[2168] in linux-net channel archive
Re: EtherExpress problems in 1.3.75
daemon@ATHENA.MIT.EDU (Philip Blundell)
Wed Mar 20 07:24:30 1996
Date: Wed, 20 Mar 1996 12:18:57 +0000 (GMT)
From: Philip Blundell <pjb27@cam.ac.uk>
To: Steven L Baur <steve@miranova.com>
cc: linux-net@vger.rutgers.edu
In-Reply-To: <m2ybowz24c.fsf@deanna.miranova.com>
On 19 Mar 1996, Steven L Baur wrote:
> The EtherExpress is practically unusable on a Pentium 100, although
> I had decent results with a 486DX33 and earlier '70's kernels. I get
> repeated
> Mar 19 10:59:30 sandy kernel: eth0: CU wedged, status 0140 0000, resetting...
>
> on the machine with EtherExpress. And UDP bad checksums on the other side:
>
> Mar 19 10:58:38 deanna kernel: UDP: bad checksum. From C0A80197:677 to C0A80101:2049 ulen 148
>
> The resetting and recovery is tragic since I was able to complete an
> entire kernel compile apparently without hard error, only to have the
> link fail due to bad object code. Source code was NFS'ed, and object
> code was created on the machine with the EtherExpress.
1.3.75's EtherExpress driver is broken. Although I wouldn't expect that
particular problem (most people saw an Oops at boot time) it's not
impossible that it could be caused. Try upgrading to 1.3.76. If that
still doesn't help, let me know and I will look into it. Certainly
nothing that should have provoked your problem has been deliberately
changed for 1.3.75 -- you ought to see practically the same results as
with earlier 1.3.7x kernels.
P.
