>Number:         18599
>Category:       port-i386
>Synopsis:       Problem with epic driver on NetBSD 1.6 with a 100Mb link
>Confidential:   no
>Severity:       non-critical
>Priority:       medium
>Responsible:    port-i386-maintainer
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Wed Oct 09 08:36:00 PDT 2002
>Closed-Date:
>Last-Modified:
>Originator:     Jerome Sobecki
>Release:        NetBSD 1.6
>Organization:
Jerome Sobecki       |   Institut Pasteur 
sobek@pasteur.fr     |   Pole Informatique - Systemes et Reseau
>Environment:
	
	
System: NetBSD chopin.sis.pasteur.fr 1.6 NetBSD 1.6 (GENERIC) #0: Sun Sep 8 19:43:40 UTC 2002 autobuild@tgm.daemon.org:/autobuild/i386/OBJ/autobuild/src/sys/arch/i386/compile/GENERIC i386
Architecture: i386
Machine: i386
>Description:
I have a SMC card (83c17) in a i386 computer.

When I use a 10 Mb connection, all is OK (I can have around 6 Mbps when I get
files from a local ftp server). But when I use a 100 Mb link, I can'thave
better than 300 kbps. And no collisions are detected. 
Also, I have some unexpected "lost carrier", and a ping to another computer
of the local network takes around 6s, against less than 0.2ms with the 10 Mb
link.

There is no difference between GENERIC kernel and home compiled.

The problem appear even if we force duplex on the router.

In fact, the ftp connection (at least) is 'by block', that is to say there are  
alternatively periods when the card works around 500kb (sometimes a little more)
and the computer is then pingable, and period when the card is blocked 
(computer is not pingable and LED doesn't react anymore). It doesn't look to 
have real periodicity in this, but working period doesn't exceed 2 or 3 seconds 
in general.

The 100Mb connection is OK when I use a 3Com card in my box (ex driver), and
so when I use a SMC card under NetBSD 1.5.1.

	
>How-To-Repeat:
I tried with 3 cards (2 from the same pack and one older) on 2 boxes and with
2 different 100Mb links (but the same router), and always had the same problem.
>Fix:
	
>Release-Note:
>Audit-Trail:
>Unformatted: