 00 > vx0 1500 128.223.170/2 gdi 76188 44915 23649 00 > > I wasn't getting these errors from an NE2000, and the card appears to be > working perfectly otherwise. Same cable, same resources. I tried > rebuilding the kernel from clean sources with no change. > > I don't see anything in the archives regarding this. :) I had similar problems just last week trying to NFS mount an SCO disk onto a FreeBSD machine, both P6/200's. The SCO box had a PCI Intel EtherExpress Pro/100, the BSD box had a PCI 3c905 Fast Etherlink XL/100. I would only get the first 2 frags from the 8k NFS transactions. NFS's retries failed, since it would resend the whole 8k packet over (and, of course, only the first 2 frags would make it to the BSD box). A "netstat -i" on the BSD end gave me lots of ierrs, but I saw no console messages. Seems the vx0 driver doesn't print error messages unless you go into debugging mode. Do an "ifconfig vx0 debug" to enable them, and watch all the "RX overrun"/"packet overrun"/"fifo underrun" messages spew out. I swapped the 3com card with an Intel EE Pro/100B (fxp driver) and my problems disappeared. I suspect one of the DEC chipset cards would do fine also. I've noticed that when I run 3Com's DOS config program that the card reports 5K of receive buffers and 3K of send buffers. Is this really all there is on the card? If so, I can understand the NFS problem, but I can't believe a 100mbit card woud have such small buffers. -Dan Nelson dnelson@emsphone.com