kern/64556: if_sis short cable fix problems with NetGear FA311's

Thomas Hurst tom at hur.st
Sun Mar 21 20:30:21 PST 2004


>Number:         64556
>Category:       kern
>Synopsis:       if_sis short cable fix problems with NetGear FA311's
>Confidential:   no
>Severity:       serious
>Priority:       low
>Responsible:    freebsd-bugs
>State:          open
>Quarter:        
>Keywords:       
>Date-Required:
>Class:          sw-bug
>Submitter-Id:   current-users
>Arrival-Date:   Sun Mar 21 20:30:21 PST 2004
>Closed-Date:
>Last-Modified:
>Originator:     Thomas Hurst
>Release:        FreeBSD 5.2.1-RELEASE-p3 i386
>Organization:
>Environment:
System: FreeBSD voi.aagh.net 5.2.1-RELEASE-p3 FreeBSD 5.2.1-RELEASE-p3 #0: Wed Mar 17 21:29:20 GMT 2004 root at voi.aagh.net:/usr/obj/usr/src/sys/VOI i386

	This is a 512MB BP6 running dual 533MHz Celerons, dual NetGear FA311 NIC's,
	and has been running reliably for years, although I recently upgraded the
	CPU's.
>Description:
I'm running a pair of NetGear FA311's on FreeBSD 5.2.1-RELEASE-p3,
updated about 5 days ago.  These have been fine on 4-STABLE for the past
couple of years, except for one or two issues which I'm happy to blame
on my Motorola SB4100 cable modem, although I'm now wondering if this is
wise (symptoms were dialup-like performance not related to ISP; temporarily
dropping to 10Mbps seemed to resolve it).

My last reboot was down to an odd network issue where, basically, my
cable modem (on sis1) stopped responding to requests; I first thought it
was an issue with my provider, but moving the CM to another machine was
fine.  Checking /var/log/all, I found:

 all.5.bz2:Mar 16 07:58:34 voi kernel: sis1: Applying short cable fix (reg=e8)
 all.5.bz2:Mar 16 07:59:03 voi kernel: sis1: Applying short cable fix (reg=0)
 all.5.bz2:Mar 16 07:59:10 voi kernel: sis1: Applying short cable fix (reg=0)
[repeat lots of times]
 all.5.bz2:Mar 16 08:36:29 voi kernel: sis1: Applying short cable fix (reg=0)
 all.5.bz2:Mar 16 08:37:00 voi kernel: sis1: Applying short cable fix (reg=b)
 all.5.bz2:Mar 16 08:37:05 voi kernel: sis1: Applying short cable fix (reg=e8)
 all.5.bz2:Mar 16 08:37:12 voi kernel: sis1: Applying short cable fix (reg=e8)
 all.5.bz2:Mar 16 08:37:22 voi kernel: sis1: Applying short cable fix (reg=0)
[repeat lots of times]
 all.5.bz2:Mar 16 08:39:16 voi kernel: sis1: Applying short cable fix (reg=0)
 all.5.bz2:Mar 16 08:41:26 voi kernel: sis0: Applying short cable fix (reg=f3)
 all.5.bz2:Mar 16 08:41:34 voi kernel: sis1: Applying short cable fix (reg=23)
 all.5.bz2:Mar 16 08:41:42 voi kernel: sis1: Applying short cable fix (reg=0)
[repeat a few more times]
 all.5.bz2:Mar 16 08:42:12 voi kernel: sis1: Applying short cable fix (reg=0)
 all.5.bz2:Mar 16 08:43:43 voi kernel: sis0: Applying short cable fix (reg=f7)
 all.5.bz2:Mar 16 08:43:43 voi kernel: sis1: Applying short cable fix (reg=23)
 all.5.bz2:Mar 16 08:43:43 voi kernel: sis1: Applying short cable fix (reg=e8)

I finally rebooted in exasperation, although next time I'll try dropping
to 10Mbit again.

e8 seems to be the normal value for reg for sis1 (my CM), btw.

Now, earlier today I was playing a FLAC file in foobar 2000; I
ReplayGained it happily at 4M/s, then gawped as I saw the update
operation write back at 400-900k/s, while my console on FreeBSD was
filled with:

 Mar 22 00:20:23 voi kernel: sis0: Applying short cable fix (reg=e8)
 Mar 22 00:20:54 voi last message repeated 21 times
 Mar 22 00:21:09 voi last message repeated 11 times

Things seem fine now, no reboot or any other change other than the
removal of the QoS packet scheduler in my Windows XP desktop.
>How-To-Repeat:
	Use one or two FA311's for an extended period using short cables. Not tested
	on longer cables.
>Fix:
	Tempt me to get some Intel NIC's?  Mmmm, GigE... :o
>Release-Note:
>Audit-Trail:
>Unformatted:


More information about the freebsd-bugs mailing list