mesa.git
14 years agonvfx: add NOUVEAU_SWTNL to force swtnl at runtime
Luca Barbieri [Mon, 22 Feb 2010 10:44:31 +0000 (11:44 +0100)]
nvfx: add NOUVEAU_SWTNL to force swtnl at runtime

Replace the FORCE_SWTNL macro with a NOUVEAU_SWTNL environment variable.

14 years agonvfx: move nv04_surface_2d.c into nvfx directory
Luca Barbieri [Mon, 22 Feb 2010 19:14:32 +0000 (20:14 +0100)]
nvfx: move nv04_surface_2d.c into nvfx directory

It is only used on pre-nv50 and nvfx is the only Gallium pre-nv50 driver.

14 years agonvfx: fix quads drawing
Luca Barbieri [Mon, 22 Feb 2010 13:13:17 +0000 (14:13 +0100)]
nvfx: fix quads drawing

The primitive splitting code is totally broken and will be rewritten.

Fix the most important bug now though.

14 years agonvfx: fix viewport state after bypass removal for swtnl
Luca Barbieri [Tue, 2 Mar 2010 21:51:39 +0000 (22:51 +0100)]
nvfx: fix viewport state after bypass removal for swtnl

The adjustment of nv30/nv40 after the removal of bypass incorrectly
removed the hardware viewport bypass code, which we still need for
swtnl and also forgot to remove NVFX_NEW_RAST from pipe.

14 years agonv30, nv40: move last files to nvfx/ and rm -rf nv30 nv40
Luca Barbieri [Sun, 21 Feb 2010 14:13:12 +0000 (15:13 +0100)]
nv30, nv40: move last files to nvfx/ and rm -rf nv30 nv40

This is the last nvfx unification patch.

nv[34]0_fragtex.c are moved to the common directory
nv[34]0_shader.h are renamed to nv[34]0_vertprog.h and moved to
the common directory

The separate nv30 and nv40 directories are removed from the build
system

14 years agonv30, nv40: partially non-trivially unify sampler state in nv[34]0_state.c
Luca Barbieri [Wed, 24 Feb 2010 14:08:48 +0000 (15:08 +0100)]
nv30, nv40: partially non-trivially unify sampler state in nv[34]0_state.c

Many things, like texture wrap modes and min/mag filters are common.

Some others, like annisotropy and lod settings, are not.

14 years agonv30, nv40: partially non-trivially unify nv[34]0_fragtex.c
Luca Barbieri [Sun, 21 Feb 2010 14:07:17 +0000 (15:07 +0100)]
nv30, nv40: partially non-trivially unify nv[34]0_fragtex.c

The bulk files cannot be unified, but the frontend can and allows to
share some code and simplify state_emit.c

14 years agonv30, nv40: unify nv[34]0_context.c
Luca Barbieri [Sun, 21 Feb 2010 13:40:32 +0000 (14:40 +0100)]
nv30, nv40: unify nv[34]0_context.c

They are now almost identical, except for nv30 vs nv40 fragtex
initialization.

14 years agonv30, nv40: fully unify nv[34]0_context.h
Luca Barbieri [Sun, 21 Feb 2010 13:38:04 +0000 (14:38 +0100)]
nv30, nv40: fully unify nv[34]0_context.h

Move the remaining content to the common header.

14 years agonv30, nv40: non-trivially unify nv[34]0_screen.c
Luca Barbieri [Sun, 21 Feb 2010 13:31:27 +0000 (14:31 +0100)]
nv30, nv40: non-trivially unify nv[34]0_screen.c

The files have the same structure but are substantially different.

They are unified with appropriate conditionals.

14 years agonv30, nv40: non-trivially unify nv[34]0_vertprog.c
Luca Barbieri [Sun, 21 Feb 2010 12:33:48 +0000 (13:33 +0100)]
nv30, nv40: non-trivially unify nv[34]0_vertprog.c

vertprog.c is similar but has substantial differences:
1. nv40 supports clip planes
2. nv40 uses a more advanced register allocator
3. Some register setup is different
4. Constants with the same name have different values

This patch unifies the two files.
nv30 gains clip plane support and the nv40 register allocator.

A new NVFX_VP(x) macro is introduced that at runtime resolved to
either the nv30 or the nv40 constant value.

nv30 clip planes are not tested and might not work

14 years agonv30, nv40: partially unify nv[34]0_state.c
Luca Barbieri [Sun, 21 Feb 2010 11:39:21 +0000 (12:39 +0100)]
nv30, nv40: partially unify nv[34]0_state.c

state.c is identical except for:
1. Sampler state creation is different
2. nv40 swtnl support
3. Separate blend equations on nv40

This patch unifies nv[34]0_state.c, except the sampler state creation code.

14 years agonv30, nv40: unify nv[34]0_vbo.c
Luca Barbieri [Sun, 21 Feb 2010 10:12:08 +0000 (11:12 +0100)]
nv30, nv40: unify nv[34]0_vbo.c

The files are identical, except for swtnl support which is commented
out on nv30 and restart being initialized on nv30 to avoid a compiler
warning.

14 years agonv30, nv40: non-trivially unify nv[34]0_draw.c
Luca Barbieri [Sun, 21 Feb 2010 10:17:55 +0000 (11:17 +0100)]
nv30, nv40: non-trivially unify nv[34]0_draw.c

nv30_draw.c is a stub.

This patch makes both nv30 and nv40 use the nv40 swtnl path.

Note that this doesn't actually work on nv30 because the vertex program is
encoded in the nv40-only layout.

However, swtnl was unimplemented before on nv30, so this is not a regression.
Furthermore, a patch to fix this is available near the end of the patchset.

14 years agonv30, nv40: non-trivially unify nv[34]0_fragprog.c
Luca Barbieri [Sun, 21 Feb 2010 09:55:41 +0000 (10:55 +0100)]
nv30, nv40: non-trivially unify nv[34]0_fragprog.c

The files are mostly the same except:
1. On NV40, some TGSI instructions are emulated with several hardware ones
2. Some instructions such as DDX/DDY, and STR were missing from nv30
3. NV40 has more sophisticated register management

nv30 now supports all instructions and uses the nv40 register management.

14 years agonv30, nv40: non-trivially partially unify nv[34]0_shader.h
Luca Barbieri [Sun, 21 Feb 2010 01:26:25 +0000 (02:26 +0100)]
nv30, nv40: non-trivially partially unify nv[34]0_shader.h

shader.h is similar, except for the following differences:
1. The instruction sets are not exactly the same, but mostly similar
2. Vertex program fields are in different bit positions

This patch unifies all parts of nv[34]0_shader.h except the vertex
program fields.

Vertex opcodes are also changed so that the constant names includes
SCA if it is a scalar opcode and VEC if it is a vector opcode.

14 years agonv30, nv40: non-trivially unify nv[34]0_state_fb.c
Luca Barbieri [Sat, 20 Feb 2010 23:59:30 +0000 (00:59 +0100)]
nv30, nv40: non-trivially unify nv[34]0_state_fb.c

The files are significantly different due to:
1. nv30 support 2 render targets, nv40 4
2. z-buffer pitch is set differently
3. nv30 has a limitation of colour_bits >= zeta_bits. This may not
   actually exist in the driver though
4. nv30 points color0 at depth in the depth-only case
5. nv30 sets NV34TCL_VIEWPORT_TX_ORIGIN to 0. This is probably
   unnecessary

This patch attempts to unify the two files and preserve the existing
behavior.

14 years agonv30, nv40: unify nv[34]0_state_viewport.c
Luca Barbieri [Sat, 20 Feb 2010 22:30:59 +0000 (23:30 +0100)]
nv30, nv40: unify nv[34]0_state_viewport.c

The files are identical, except for an extra comment in nv30.

14 years agonv30, nv40: unify nv[34]0_query.c
Luca Barbieri [Sat, 20 Feb 2010 18:39:24 +0000 (19:39 +0100)]
nv30, nv40: unify nv[34]0_query.c

The files are identical except formatting.

14 years agonv30, nv40: unify nv[34]0_miptree.c
Luca Barbieri [Sat, 20 Feb 2010 18:32:29 +0000 (19:32 +0100)]
nv30, nv40: unify nv[34]0_miptree.c

The only difference between nv30 and nv40 is that nv30 allowed swizzling
for more texture types.

This patch preserves the existing behavior, using conditional code.

Note however that this does not make sense, since all texture types can
be swizzled on nv40 and probably on nv30 too.

However, the handling of swizzled surfaces in the current 2D code is
partially broken, so it's best not to touch this.

A whole rewrite of the 2D code will be submitted, which will solve this
problem.

14 years agonv30, nv40: unify nv[34]0_state_stipple.c
Luca Barbieri [Sat, 20 Feb 2010 18:22:57 +0000 (19:22 +0100)]
nv30, nv40: unify nv[34]0_state_stipple.c

The files are identical, except for the fact that the nv40 version
forgets to unreference the stateobj.

Unified to the correct nv30 version.

14 years agonv30, nv40: unify identical nv[34]0_state_zsa.c
Luca Barbieri [Sat, 20 Feb 2010 22:19:40 +0000 (23:19 +0100)]
nv30, nv40: unify identical nv[34]0_state_zsa.c

14 years agonv30, nv40: unify identical nv[34]0_state_scissor.c
Luca Barbieri [Sat, 20 Feb 2010 22:17:41 +0000 (23:17 +0100)]
nv30, nv40: unify identical nv[34]0_state_scissor.c

14 years agonv30, nv40: unify identical nv[34]0_state_rasterizer.c
Luca Barbieri [Sat, 20 Feb 2010 22:16:01 +0000 (23:16 +0100)]
nv30, nv40: unify identical nv[34]0_state_rasterizer.c

14 years agonv30, nv40: unify identical nv[34]0_state_blend.c
Luca Barbieri [Sat, 20 Feb 2010 19:37:17 +0000 (20:37 +0100)]
nv30, nv40: unify identical nv[34]0_state_blend.c

14 years agonv30, nv40: non-trivially unify nv[34]0_state_emit.c
Luca Barbieri [Sun, 21 Feb 2010 10:33:15 +0000 (11:33 +0100)]
nv30, nv40: non-trivially unify nv[34]0_state_emit.c

The files are the same except for swtnl support on nv40 and for
texture cache flushing on nv40.

Unify them, and use a macro to define 4 versions of render_states,
for all combinations of nvfx and hwtnl/swtnl.

14 years agonv30, nv40: unify identical nv[34]0_transfer.c
Luca Barbieri [Sat, 20 Feb 2010 19:07:10 +0000 (20:07 +0100)]
nv30, nv40: unify identical nv[34]0_transfer.c

14 years agonv30, nv40: unify identical nv[34]0_clear.c
Luca Barbieri [Sat, 20 Feb 2010 19:04:54 +0000 (20:04 +0100)]
nv30, nv40: unify identical nv[34]0_clear.c

14 years agonvfx: add nvfx directory to build system
Luca Barbieri [Sat, 20 Feb 2010 18:52:22 +0000 (19:52 +0100)]
nvfx: add nvfx directory to build system

Will be used to hold source files unified between nv30 and nv40.

Eventually all nv30 and nv40 code will be moved there and the
nv30 and nv40 directories will be removed.

14 years agonv30, nv40: add is_nv4x member to context and screen structs
Luca Barbieri [Sun, 21 Feb 2010 12:40:49 +0000 (13:40 +0100)]
nv30, nv40: add is_nv4x member to context and screen structs

This will make it faster to check for nv40.

14 years agonv30, nv40: unify all structures and headers, except shaders
Luca Barbieri [Sat, 20 Feb 2010 17:37:43 +0000 (18:37 +0100)]
nv30, nv40: unify all structures and headers, except shaders

This patch unifies nv[34]0_screen.h, nv[34]0_context.h and
nv[34]0_state.h

The unified files are put in a new "nvfx" directory.

nv30_context.h and nv40_context.h still exist to hold the function
prototypes and include nvfx_context.h

nv[34]0_screen.h and nv[34]0_state.h are deleted, replaced by the
unified versions.

nv40 includes some extra fields for swtnl and user clip planes
support.

These fields will be unused on nv30 until that functionality gets
added to it too (by unification with nv40).

14 years agonouveau: s/rankine/eng3d/g; s/curie/eng3d/g
Luca Barbieri [Sat, 20 Feb 2010 17:34:00 +0000 (18:34 +0100)]
nouveau: s/rankine/eng3d/g; s/curie/eng3d/g

Result of running:
perl -i -p -e 's/rankine/eng3d/g; s/curie/eng3d/g;' nv[34]0/*.[ch]

This will allow to more easily unify nv30 and nv40.

14 years agonv30: remove unused on_hw field and constant fp_reg_control field
Luca Barbieri [Sat, 20 Feb 2010 17:18:48 +0000 (18:18 +0100)]
nv30: remove unused on_hw field and constant fp_reg_control field

This makes nv30_state.h equivalent to nv40_state.h

14 years agonv40: use NV34TCL_ constants where available
Luca Barbieri [Sat, 20 Feb 2010 16:43:31 +0000 (17:43 +0100)]
nv40: use NV34TCL_ constants where available

It was decided to just use the NV34TCL_ constants for constants
common between nv30 and nv40, and deprecate the NV40TCL_ versions.

This patch changes the nv40 driver to use NV34TCL_ constants for
common functionality.

This reduces differences between nv30 and nv40 to ease further
unification.

14 years agor300g: Remove unnecessary headers.
Vinson Lee [Sun, 14 Mar 2010 23:43:01 +0000 (16:43 -0700)]
r300g: Remove unnecessary headers.

14 years agodri/common: mv __driUtilMessage to utils.c
George Sapountzis [Sun, 14 Mar 2010 09:36:47 +0000 (11:36 +0200)]
dri/common: mv __driUtilMessage to utils.c

allows to link with xmlconfig without dri_util, and has nothing drm-specific.

14 years agodri/swrast: add comment in case it's not clear
George Sapountzis [Sun, 14 Mar 2010 09:36:46 +0000 (11:36 +0200)]
dri/swrast: add comment in case it's not clear

14 years agodri/swrast: drop mtypes.h from dri_sw
George Sapountzis [Sun, 14 Mar 2010 09:36:46 +0000 (11:36 +0200)]
dri/swrast: drop mtypes.h from dri_sw

14 years agodri/swrast: update copyright email
George Sapountzis [Sun, 14 Mar 2010 09:36:46 +0000 (11:36 +0200)]
dri/swrast: update copyright email

14 years agodri/swrast: port to dri_sw (drawable)
George Sapountzis [Sun, 14 Mar 2010 09:36:46 +0000 (11:36 +0200)]
dri/swrast: port to dri_sw (drawable)

14 years agodri/swrast: port to dri_sw (context)
George Sapountzis [Sun, 14 Mar 2010 09:36:45 +0000 (11:36 +0200)]
dri/swrast: port to dri_sw (context)

14 years agodri/swrast: port to dri_sw
George Sapountzis [Sun, 14 Mar 2010 09:36:45 +0000 (11:36 +0200)]
dri/swrast: port to dri_sw

14 years agodri: add dri_sw.c helper
George Sapountzis [Sun, 14 Mar 2010 09:36:45 +0000 (11:36 +0200)]
dri: add dri_sw.c helper

This is dri_util.c stripped from the drm-specific bits and will be used for
both classic and gallium swrast_dri.so

14 years agotrace: Remove unnecessary headers.
Vinson Lee [Sun, 14 Mar 2010 22:47:05 +0000 (15:47 -0700)]
trace: Remove unnecessary headers.

14 years agosvga: Remove unnecessary header.
Vinson Lee [Sun, 14 Mar 2010 22:38:04 +0000 (15:38 -0700)]
svga: Remove unnecessary header.

14 years agoglapi: exec_malloc for dispatch stubs
George Sapountzis [Thu, 11 Mar 2010 16:04:03 +0000 (18:04 +0200)]
glapi: exec_malloc for dispatch stubs

14 years agoglapi: this one should be by THREAD
George Sapountzis [Thu, 11 Mar 2010 16:04:03 +0000 (18:04 +0200)]
glapi: this one should be by THREAD

14 years agoglapi: more organize code by arch
George Sapountzis [Thu, 11 Mar 2010 16:04:03 +0000 (18:04 +0200)]
glapi: more organize code by arch

It seems that x86-64 with tls will fail to compile or load due to a missining
gl_dispatch_functions_start symbol. Not changing though, since this is how it
used to be and cannot test.

14 years agoglapi: organize arch-specific code by arch (2)
George Sapountzis [Thu, 11 Mar 2010 16:04:02 +0000 (18:04 +0200)]
glapi: organize arch-specific code by arch (2)

14 years agoglapi: organize arch-specific code by arch
George Sapountzis [Thu, 11 Mar 2010 16:04:02 +0000 (18:04 +0200)]
glapi: organize arch-specific code by arch

14 years agor300g: remove pipe_context from r300_screen
Marek Olšák [Sun, 14 Mar 2010 13:32:50 +0000 (14:32 +0100)]
r300g: remove pipe_context from r300_screen

14 years agosoftpipe: Remove unnecessary header.
Vinson Lee [Sun, 14 Mar 2010 02:05:39 +0000 (18:05 -0800)]
softpipe: Remove unnecessary header.

14 years agogallium: Respect user's CFLAGS for including X headers
Dan Nicholson [Sat, 13 Mar 2010 19:56:59 +0000 (11:56 -0800)]
gallium: Respect user's CFLAGS for including X headers

This can break on systems that don't have a system X installation.

Signed-off-by: Dan Nicholson <dbn.lists@gmail.com>
14 years agoUse X_LIBS from pkg-config, instead of libdir, for locating libX11
Jeff Smith [Sat, 13 Mar 2010 00:55:09 +0000 (18:55 -0600)]
Use X_LIBS from pkg-config, instead of libdir, for locating libX11

Signed-off-by: Jeff Smith <whydoubt@yahoo.com>
Signed-off-by: Dan Nicholson <dbn.lists@gmail.com>
14 years agollvmpipe: Switch to PIPE_TEX_MIPFILTER_NONE when texture has no mipmaps.
José Fonseca [Sat, 13 Mar 2010 19:11:08 +0000 (19:11 +0000)]
llvmpipe: Switch to PIPE_TEX_MIPFILTER_NONE when texture has no mipmaps.

14 years agor300g: add high quality anisotropic filtering for R5xx (disabled by default)
Marek Olšák [Sat, 13 Mar 2010 16:38:43 +0000 (17:38 +0100)]
r300g: add high quality anisotropic filtering for R5xx (disabled by default)

Oh look, an undocumented feature. It's a nice tool for benchmarking
texturing.

14 years agor300g: fix anisotropic filtering, fix macrotiling
Marek Olšák [Sat, 13 Mar 2010 15:24:50 +0000 (16:24 +0100)]
r300g: fix anisotropic filtering, fix macrotiling

Two bug fixes at the same time. :)

14 years agor300g: refrain from using immediate mode if it causes slowdown
Marek Olšák [Sat, 13 Mar 2010 05:07:33 +0000 (06:07 +0100)]
r300g: refrain from using immediate mode if it causes slowdown

E.g. when mapping buffers could flush CS or cause waiting
for a busy buffer.

The side effect of this is it also fixes progs/demos/arbocclude however
a separate fix should be proposed to address this issue in other cases
it might occur.

14 years agor300g: fix up function names
Marek Olšák [Sat, 13 Mar 2010 03:43:46 +0000 (04:43 +0100)]
r300g: fix up function names

14 years agoradeon: add some debuging info for glCopyTex(Sub)Image
Maciej Cencora [Sat, 13 Mar 2010 16:33:48 +0000 (17:33 +0100)]
radeon: add some debuging info for glCopyTex(Sub)Image

14 years agor300: blits for small dst pitch work just fine
Maciej Cencora [Sat, 13 Mar 2010 16:28:33 +0000 (17:28 +0100)]
r300: blits for small dst pitch work just fine

14 years agoradeon: fix gl format to mesa format mapping and add GL_BGRA formats
Maciej Cencora [Sat, 13 Mar 2010 16:27:51 +0000 (17:27 +0100)]
radeon: fix gl format to mesa format mapping and add GL_BGRA formats

14 years agoradeon: fix glCopyTex(Sub)Image
Maciej Cencora [Sat, 13 Mar 2010 15:46:19 +0000 (16:46 +0100)]
radeon: fix glCopyTex(Sub)Image

Fallback to swrast for software renderbuffers

14 years agollvmpipe: Ensure the context is flushed before modifying textures.
José Fonseca [Sat, 13 Mar 2010 16:13:26 +0000 (16:13 +0000)]
llvmpipe: Ensure the context is flushed before modifying textures.

14 years agollvmpipe: Don't use texture transfer internally.
José Fonseca [Sat, 13 Mar 2010 16:04:06 +0000 (16:04 +0000)]
llvmpipe: Don't use texture transfer internally.

Now that transfers are context objects their sideeffects must happen in
order when used by the state tracker, but that synchronization must be
bypassed when used inside the driver, or it would cause infinite
recursion.

14 years agollvmpipe: Obey rasterization rules.
José Fonseca [Sat, 13 Mar 2010 11:22:39 +0000 (11:22 +0000)]
llvmpipe: Obey rasterization rules.

Replicates softpipe.

14 years agollvmpipe: setup_context -> lp_setup_context
José Fonseca [Sat, 13 Mar 2010 10:45:52 +0000 (10:45 +0000)]
llvmpipe: setup_context -> lp_setup_context

Otherwise IDEs and debuggers have trouble distinguishing from softpipe's
setup_context.

14 years agonv50: fixup after gallium-context-transfers merge
Xavier Chantry [Sat, 13 Mar 2010 10:33:27 +0000 (11:33 +0100)]
nv50: fixup after gallium-context-transfers merge

14 years agopython: Temporarily disable python state tracker until transfers are done by contexts
José Fonseca [Fri, 12 Mar 2010 13:12:15 +0000 (13:12 +0000)]
python: Temporarily disable python state tracker until transfers are done by contexts

It requires changes not ony to the state tracker but to the python
scripts too.

14 years agost/egl: Cache the pipe surface used in flush_frontbuffer.
Chia-I Wu [Fri, 12 Mar 2010 16:48:23 +0000 (00:48 +0800)]
st/egl: Cache the pipe surface used in flush_frontbuffer.

It is very likely that the same surface will be flushed again and again.
Caching the surface should reduce the overhead of surface creation.

14 years agomklib: Fix amd64 builds on Solaris when using Sun compilers
Alan Coopersmith [Sat, 13 Mar 2010 03:03:41 +0000 (19:03 -0800)]
mklib: Fix amd64 builds on Solaris when using Sun compilers

Signed-off-by: Alan Coopersmith <alan.coopersmith@sun.com>
14 years agor300g: Forcibly sanitize max index.
Corbin Simpson [Sat, 13 Mar 2010 02:13:18 +0000 (18:13 -0800)]
r300g: Forcibly sanitize max index.

Not perfect, but the kernel should shut up now.

14 years agoRevert "st/mesa: Always recalculate invalid index bounds."
Corbin Simpson [Sat, 13 Mar 2010 01:58:59 +0000 (17:58 -0800)]
Revert "st/mesa: Always recalculate invalid index bounds."

This reverts commit 50876ddaaff72a324ac45e255985e0f84e108594.

Per ML discussion.

14 years agogallivm: cube map sampling works now
Brian Paul [Fri, 12 Mar 2010 23:54:55 +0000 (16:54 -0700)]
gallivm: cube map sampling works now

14 years agogallivm: fix bugs in nested if/then/else codegen
Brian Paul [Fri, 12 Mar 2010 23:53:51 +0000 (16:53 -0700)]
gallivm: fix bugs in nested if/then/else codegen

Quite a bit a hair pulling was involved...

14 years agogallivm: support non-vector float in lp_build_sgn()
Brian Paul [Fri, 12 Mar 2010 21:24:03 +0000 (14:24 -0700)]
gallivm: support non-vector float in lp_build_sgn()

14 years agoi965: Clarify the roles of emit_pixel_xy(), emit_delta_xy(), emit_wpos_xy().
Eric Anholt [Fri, 12 Mar 2010 23:10:22 +0000 (15:10 -0800)]
i965: Clarify the roles of emit_pixel_xy(), emit_delta_xy(), emit_wpos_xy().

14 years agoi965: Clarify that DELTAXY always occurs for both X and Y.
Eric Anholt [Thu, 11 Mar 2010 02:48:23 +0000 (18:48 -0800)]
i965: Clarify that DELTAXY always occurs for both X and Y.

14 years agoi965: Do FS SLT, SGT, and friends using CMP, SEL instead of CMP, MOV, MOV.
Eric Anholt [Thu, 11 Mar 2010 01:38:33 +0000 (17:38 -0800)]
i965: Do FS SLT, SGT, and friends using CMP, SEL instead of CMP, MOV, MOV.

14 years agoi965: Do VS SGT, SLT, and friends using CMP, SEL instead of CMP, MOV, MOV.
Eric Anholt [Thu, 11 Mar 2010 01:35:02 +0000 (17:35 -0800)]
i965: Do VS SGT, SLT, and friends using CMP, SEL instead of CMP, MOV, MOV.

14 years agoi965: Fix up VS DP4 sequences to avoid dependency control.
Eric Anholt [Thu, 11 Mar 2010 01:16:39 +0000 (17:16 -0800)]
i965: Fix up VS DP4 sequences to avoid dependency control.

This is recommended by the B-Spec.  I wasn't able to measure any
difference in ETQW.

14 years agoi965: When doing a swizzled kill pixel, don't do redundant channel compares.
Eric Anholt [Wed, 10 Mar 2010 23:44:32 +0000 (15:44 -0800)]
i965: When doing a swizzled kill pixel, don't do redundant channel compares.

This was obvious when looking at the compiled output of ETQW's
shaders.

14 years agoi965: Use the SEL instruction to implement MIN and MAX.
Eric Anholt [Wed, 10 Mar 2010 23:32:05 +0000 (15:32 -0800)]
i965: Use the SEL instruction to implement MIN and MAX.

Saves an instruction over doing conditional moves.

14 years agosvga: Fix up for context transfers.
Michel Dänzer [Fri, 12 Mar 2010 18:52:24 +0000 (19:52 +0100)]
svga: Fix up for context transfers.

14 years agost/xorg: Remove flushes no longer necessary thanks to context transfers.
Michel Dänzer [Fri, 12 Mar 2010 18:52:24 +0000 (19:52 +0100)]
st/xorg: Remove flushes no longer necessary thanks to context transfers.

14 years agoglx/single2.c: Don't call __builtin_expect on non-gnu compilers
Alan Coopersmith [Fri, 12 Mar 2010 18:27:03 +0000 (10:27 -0800)]
glx/single2.c: Don't call __builtin_expect on non-gnu compilers

Signed-off-by: Alan Coopersmith <alan.coopersmith@sun.com>
14 years agoMerge branch '7.8'
Michel Dänzer [Fri, 12 Mar 2010 18:00:47 +0000 (19:00 +0100)]
Merge branch '7.8'

14 years agovmwgfx/dri: Fix SCons build.
Michel Dänzer [Fri, 12 Mar 2010 17:54:23 +0000 (18:54 +0100)]
vmwgfx/dri: Fix SCons build.

14 years agost/dri: Always try to set up R5G6B5 configs.
Michel Dänzer [Fri, 12 Mar 2010 17:24:34 +0000 (18:24 +0100)]
st/dri: Always try to set up R5G6B5 configs.

Allows compiz to work in depth 16.

The DRI2 getBuffersWithFormat hook is only required for 16/32 bit depth
buffers, for colour buffers the only requirement is that the format matches
the drawable depth, which we can't check here.

(cherry picked from commit c50477c255a34444720fb944c54373462ef39fb9)

14 years agor100/r200/r300/r300: only enable accelerated pixel ops with kms
Alex Deucher [Fri, 12 Mar 2010 16:16:50 +0000 (11:16 -0500)]
r100/r200/r300/r300: only enable accelerated pixel ops with kms

fixes fdo bug 27043

14 years agodri/r700: include shader/programopt.h instead of programopt.c.
Luc Verhaegen [Fri, 12 Mar 2010 07:35:22 +0000 (08:35 +0100)]
dri/r700: include shader/programopt.h instead of programopt.c.

Signed-off-by: Brian Paul <brianp@vmware.com>
14 years agoAdd programs to .gitignore in redbook
Jeff Smith [Fri, 12 Mar 2010 02:19:49 +0000 (20:19 -0600)]
Add programs to .gitignore in redbook

Signed-off-by: Jeff Smith <whydoubt@yahoo.com>
Signed-off-by: Brian Paul <brianp@vmware.com>
14 years agoAdd programs to .gitignore in xdemos
Jeff Smith [Fri, 2 Mar 2007 06:14:48 +0000 (00:14 -0600)]
Add programs to .gitignore in xdemos

Signed-off-by: Jeff Smith <whydoubt@yahoo.com>
Signed-off-by: Brian Paul <brianp@vmware.com>
14 years agoAdd -L$(libdir) for xdemos and egl so that the right libX11 is found
Jeff Smith [Tue, 9 Mar 2010 18:40:44 +0000 (12:40 -0600)]
Add -L$(libdir) for xdemos and egl so that the right libX11 is found

Signed-off-by: Jeff Smith <whydoubt@yahoo.com>
Signed-off-by: Brian Paul <brianp@vmware.com>
14 years agoGrammar and spelling fixes
Jeff Smith [Fri, 13 Jun 2008 14:50:43 +0000 (09:50 -0500)]
Grammar and spelling fixes

Signed-off-by: Jeff Smith <whydoubt@yahoo.com>
Signed-off-by: Brian Paul <brianp@vmware.com>
14 years agost/xorg: context transfers
Keith Whitwell [Fri, 12 Mar 2010 12:13:32 +0000 (12:13 +0000)]
st/xorg: context transfers

14 years agogallium: fix BGRA vertex color swizzles
Marek Olšák [Tue, 9 Mar 2010 01:55:42 +0000 (01:55 +0000)]
gallium: fix BGRA vertex color swizzles

The mapping for vertex_array_bgra:
(gl -> st -> translate)
GL_RGBA -> PIPE_FORMAT_R8G8B8A8 (RGBA) -> no swizzle (XYZW)
GL_BGRA -> PIPE_FORMAT_A8R8G8B8 (ARGB) -> ZYXW (BGRA again??)

Iẗ́'s pretty clear that PIPE_FORMAT_A8R8G8B8 here is wrong. This commit
fixes the pipe format and removes obvious workarounds in util/translate.

Tested with: softpipe, llvmpipe, r300g.

Signed-off-by: José Fonseca <jfonseca@vmware.com>
14 years agost/mesa: Always recalculate invalid index bounds.
Corbin Simpson [Fri, 12 Mar 2010 10:51:40 +0000 (02:51 -0800)]
st/mesa: Always recalculate invalid index bounds.

These should always be sanitized before heading towards the pipe driver,
and if the calling function explicitly marked them as invalid, we need
to regenerate them.

Allows r300g to properly pass a bit more of Wine's d3d9 testing without
dropping stuff on the floor.

14 years agost/mesa: Update debug infos.
Corbin Simpson [Fri, 12 Mar 2010 10:51:10 +0000 (02:51 -0800)]
st/mesa: Update debug infos.

Still commented out, of course.

14 years agor300g: Properly clamp index limits.
Corbin Simpson [Fri, 12 Mar 2010 10:47:44 +0000 (02:47 -0800)]
r300g: Properly clamp index limits.

Make the kernel even less likely to barf on us.