Nicolai Haehnle [Fri, 13 May 2005 21:29:17 +0000 (21:29 +0000)]
Make software fallbacks work again:
We need to plug in swsetup functions into the TNL module's callbacks.
Also, resize the framebuffer size when the drawable size changes.
Nicolai Haehnle [Fri, 13 May 2005 21:00:11 +0000 (21:00 +0000)]
Remove quiescence sequence from r300ClearBuffer.
The DRM is responsible for emitting this quiescence sequence when
appropriate.
Jon Smirl [Fri, 13 May 2005 18:31:35 +0000 (18:31 +0000)]
First attempt at getting egl support up on dumb framebuffer.
Seems to be mostly working. Not all of egl API is implemented.
Nicolai Haehnle [Fri, 13 May 2005 18:03:24 +0000 (18:03 +0000)]
Patch from Jeff Smith:
The type of vertex data is opaque in the emit_* functions, so use GLvoid*
instead of char*. This also eliminates some compiler warnings.
Nicolai Haehnle [Fri, 13 May 2005 17:44:22 +0000 (17:44 +0000)]
Patch from Jeff Smith:
Fix a WARN_ONCE format string.
Alan Hourihane [Fri, 13 May 2005 11:56:13 +0000 (11:56 +0000)]
Remove MAX_VISUAL limitations of 100. If running dual head this can
be easily consumed. There are no limitations now and the VisualTable
is realloced when necessary.
Keith Whitwell [Fri, 13 May 2005 09:15:50 +0000 (09:15 +0000)]
Calculate render inputs for tcl stage correctly (hint: they aren't the
same as tnl->render_inputs). Fixes recent tcl problems.
Ian Romanick [Thu, 12 May 2005 23:15:38 +0000 (23:15 +0000)]
Remove some unused structure fields, add some comments, and ifdef out some
dead code. This is just some clean-up work which should not have any
functional impact.
Keith Whitwell [Thu, 12 May 2005 10:28:43 +0000 (10:28 +0000)]
new MESA_DEBUG option: disassem
Keith Whitwell [Thu, 12 May 2005 10:22:29 +0000 (10:22 +0000)]
Rein in debug slightly
Keith Whitwell [Thu, 12 May 2005 09:31:33 +0000 (09:31 +0000)]
Turn disassem off (oops).
Keith Whitwell [Thu, 12 May 2005 09:07:12 +0000 (09:07 +0000)]
Add check for mask == NULL to the one routine missing it.
Keith Whitwell [Thu, 12 May 2005 08:54:13 +0000 (08:54 +0000)]
When lighting is enabled, but no lights are enabled, the scenecolor
becomes the result of lighting. When lighting is disabled,
pass-through incoming color value. Likewise, pass through incoming
texture values. (Based on patch by Aapo Tahkola)
Add compile-time configuration to switch between DP4 and MUL/MAD for
matrix-vector multiplications.
Ben Skeggs [Thu, 12 May 2005 03:53:18 +0000 (03:53 +0000)]
fix warnings in r300_texstate.c (Jeff Smith)
Karl Schultz [Wed, 11 May 2005 22:01:20 +0000 (22:01 +0000)]
The "P" suffix on APIENTRY and GLAPIENTRY is to be used for pointers to
functions, not for functions that return pointers. Restore protoype
for glGetString back to its correct form.
Aapo Tahkola [Wed, 11 May 2005 17:28:13 +0000 (17:28 +0000)]
VBOs.
Keith Whitwell [Wed, 11 May 2005 16:30:02 +0000 (16:30 +0000)]
Don't let FragmentProgram._Active influence choice of vertex vs pixel
fog.
Keith Whitwell [Wed, 11 May 2005 16:28:33 +0000 (16:28 +0000)]
Previously in TXP and TEX instructions, lambda was passed to
fetch_texel as zero, but I believe this is incorrect. The spec uses a
pseudocode function:
vec4 TextureSample(float s, float t, float r, float lodBias,
int texImageUnit, enum texTarget);
to specify the behaviour of TEX, TXB and TXP instructions. For TEX
and TXP, lodBias is passed as zero, TXB is passed with texcoord[4]. In our code we have
static void
fetch_texel( GLcontext *ctx, const GLfloat texcoord[4], GLfloat lambda,
GLuint unit, GLfloat color[4] )
and were passing zero and a biased lambda value respectively. The
difference is that TextureSample() would add in the lambda term
itself, while in our code the caller is expected to do this. Thus in
the TEX and TXP cases, it is necessary to pass an unbiased lambda
value for things to work out correctly.
Keith Whitwell [Wed, 11 May 2005 15:18:59 +0000 (15:18 +0000)]
Use _mesa_memset rather than memset
Keith Whitwell [Wed, 11 May 2005 15:18:23 +0000 (15:18 +0000)]
Don't let FragmentProgram._Active influence choice of vertex vs pixel
fog.
Keith Whitwell [Wed, 11 May 2005 15:17:31 +0000 (15:17 +0000)]
Test for texcoord[3] == zero before dividing. Not so sure about this
test - if texcoord[3] is zero, we'd probably be fine except for an
ASSERT in IROUND_POS() which gets triggered by the inf values
created.
Keith Whitwell [Wed, 11 May 2005 15:16:35 +0000 (15:16 +0000)]
Fix glean/conform problems. Don't generate so much output when
disassembling.
Keith Whitwell [Wed, 11 May 2005 10:30:13 +0000 (10:30 +0000)]
Fix valgrind complaint, conform vertex order test.
Keith Whitwell [Wed, 11 May 2005 08:34:19 +0000 (08:34 +0000)]
Rename temp_flag to temp_in_use. Use ctx->Const.MaxVertexProgramTemps
rather than MAX_NV_VERTEX_PROGRAM_TEMPS and deal with this possibly
being greater than 32.
Keith Whitwell [Wed, 11 May 2005 08:30:23 +0000 (08:30 +0000)]
Fix logic for allocating texture temporaries (Ben Skeggs).
Rename temp_flag, tex_temp_flag to make this clearer.
Respect ctx->Const.MaxFragmentProgramTemps limit.
Adam Jackson [Wed, 11 May 2005 01:33:50 +0000 (01:33 +0000)]
int -> size_t to fix compile on LP64.
Keith Whitwell [Tue, 10 May 2005 18:24:50 +0000 (18:24 +0000)]
Ensure programs don't overflow allocated instruction store.
Keith Whitwell [Tue, 10 May 2005 18:22:19 +0000 (18:22 +0000)]
Fix some valgrind complaints
Keith Whitwell [Tue, 10 May 2005 18:10:32 +0000 (18:10 +0000)]
fix some valgrind complaints
Keith Whitwell [Tue, 10 May 2005 18:10:00 +0000 (18:10 +0000)]
Improved detection of program changes.
Keith Whitwell [Tue, 10 May 2005 18:09:17 +0000 (18:09 +0000)]
Double-buffer generated instructions and only notify driver when the
generated program differs from the previous one.
Keith Whitwell [Tue, 10 May 2005 18:08:25 +0000 (18:08 +0000)]
Fix some valgrind complaints
Keith Whitwell [Tue, 10 May 2005 13:57:50 +0000 (13:57 +0000)]
Double-buffer generated instructions and only notify driver when the
generated program differs from the previous one.
Keith Whitwell [Tue, 10 May 2005 13:56:23 +0000 (13:56 +0000)]
reduce the use of malloc and strdup for parameter lists
Keith Whitwell [Tue, 10 May 2005 13:54:52 +0000 (13:54 +0000)]
don't be fooled into emitting padding for wpos when not used
Keith Whitwell [Tue, 10 May 2005 11:41:33 +0000 (11:41 +0000)]
Missing from previous commit
Keith Whitwell [Tue, 10 May 2005 11:40:52 +0000 (11:40 +0000)]
Update for FragmentProgram._Active
Keith Whitwell [Tue, 10 May 2005 11:39:50 +0000 (11:39 +0000)]
Better driver notification on changes.
Keith Whitwell [Tue, 10 May 2005 11:38:56 +0000 (11:38 +0000)]
get fog working with _TexEnvProgram
Keith Whitwell [Tue, 10 May 2005 10:25:16 +0000 (10:25 +0000)]
Temporary fix - delete and recreate texenvprogram so that drivers
notice when it changes.
Jouk Jansen [Tue, 10 May 2005 10:15:30 +0000 (10:15 +0000)]
Committing in .
OpenVMS compile support update for new buffer-code
Modified Files:
Mesa/src/mesa/drivers/x11/descrip.mms
Mesa/src/mesa/main/descrip.mms
Mesa/src/mesa/swrast/descrip.mms
----------------------------------------------------------------------
Keith Whitwell [Tue, 10 May 2005 09:56:02 +0000 (09:56 +0000)]
Fix texenv segfault. Reported by Ben Skeggs.
Keith Whitwell [Tue, 10 May 2005 08:58:44 +0000 (08:58 +0000)]
Try not to use the same temp reg as a TXP destination more than once,
as this also constitutes a texture indirection. Reported by Ben
Skeggs.
Daniel Borca [Mon, 9 May 2005 22:15:24 +0000 (22:15 +0000)]
pathetic attempt to accomodate new frambuffer changes (still some work to do)
Daniel Borca [Mon, 9 May 2005 22:14:58 +0000 (22:14 +0000)]
accomodate new frambuffer changes
Daniel Borca [Mon, 9 May 2005 22:14:32 +0000 (22:14 +0000)]
fix some typos in FX code
Keith Whitwell [Mon, 9 May 2005 17:59:13 +0000 (17:59 +0000)]
i915 will use _TexEnvProgram (if active)
Keith Whitwell [Mon, 9 May 2005 17:58:13 +0000 (17:58 +0000)]
Use driver functions to create TexEnvProgram, TnlProgram
Keith Whitwell [Mon, 9 May 2005 17:42:18 +0000 (17:42 +0000)]
check for null DrawBuffer values
Keith Whitwell [Mon, 9 May 2005 17:38:52 +0000 (17:38 +0000)]
Fallback (rather than fail) on unknown program targets.
Jon Smirl [Sun, 8 May 2005 05:01:03 +0000 (05:01 +0000)]
dri fb works using renderbuffers now in RGBA mode
Jon Smirl [Sat, 7 May 2005 19:56:44 +0000 (19:56 +0000)]
fbdri can draw double buffered now. Next step is to convert it
to use RenderBuffers.
Brian Paul [Sat, 7 May 2005 18:27:58 +0000 (18:27 +0000)]
remove some dead code
Brian Paul [Sat, 7 May 2005 17:27:57 +0000 (17:27 +0000)]
x86_64 stuff
Brian Paul [Sat, 7 May 2005 17:21:11 +0000 (17:21 +0000)]
added linux-x86-xcb config (Jeremy Kolb)
Brian Paul [Sat, 7 May 2005 17:14:50 +0000 (17:14 +0000)]
Patches for XCB (Jeremy Kolb)
Brian Paul [Sat, 7 May 2005 17:06:49 +0000 (17:06 +0000)]
assorted warning clean-ups for x86_64, etc (Mikko T.)
Brian Paul [Sat, 7 May 2005 16:59:58 +0000 (16:59 +0000)]
x86-64 transform optimizations (Mikko T.)
Brian Paul [Fri, 6 May 2005 22:17:24 +0000 (22:17 +0000)]
fix a bunch of html errors
Brian Paul [Fri, 6 May 2005 19:28:31 +0000 (19:28 +0000)]
fix cpp / bytesPerPixel problem (bug
1196554)
Brian Paul [Fri, 6 May 2005 15:23:54 +0000 (15:23 +0000)]
remove bad assertion, added new ones in clear_pixmap()
Brian Paul [Fri, 6 May 2005 14:44:16 +0000 (14:44 +0000)]
restore call to SetBuffer() in clear_color_buffers(), only temporary
Ben Skeggs [Fri, 6 May 2005 08:41:50 +0000 (08:41 +0000)]
Updated for EXT_framebuffer_object changes. I don't know if this is correct, but it seems to work
Jon Smirl [Fri, 6 May 2005 05:50:41 +0000 (05:50 +0000)]
Fix problem generating mode db. Miniglxtest works now.
Miniglxsample is still broken with double buffering issues.
Aapo Tahkola [Fri, 6 May 2005 03:18:40 +0000 (03:18 +0000)]
Missing setups that cause all kinds of problems with hw tnl.
Karl Schultz [Thu, 5 May 2005 21:08:07 +0000 (21:08 +0000)]
Port Mesa to build on a P64 platform (e.g., Win64). P64 platforms
use 64-bit pointers and 32-bit longs. So, operations like casting pointers
to unsigned long and back to pointer won't work. glheader.h now
includes files to define uintptr_t, which should instead be used for
this sort of operation. It is an integer type that is the same size
as a pointer.
Brian Paul [Thu, 5 May 2005 17:13:18 +0000 (17:13 +0000)]
fix for renderbuffer (Andreas Stenglein)
Brian Paul [Thu, 5 May 2005 15:09:16 +0000 (15:09 +0000)]
remove some test code
Brian Paul [Thu, 5 May 2005 14:18:31 +0000 (14:18 +0000)]
minor face-lift for web pages
Brian Paul [Thu, 5 May 2005 14:18:04 +0000 (14:18 +0000)]
fix typo
Brian Paul [Thu, 5 May 2005 13:08:06 +0000 (13:08 +0000)]
added a comment
Brian Paul [Thu, 5 May 2005 13:07:52 +0000 (13:07 +0000)]
add PROGRAM_DIRS line
Aapo Tahkola [Thu, 5 May 2005 06:51:00 +0000 (06:51 +0000)]
Getting rusty... 4278 was already there.
Jon Smirl [Thu, 5 May 2005 05:50:19 +0000 (05:50 +0000)]
Fix up origin in dri/fb driver
Calculated fbBPP
Aapo Tahkola [Thu, 5 May 2005 03:40:47 +0000 (03:40 +0000)]
Flat shading and bunch of clarifications to r300_reg.h .
Adam Jackson [Thu, 5 May 2005 03:15:08 +0000 (03:15 +0000)]
Refactor COMMON_SOURCES up to Makefile.template
Jon Smirl [Thu, 5 May 2005 00:28:37 +0000 (00:28 +0000)]
Minor touchups to fbdev renderbuffer support
Aapo Tahkola [Thu, 5 May 2005 00:28:31 +0000 (00:28 +0000)]
Pack tex coord regs at vertex program backend.
Brian Paul [Thu, 5 May 2005 00:10:18 +0000 (00:10 +0000)]
more updates for users of s_spantemp.h
Adam Jackson [Wed, 4 May 2005 22:05:02 +0000 (22:05 +0000)]
Add drirenderbuffer.c to COMMON_SOURCES
Brian Paul [Wed, 4 May 2005 20:20:02 +0000 (20:20 +0000)]
fix s_spantemp.h includes
Brian Paul [Wed, 4 May 2005 20:18:23 +0000 (20:18 +0000)]
FBO changes
Brian Paul [Wed, 4 May 2005 20:13:59 +0000 (20:13 +0000)]
fix xm_span2.c reference
Brian Paul [Wed, 4 May 2005 20:11:35 +0000 (20:11 +0000)]
Major check-in of changes for GL_EXT_framebuffer_object extension.
Main driver impacts:
- new code for creating the Mesa GLframebuffer
- new span/pixel read/write code
Some drivers not yet updated/tested.
Brian Paul [Wed, 4 May 2005 20:05:21 +0000 (20:05 +0000)]
html fix
Brian Paul [Wed, 4 May 2005 20:00:15 +0000 (20:00 +0000)]
EGLScreenMESA updates
Brian Paul [Wed, 4 May 2005 14:50:45 +0000 (14:50 +0000)]
don't call XTranslateCoordinates() if rendering to pixmap
Keith Whitwell [Wed, 4 May 2005 11:44:44 +0000 (11:44 +0000)]
Add a facility to route all rasterization through a fragment program
which is automatically generated to match the current texture environment
state. Introduces a new value ctx->FragmentProgram._Active which is
true when either _Enabled is true or there is such a fragment program
ready to run.
To test out on a driver running the software rasterizer, set
MESA_TEX_PROG=t in the environment. It goes without saying that performance
is lower for the software rasterizer in this mode.
Keith Whitwell [Wed, 4 May 2005 11:21:46 +0000 (11:21 +0000)]
Sort instructions for clarity.
Keith Whitwell [Wed, 4 May 2005 11:21:25 +0000 (11:21 +0000)]
Include negate information in disassembly output.
Keith Whitwell [Wed, 4 May 2005 11:20:59 +0000 (11:20 +0000)]
Mark unused registers as undefined so dissassemblers can recognized
them easily.
Keith Whitwell [Wed, 4 May 2005 11:20:20 +0000 (11:20 +0000)]
Improve the quality of the disassembly output for fragment programs.
Jouk Jansen [Wed, 4 May 2005 06:33:58 +0000 (06:33 +0000)]
Committing in .
OpenVMS compile update
Modified Files:
Mesa/src/mesa/shader/slang/descrip.mms
----------------------------------------------------------------------
Jon Smirl [Wed, 4 May 2005 06:12:22 +0000 (06:12 +0000)]
DRI fbdev driver getting better. Now I can see something on the screen
but it isn't the right something yet.
Brian Paul [Wed, 4 May 2005 03:33:21 +0000 (03:33 +0000)]
sync with latest EGL_MESA_screen_surface spec (EGLScreenMESA handles)
Brian Paul [Wed, 4 May 2005 03:32:39 +0000 (03:32 +0000)]
updates for latest EGLScreenMESA changes
Keith Whitwell [Tue, 3 May 2005 15:11:23 +0000 (15:11 +0000)]
Import all texture coordinate arrays. Import them only once.
Aapo Tahkola [Sun, 1 May 2005 23:36:43 +0000 (23:36 +0000)]
Add support to external tnl switcher and disable hw tnl by default.
Felix Kuehling [Sun, 1 May 2005 14:37:23 +0000 (14:37 +0000)]
Bumped the Savage driver date (a bit late) to mark S3TC support.
Felix Kuehling [Sun, 1 May 2005 14:29:16 +0000 (14:29 +0000)]
Fixed and cleaned up programming of watermark registers. There may be
a marginal speedup, but I'm not sure this has the same effect on all
hardware. Tested on Savage IX and ProSavageDDR. For experimenting with
different values see the macros at the start of savagestate.c.