Oversized ethernet problems

P.A.M. van Dam (Pascal) pam@ramoth.xs4all.nl
Sun Oct 17 12:02:53 1999


On Sun, Oct 17, 1999 at 11:07:46AM -0400, Donald Becker wrote:
> On Sun, 17 Oct 1999, P.A.M. van Dam (Pascal) wrote:
> 
> > Like some others on this list I'm also experiencing the oversized ethernet
> > frame phenomenon. My setup is a AMD K6/2 300Mhz with two rtl8139 cards,
> > kernel is 2.2.12 with stock rtl8139 driver. Also tried the latest driver
> > from cesdis with no better results.
> 
> Which driver version?
> (People are often wrong when they say "the latest driver verison".)

[/var/log/messages]
Oct 17 16:08:45 ladystrange kernel: rtl8139.c:v1.08 6/25/99 Donald Becker http://cesdis.gsfc.nasa.gov/linux/drivers/rtl8139.html 
Oct 17 16:08:45 ladystrange kernel: rtl8139.c:v1.08 6/25/99 Donald Becker http://cesdis.gsfc.nasa.gov/linux/drivers/rtl8139.html 
Oct 17 16:08:45 ladystrange kernel: eth0: RealTek RTL8139 Fast Ethernet at 0xe800, IRQ 11, 00:00:b4:90:a8:d4. 
Oct 17 16:08:45 ladystrange kernel: eth1: RealTek RTL8139 Fast Ethernet at 0xec00, IRQ 10, 00:00:b4:93:e6:90. 

----

> 
> And which chip version?  The rtl8139 or rtl8139A?  (It matters in this
> case.)

Driver is v1.08, chip version appears to be rtl8139 according to the driver's
output.

> 
> 
> > Oct 17 13:13:50 ladystrange kernel: eth1: Oversized Ethernet frame, status cccccccc!
> > Oct 17 13:13:50 ladystrange kernel: eth1: Abnormal interrupt, status 00000021.
> 
> With the RTL8139A chip this may indicate a link change interrupt, not a Rx
> underflow.  Is the link LED status steady while this is happening?

I'll force an oversized frame now; wait a minute...
Ok... Let's check the LEDs.

LINK LED is steady green. Which means. LINK == OK.

> 
> > 		if (rx_status & RxTooLong) {
> > 			if (rtl8129_debug > 0)
> > 				printk(KERN_NOTICE"%s: Oversized Ethernet frame, status %4.4x!\n",
> > 					   dev->name, rx_status);
> > 			tp->stats.rx_length_errors++;
> > 			/* A.C.: The chip hangs here. */
> 
> I believe that Alan's comment is wrong: the chip doesn't have a bug, but if
> there is a data transfer problem, e.g. corrupted PCI transfers, the driver
> ends up at this point.
> 
> Are you running an overclocked system?

Nope. Just clean 300Mhz AMD K6/II on an Epox MVP3 mobo.

Hope this helps a little.

Best regards,

	Pascal

> 
> Donald Becker
> Scyld Computing Corporation, and
> USRA-CESDIS,   becker@cesdis.gsfc.nasa.gov
> 
 | To unsubscribe, send mail to Majordomo@cesdis.gsfc.nasa.gov, and within the
 |  body of the mail, include only the text:
 |   unsubscribe this-list-name youraddress@wherever.org
 | You will be unsubscribed as speedily as possible.