Age | Commit message (Collapse) | Author |
|
|
|
|
|
|
|
|
|
texture sampling and texture environment, optimize away texture sampling for units if the result is not used, always emit the env instructions in-order and try to eliminate GL_REPLACE env instructions.
|
|
|
|
_mesa_init_driver_functions()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
GL_ARB_texture_env_crossbar for i810. This passes both demos/texenv and all
of glean's texCombine tests.
|
|
|
|
|
|
hardware format of the texture, and SetTexImages is supposed to use the
format selected by ChooseTextureFormat. However, both routines were making
their choices based on the texture's BaseFormat. This is wrong.
ChooseTextureFormat uses BaseFormat and SetTexImages uses
TexFormat->MesaFormat.
Once SetTexImages was fixed to use the right format values,
ChooseTextureFormat was cleaned up. It now uses the few available texture
formats supported by the i810 in a smarter way. This should improve the
quality of LUMINANCE, LUMINANCE_ALPHA, and INTENSITY textures.
I tested this by cycling through all the texture formats in demos/texenv and
tests/yuvsquare.
|
|
|
|
erroring out when it shouldn't. The errno could be changed by usleep() between
the ioctl call and the loop check, if a signal was received. This could cause
an EAGAIN return from the DRM_RADEON_TEXTURE ioctl to not loop again.
Instead of checking errno, check thevalue of ret itself, since it is a saved
(and sign-flipped) copy of errno from the ioctl call.
|
|
libdrm can be had from:
http://people.freedesktop.org/~ajax/libdrm/
|
|
|
|
but, for whatever reason, the extension wasn't enabled.
Add some comments to i810ChooseTextureFormat. There's some strangeness with
i810 texture formats.
|
|
progs/tests/blendsquare.
|
|
primarilly needed to support functions that are internally implemented using
extension entry points (e.g., glBlendFunc uses glBlendFuncSeparate).
|
|
This can happen when all the texture enables for a texture unit are disabled.
This fixes bugzilla #3195.
|
|
The GL_RENDERER string now accurately contains the chip name. This was only
tested on a "plain" i810 (PCI ID 8086:7121).
|
|
Add additional checks for the *DRIRec info structure passed in from the
device driver. This ensures that things fallback to indirect rendering if
the DDX driver has had modifications (i.e. removal of the drmAddress field).
|
|
structure to the 3D drivers view of it, and abort if they don't match.
This traps the case with the drmAddress removal in the DDX driver.
|
|
extension wasn't enabled. I have verified its correct function with Mesa's
stencil_wrap test. It is enabled on both Savage4 and Savage3D. Since
Savage3D uses a software fallback for *all* stencil operations, this is safe.
|
|
'tcl_mode=1'. This fixes bugzilla #3972.
|
|
proper DRM_CAS_RESULT to declare the variable used to store the result
of DRM_CAS. Second, only use the "real" versions of LOCK_HARDWARE and
UNLOCK_HARDWARE on SPARC. That's the only platform where the hardware
can really exist.
|
|
attributes in display lists. If a driver doesn't add entry points for
GL_NV_vertex_program, it will explode. This patch adds GL_NV_vertex_program
to the list of default extensions whose entry points are added by
driInitExtensions.
Thanks go to Roland Scheidegger for tracking the problem down and suggesting
the fix.
|
|
IN_DRI_DRIVER cruft. This was suggested by George Fufutos.
|
|
|
|
|
|
|
|
|
|
Fix suggested by Ian Romanick.
R300 driver now works without segfault.
|
|
The weak version of _glapi_DispatchTSD was removed from dri_util.c. As
stated in the bug report, this was only needed to support versions of libGL
that did not have this symbol. Versions of libGL without this symbol are no
longer supported at all, so it is totally irrelevant.
|
|
|
|
CALL_by_offset, SET_by_offset, and GET_by_offset all had various problems.
The core issue is that parts of the device-independent code in Mesa assumes
that all functions have slots in the dispatch table. This is especially
true in the display list code. It will merrilly try to set dispatch
pointers for glVertexAttrib1fARB even if GL_ARB_vertex_program is not
supported. When the GET/SET/CALL macros are invoked, they would read a 0
from the remap table. The problem is that 0 is the dispatch offset for
glNewList!
One change is that the remap table is now initialized to be full of -1
values. In addtion, all of the *_by_offset marcos misbehave in an obvious
way if the specified offset is -1. SET_by_offset will do nothing,
GET_by_offset will return NULL, and CALL_by_offset, since it uses
GET_by_offset, will segfault.
I also had to add GL_EXT_blend_func_separate to the list of default
extensions in all_mesa_extensions (src/mesa/drivers/dri/common/utils.c).
Even though many drivers do not export this extension, glBlendFunc is
internally implemented by calling glBlendFuncSeparate. Without this
addition, glBlendFunc stopped working on drivers (such as mga) that do not
export GL_EXT_blend_func_separate.
There are still a few assertions / crashes in GL_ARB_vertex_program tests,
but I don't think that these are related to any of my changes.
|
|
|
|
Still a work in progress.
|
|
(Stephane Marchesin + Dave Airlie)
|
|
for X on a miniglx fix. It can be removed later.
|
|
Code is already in DRM driver to do this.
|
|
I have tested these on miniglx and they seem to be working.
I'm sure Dave will give them a better test.
|
|
|
|
destroy __GLcontextModes structures. This eliminates the need for DRI
drivers to ever link with glcontextmodes.c.
|