Index: [thread] [date] [subject] [author]
  From: Jon M. Taylor <taylorj@gaia.ecs.csus.edu>
  To  : GGI mailing list <ggi-develop@eskimo.com>
  Date: Tue, 25 May 1999 23:01:40 -0700 (PDT)

GGIMesa + KGI Savage4 driver update

	I did some more work on the GGI target in the Mesa CVS tree today. 
The missing top-level GGI tree which contains the demos and ggiglut.c was
comitted and a bunch of makefile fixes were also.  Everything builds and
installs correctly, all the shared libraries load properly, etc.  gears.c
runs but some of the Mesa internal structs have changed so it still
segfaults.  The fault is in some _very_ strange code where the 
preprocessor is used to create template functions.  Hard to debug, but I 
should have it licked tomorrow.

	After that I can start working on a GGIMesa target for my
"Creative driver" as I have been calling it, which I can now reveal is for
the S3 Savage4 chipset.  I just recently got multitextured triangle
drawing and running in the driver, so now it is time to do the Mesa code. 
That article by John Carmack that was posted a while back tells exactly
how to optimize the Mesa-driver bridge.  All I need to be able to do is
make an extension stub for glDrawElements() and implement it in the KGI 
target as buffer fills and flush ioctls.

	If anyone out there has or is getting a 3D Blaster Savage4, let me
know and I'll tidy the driver up a bit and release a binary and a header
file.  The interface is pretty minimal right now - ioctl-per-accelop and 
no DMA buffers.  Basic 2D accels (ACCEL_PUTBOX, ACCEL_DRAWBOX, and 
ACCEL_DRAWLINE) are in, as well as some new 3D ioctls:

* Download texture 0
* Download texture 1
* Set state (texture and blendinfo)
* DrawPrimitive() [draw multitextured triangle]

slow as all get out, but I can implement a fully decoupled DMA buffer
approach, strip/fan/quad drawing, userspace buffer mapping via /proc, etc
as soon as I have the Mesa framework in place to make use of it and the
driver is a bit less buggy than it is right now.  'Alpha' is strictly the 
word for now.  Nevertheless, it all looks pretty promising.

Jon

---
'Cloning and the reprogramming of DNA is the first serious step in 
becoming one with God.'
	- Scientist G. Richard Seed


Index: [thread] [date] [subject] [author]