Subject: Re: elink3 and ssto weirdness
To: None <port-i386@NetBSD.ORG>
From: Christos Zoulas <christos@nyc.deshaw.com>
List: port-i386
Date: 08/06/1997 16:16:33
In article <199708061433.KAA11831@sparky.ic.sunysb.edu> cshapiro@sparky.ic.sunysb.edu (Carl S Shapiro) writes:
>I have been trying to get the 1.2G snapshot installed on a number of
>machines with 3Com network cards (both ISA 3c509 and PCI 3c590 cards)
>that use the ep driver.
>
>I am able to boot off of the SSTO floppy and the ep driver successfully
>probles my ethernet card.  I am able to configure the interface and
>ping various parts of my network.  However, if I try to ping the
>interface, I get error after the first ICMP response that says
>something to the extent of "failed to clear cached route".  I still
>able to get out onto the wire, but the machine panics when start an ftp
>session in order to pull over the distribution sets (I never get as far
>as being able to do a "get (file)", usually dies when I am ls'ing
>around).

There was a bug in libc where 127.1 was not being parsed, thus
the loopback interface did not get set-up. This causes the above problem.

>I have gotten the exact same results with a several 386 25's and 486's
>that have 3c509's in addition to a couple of Pentiums with 3c590's.  I
>do not suffer from these problems when using the 1.2.1 boot disks, nor
>does this problem occur if I boot off of the 1.2.1 disks, and then
>install the 1.2G snapshot and boot a 1.2G generic kernel off the hard 
>disk.

This is a very elusive bug that we've seen on the 3com cards related
to the MGET macros. Try using the non macro versions.

christos