[interfacekit] Re: R3 legacy code in R5 drivers?!?

Hi,

One more update on this subject (I expect).

In order to get the matrox driver going, I need to correct something stupid I 
did. If you actually use BWindowScreen acceleration you'll lock up the 
machine or app_server or so: 
I manage to kill the acc_engine 'pulse'. 

I have to place some static vars I put into mga_acc.c into the shared_info 
struct, so the clone has the info to which the engine was inited as well.

Anyone working with the nVidia driver shouldn't have this problem as I did 
not declare static global vars within that file there.

I'll update the Matrox driver to 0.15 first before I'll continue on the 
nVidia driver.


It's nice BTW to create two logfiles: one for the main accelerant and one for 
the clone. This is how I found it ;-)
The clone does not set a mode or something: it's just directly used to do acc 
(after exporting the required hooks).

--------
Just for fun:
Here's a snippet from the clone log with _some_ acc engine logging which I'll 
remove again (machine _very_ slow....):

init_common: logmask 0xffffffff, memory 0MB, hardcursor 0, usebios 0, 
greensync 0
CLONE_ACCELERANT: called!
CLONE_ACCELERANT: cloning was succesfull.
WAI
DAC: setting palette
Acc: Exporting hook B_SCREEN_TO_SCREEN_BLIT.
DAC: setting palette
ACQ
BLT: cnt 1
REL
WAI
ACQ
BLT: cnt 1

On the other logfile I could see the pageflipping working 'simultaneously', 
done by the main accelerant (PageFlipper demo app) using 'triple buffering' 
in a virtualscreen of 640 (W) x2048 (H) in a 640x480 screen:

DAC: setting palette
MOVE_DISPLAY: h 0, v 0
CRTC: setting card RAM to be displayed bpp 8
CRTC: startadd: 0
CRTC: frameRAM: 30c00000
CRTC: framebuffer: 30c00000
MOVE_DISPLAY: h 0, v 480
CRTC: setting card RAM to be displayed bpp 8
CRTC: startadd: 9600
CRTC: frameRAM: 30c00000
CRTC: framebuffer: 30c00000
MOVE_DISPLAY: h 0, v 960
CRTC: setting card RAM to be displayed bpp 8
CRTC: startadd: 12c00
CRTC: frameRAM: 30c00000
CRTC: framebuffer: 30c00000
MOVE_DISPLAY: h 0, v 0

-----


Best regards,

Rudolf.

Other related posts: