Subject: LANCE driver now fixed.
To: None <port-vax@NetBSD.ORG>
From: Anders Magnusson <ragge@ludd.luth.se>
List: port-vax
Date: 03/21/1998 15:32:14
I have just finised my rewriting of the lance code, for VAXststions.
It doesn't get any "device timeout" anymore, and ftp on my 3100/m38
can send/receive rather constantly with 250K/s, which seems appropriate
for such slow machine. Also, the device driver became much faster than
the generic one; a receive interrupt is now handled in 65 uS instead of
280uS. This will gain performance on diskless machines quite a bit.

Troubles with the MI device driver:
1) DEC wants a special order of touching the register at interrupt time
   to avoid loosing interrupts.
2) All function calls caused significant overhead in the interrupt handler.
3) If the lance misses a packet the chip must be reset(!), otherwise
   the lance internal buffers will be clobbered. This is not done in the 
   MI driver and may be the actual reason to the LANCE_REVC_BUG fiddling.

Also, because there now a MD driver, the device name cannot be le, so
I have changed it to ln (Ultrix naming).

I will check it in as soon as I have cleaned it up a bit.

-- Ragge