Subject: Re: LCSPX Driver
To: Reinhard Radtke <fam.radtke@t-online.de>
From: Blaz Antonic <blaz.antonic@siol.net>
List: port-vax
Date: 04/11/2005 15:33:00
Hello,

> I am not longer found the thread Accelerated LCSPX driver started by
> Blaz Antonic. Is this project killed?
> Is the driver ready for X11 ?

The driver is pretty much finished as far as console is concerned. It
uses acceleration in console mode which increases performnce
approximately 10-100 times over dumb operation mode and brings it on par
with accelerated LCG graphics of 4000/60 and VLC machines, not to
mention support for all the hardware gimmicks BT459 offers (namely
hardware cursor, full 8 bit read/write accress and palette changing - 8
bit planes out of 24 bit palette are supported). 

The acceleration used is more crude than that of LCG driver bnut should
nevertheless be useful for X eventually, if somebody can finally move
their behind and add X support for it - i can provide the technical
details and/or kernel interface (the latter seems like a better idea).
Dumb framebuffer wscons X server shoudl work out of the box. My idea was
to add an interface to kernel driver which would handle X server
requests (for all 16 basic rasterops, palette changing and cursor
changing and movement operations). LCG driver woudl use same kernel
interface so same X server would work on both types of VS 4000 low-cost
graphics boards without modifications (LCG accelerated functions can
bring the system down completely by deliberately issuing malicious
command packet sequence to LCG hardware so any parsing and handling of
such requests should IMHO be done in kernel and not by a user-space
application such as X server).

You don't see the driver in main source tree because nobody bothered to
submit the driver yet and i don't have access so i cannot myself (not
that i'd want the access anyway, i'd just like more people to test it so
i can finally refine it, add accelerated X support and then have it
submitted to the main tree).

BTW: note that there is a bug in keyboard code which results in keyboard
not being recognized on 4000/9x when you use "glass terminal" rather
than serial console. Consequently the keyboard just won't work under
NetBSD (unless you boot with serial console, of course). While this has
absolutely nothing to do with LCSPX driver it does impair its usefulness
somewhat because one usually still needs serial console attached even if
working with monitor and keyboard (= using LCSPX). I don't know anything
about serial port programming on 4000/0x so i have absolutely no idea
how to fix it - it is a trivial bug though: the sequence that triggers
an interrupt (for "auto" detection) on non 4000/9x class machines
doesn't work on 4000/9x so system assumes there's no keyboard attached.
A wordaroudn is simple (as rpeorted by Michael Hitch) but i'd rather see
the actual problem fixed.

Blaz Antonic
-- 
Hi! I'm a signature virus!
Copy me into your signature to help me spread!