Subject: Re: DEQNA trouble.
To: None <PORT-VAX@netbsd.org>
From: None <carlini@andnot.lkg.dec.com>
List: port-vax
Date: 06/17/1999 17:28:28
> The problem is to get the deqna into ALLMULTI mode. The programmers
> manual states that:

>
>If the setup packet len is 128 bytes, then setting bit 0 will turn on 
>allmulti reception and setting bit 1 will put the interface in 
>promiscous mode. BUT:


I only have the User's Guide not the Programmer's Guide so there may be some
differences here. But reading from section 4.3.3:

> The packet len is stored -(len/2) in the packet, which looses bit 0.


You should never loose bit 0 - you never know where a loose bit might end up
:-) But seriously the length is the length in words. Except that 4.3.3/4.3.3.1
conspire to make me believe that the buffer length word is interpreted
differently. Values of 000 to 177 (octal) define a packet which tells the DEQNA
which addresses to respond to. ,values greater than 128 (it then says 177
octal, so I guess it means values greater than 127...) get interpreted
differently. So given that
bit 0 - all multicast
bit 1 - promiscuous
bits 3:2 - plays with LEDs
bits 6:4 - sanity timer
if you want the effect of just turning on all multicast addresses you would:

1) use a "normal" setup packet (with a length between 000 and 177 octal) to
tell the DEQNA which address(es) to respond to (set all unused ones to the
physical address)

2) next build a setup packet with length 0x81 and see what happens (it should
turn on multicast packet reception).

Take this with a suitable grain of salt - I've never played at programming a
DEQNA.

Antonio