Age | Commit message (Collapse) | Author |
|
__driUtilCreateNewScreen returns NULL.
|
|
|
|
- Don't lock in savageDDUpdateStatte, don't call savageEmitHwStateLocked
- Need to grab the lock for texture uploads now
- New SAVAGE_NEW_CULL bit in new_state to make sure that culling state and
raster_primitive are not examined at the wrong times.
|
|
points and lines.
|
|
|
|
commands, no vertex DMA.
|
|
|
|
> of levels, including 0.
|
|
texture heaps into account.
|
|
Only available with Xlib driver for now.
Assorted clean-ups related to Draw/ReadBuffer().
Renamed FRONT_LEFT_BIT -> DD_FRONT_LEFT_BIT, etc.
|
|
|
|
|
|
A bit more cosmetics.
Improved state emit on Savage 3D/IX/MX.
|
|
|
|
in another small speedup. Fixed a problem that was masked by that
WAIT_IDLE_EMPTY:
- flush command buffer and WAIT_IDLE_EMPTY before uploading textures
|
|
|
|
|
|
|
|
|
|
|
|
|
|
buffer ..
|
|
|
|
|
|
|
|
|
|
and r128_sarea.h since they are redundant now.
|
|
|
|
of glx/mini. removes glx/mini/drm.h glx/mini/sarea.h
|
|
|
|
fine ..
|
|
|
|
This finally really fixes assertion failures in savageBindTexture.
I think the same situation with several contexts is still broken though.
|
|
|
|
|
|
- make sure that vertex setup is initialized
- Savage seems to need the W coordinate for smooth shading
|
|
were getting in savageBindTexture.
|
|
|
|
|
|
|
|
|
|
|
|
ported from the r200 driver, r200_screen.c rev. 1.8 - 1.12)
|
|
|
|
which registers changed in all the functions that change the state. Instead
check which registers changed in the end when the hardware state is updated.
Tried to avoid cosmetic (whitespace) changes in this commit.
|
|
first place, and make the message a bit clearer.
+ Remove some extraneous whitespace.
|
|
|
|
erroneously marked as GLX_SLOW_CONFIG.
|
|
|
|
savage vertex formats.
|