Duplex mode

Donald Becker becker@cesdis.gsfc.nasa.gov
Tue Nov 2 09:34:10 1999

On Tue, 2 Nov 1999, Dirk Raebiger wrote:

> I'm using v1.08a of rtl8139 together with linux kernel 2.2.13. The network
> administrator at the institute here says that my machine (gollum) sends in
> full-duplex mode. This is also what the led of the card says (full duplex).

Incorrect duplex is not a typical problem for a rtl8139.

> The card (Mercy with RealTek 8139) can be configured by a DOS tool and its
> EEPROM is set to 10MBIT/half duplex, how it should be.

Hmmm, the Linux driver doesn't know about a field defined for the default
media type type.

> The 'rtl8139-diag' says the following:
>  The chip configuration is 0x10 0x2c, MII half-duplex mode.

Then the chip is in half duplex mode.

> Parsing the EEPROM of a RealTek chip:
>   Configuration register 0/1 -- 0x5c / 0xc2.
> RealTek chip registers at 0xc000
>  0x060: 0000600e 05e1780d 00000000 00000000 00000000 000f77c0 38fa8388 ad38de43.

Run 'mii-diag' to see the current transceiver status.

> ifconfig reports errors, matching the Cisco analysis of our network
> administrator:
> eth0      Link encap:Ethernet  HWaddr 00:E0:7D:02:9D:4A
>           RX packets:42987 errors:1 dropped:0 overruns:0 frame:0
>           TX packets:42063 errors:0 dropped:0 overruns:0 carrier:0
>           Collisions:6716

Please report the contents of /proc/net/dev rather than the summarized
output of 'ifconfig'.

I would hardly call this reporting errors -- a single Rx error could be from
many sources.

You are getting collisions which is a certain indication that you are *not*
in full duplex mode.

> Can anyone help me, why the card thinks it's in full duplex? What can I
> do?

There is no indication that the card is in full duplex mode.
None at all.

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.