[realtek] Problem with Accton EN1207D-TX

[Donald Becker]

On Tue, 9 Jan 2001, Ville Herva wrote:
> On Tue, Jan 09, 2001 at 03:35:09AM -0500, you [Donald Becker] claimed:

> > Adding the table entry is trivial, but I won't do it to the distributed
> > version until we verify that this is not just a hardware or EEPROM
> > programming problem.

> +       {"Accton/HP EN1207D-TX (Realtek RTL8139)", { 0x12101113, 0xffffffff,},
> +        RTL8139_IOTYPE, 0x80, RTL8139_CAPS, },

Yes, this is correct.

> (Of course, I don't a slightest clue on how one should add that table
> entry, but that's hardly an excuse for not trying, is it? ;) (In
> particular, what are the last 4 hexdigits supposed to be? - Well, forget
> it... :) )

They are a bitmask that says "only pay attention to these bits".
The full form also has chip version and subsystem ID.  Since unspecified
fields default to zero, omitting the additional pairs results in
match-anything behavior.

> The card was detected, but immediately after ifup eth0, it gave these
> errors:

What was the detection message?  Was the station address reasonable (not all
0xff)? 

> and while pinging (which obviously did not succeed):
> 
> eth0: Transmit timed out, status 0d 0004 media 00
> eth0: Oversized Ethernet packet, status 2dc8

Ahhh, this could very well be a RTL8139!  That status reports normal
rtl8139 operation with a Tx-done interrupt outstanding.

> eth0: RTL8139 interrupt line blocked, status ffff

.... But with a power or bus bridge problem that causes it to reset itself
or not respond.

> Then the machine spontaneously rebooted itself :).

Donald Becker				becker@scyld.com
Scyld Computing Corporation		http://www.scyld.com
410 Severn Ave. Suite 210		Second Generation Beowulf Clusters
Annapolis MD 21403			410-990-9993