Keith Whitwell [Thu, 20 Aug 2009 14:46:51 +0000 (15:46 +0100)]
softpipe: fix glitch in texel lookups on fastpaths
Fixes two issues - firstly for mipmap levels with one or more
dimensions smaller than tilesize, the code was sampling off the edge
of the texture (but still within the tile).
Secondly, in the linear_mipmap_linear case, both the default code and
new fastpath were incorrect. This change fixes the fastpath and adds
a comment to the default path, which still needs to be fixed.
Basically the issue is that the coordinates in the smaller texture
level are/were being computed by just dividing thecoordinates from the
larger texture level by two, as in:
x0[j] /= 2;
y0[j] /= 2;
x1[j] /= 2;
y1[j] /= 2;
The issues with this are signficant. Initially x1 is most often equal
to x0+1, but after this, it will likely be equal to x0, so we will not
actually be performing the linear blend within the smaller mipmap.
The fastpath code avoided this (recalculated x1), but was still using
the weighting factors from the larger mipmap level (xw, yw), which
were incorrect.
Change the fastpath code to do two full, independent linear samples of
the two mipmap levels before blending. The default code needs to do
the same thing.
Keith Whitwell [Thu, 20 Aug 2009 10:25:20 +0000 (11:25 +0100)]
softpipe: optimized path for simple mipmap sampling
linear-mip-linear-repeat-POT sampling faspath, provides a very nice
speedup to apps that do this common type of texturing.
Test case: demos/terrain, turn fog off, turn texturing on.
Without patch: 12 fps
With patch: 20 fps.
Keith Whitwell [Tue, 18 Aug 2009 15:21:12 +0000 (16:21 +0100)]
softpipe: move flatshade-first check out of loop
Keith Whitwell [Tue, 11 Aug 2009 17:23:28 +0000 (18:23 +0100)]
Revert "softpipe: rearrange blend fastpaths"
This reverts commit
1295cf423e21dad04a947960782ffa8db2739709.
The original formulation was easier to understand & work with. Will
revisit this later.
Keith Whitwell [Tue, 11 Aug 2009 17:06:16 +0000 (18:06 +0100)]
softpipe: reduce textual differences between exec and sse shader paths
Unshare one function (setup_pos_vector) as we want to push this code
into the generated shader in the SSE case.
Keith Whitwell [Tue, 11 Aug 2009 17:03:01 +0000 (18:03 +0100)]
softpipe: remove gallivm fragment shaders
However we do llvm integration, it will be different & more comprehensive
than this.
Keith Whitwell [Thu, 30 Jul 2009 10:59:32 +0000 (11:59 +0100)]
softpipe: setup quad outputs from with fs->run
Keith Whitwell [Thu, 30 Jul 2009 10:35:50 +0000 (11:35 +0100)]
softpipe: rearrange blend fastpaths
Keith Whitwell [Thu, 30 Jul 2009 10:35:08 +0000 (11:35 +0100)]
softpipe: add depth-lequal z16 path
Keith Whitwell [Thu, 30 Jul 2009 10:34:36 +0000 (11:34 +0100)]
softpipe: remove unused variable in shade_quad
Keith Whitwell [Mon, 27 Jul 2009 14:51:15 +0000 (15:51 +0100)]
softpipe: fix off-by-one in nearest texcoord routines
Stray '- 0.5' copied from linear versions.
Keith Whitwell [Mon, 27 Jul 2009 11:44:58 +0000 (12:44 +0100)]
softpipe: example fast paths for simple samplers
All these fastpaths are examples of the types of things we'd code-generate
in a more sophisticated version of softpipe.
Keith Whitwell [Mon, 27 Jul 2009 11:11:16 +0000 (12:11 +0100)]
softpipe: fastpath for interpolated z16 less depthtesting
Because this is interpolated (ie. early) depth, we can build in an
assumption about the quads emitted by triangle setup, ie that they
are actually linear spans. Interpolate z over those spans in z16
format to save on math & conversion.
Keith Whitwell [Mon, 27 Jul 2009 10:23:51 +0000 (11:23 +0100)]
softpipe: cope with nr_cbufs == 0
Disable blend code when no color buffer
Keith Whitwell [Mon, 27 Jul 2009 07:17:45 +0000 (08:17 +0100)]
softpipe: move all depth/stencil/alpha pixel processing into one stage
Keith Whitwell [Sat, 25 Jul 2009 10:01:48 +0000 (11:01 +0100)]
softpipe: fix error in scissor state dependencies
Keith Whitwell [Sat, 25 Jul 2009 09:01:06 +0000 (10:01 +0100)]
softpipe: cleanup framebuffer state routine slightly
Keith Whitwell [Fri, 24 Jul 2009 19:19:18 +0000 (20:19 +0100)]
softpipe: move all color-combine code into sp_quad_blend.c
Consolidate the read-modify-write color combining code from
the blend, colormask and output stages.
Keith Whitwell [Fri, 24 Jul 2009 19:18:52 +0000 (20:18 +0100)]
softpipe: fix typo
Keith Whitwell [Fri, 24 Jul 2009 17:46:17 +0000 (18:46 +0100)]
softpipe: example fastpaths in blending
Keith Whitwell [Fri, 24 Jul 2009 17:17:05 +0000 (18:17 +0100)]
softpipe: actually pass >1 quad from triangle routine
First attempt
Keith Whitwell [Fri, 24 Jul 2009 15:49:35 +0000 (16:49 +0100)]
softpipe: expand quad pipeline to process >1 quad at a time
This is part one -- we still only pass a single quad down, but
the code can now cope with more. The quads must all be from the same
tile.
Keith Whitwell [Fri, 24 Jul 2009 15:12:48 +0000 (16:12 +0100)]
softpipe: rip out old mulithread support
Keith Whitwell [Thu, 23 Jul 2009 10:14:39 +0000 (11:14 +0100)]
softpipe: avoid flushing depth buffer cache on swapbuffers
There's no need to push out depth buffer contents on swapbuffers.
Note that this change doesn't throw away depth buffer changes, it simply
holds them in the cache over calls to swapbuffers. The hope is
that swapbuffers will be followed by a clear() which means in that case
we won't have to write the changes out.
Keith Whitwell [Wed, 22 Jul 2009 14:36:25 +0000 (15:36 +0100)]
softpipe: also shortcircuit non-texture tile lookups
Keith Whitwell [Wed, 22 Jul 2009 14:08:42 +0000 (15:08 +0100)]
softpipe: shortcircuit repeated lookups of the same tile
The sp_tile_cache is often called repeatedly to look up the same
tile. Add a cache (to the cache) of the single tile most recently
retreived and make a quick inline check to see if this matches the
subsequent request.
Add a tile_address bitfield struct to make this check easier.
Keith Whitwell [Fri, 17 Jul 2009 11:12:04 +0000 (12:12 +0100)]
softpipe: remove unused vars in sp_setup.c
Keith Whitwell [Fri, 17 Jul 2009 11:03:51 +0000 (12:03 +0100)]
softpipe: use bitwise logic to setup quad masks in sp_setup
Keith Whitwell [Fri, 17 Jul 2009 09:47:32 +0000 (10:47 +0100)]
softpipe: simplify flush_spans
No loss of performance, but simpler code.
Keith Whitwell [Fri, 17 Jul 2009 09:44:22 +0000 (10:44 +0100)]
softpipe: make some small steps to flush texture cache less frequently
No performance gain yet, but the code is a bit cleaner.
Keith Whitwell [Thu, 16 Jul 2009 16:57:00 +0000 (17:57 +0100)]
gallium/xlib: use XSHM for swapbuffers
Makes some difference, but suprisingly little. Barely worth the effort.
Keith Whitwell [Thu, 16 Jul 2009 16:51:02 +0000 (17:51 +0100)]
util: _debug_printf should print even when DEBUG is not defined
The leading underscore is meaningful... This function is used by
_warning and _error functions as well as the more common
debug_printf().
debug_printf (without underscore) gets turned off when DEBUG is
disabled, but warning/error messages still use this function to get
their message out.
Keith Whitwell [Thu, 16 Jul 2009 13:14:32 +0000 (14:14 +0100)]
softpipe: remove backwards dependency from tilecache to softpipe
The tile cache is a utility, it shouldn't know anything about the
entity which is making use of it (ie softpipe).
Remove softpipe parameter to all the tilecache function calls, and
also remove the need to keep a softpipe pointer in the sampler structs.
Keith Whitwell [Fri, 12 Jun 2009 10:59:01 +0000 (11:59 +0100)]
gallium: remove multiple aliases for TGSI opcodes
This is a source of ongoing confusion. TGSI has multiple names for
opcodes where the same semantics originate in multiple shader APIs.
For instance, TGSI includes both Mesa/GLSL and DX/SM30 names for
opcodes with the same semantics, but aliases those names to the same
underlying opcode number.
This makes it very difficult to visually inspect two sets of opcodes
(eg in state tracker & driver) and check if they implement the same
functionality.
This patch arbitarily rips out the versions of the opcodes not currently
favoured by the mesa state tracker and leaves us with a single name
for each distinct operation.
Keith Whitwell [Tue, 21 Jul 2009 23:39:00 +0000 (00:39 +0100)]
gallium: simplify tgsi_full_immediate struct
Remove the need to have a pointer in this struct by just including
the immediate data inline. Having a pointer in the struct introduces
complications like needing to alloc/free the data pointed to, uncertainty
about who owns the data, etc. There doesn't seem to be a need for it,
and it is unlikely to make much difference plus or minus to performance.
Added some asserts as we now will trip up on immediates with more
than four elements. There were actually already quite a few such asserts,
but the >4 case could be used in the future to specify indexable immediate
ranges, such as lookup tables.
Richard Li [Tue, 21 Jul 2009 21:56:06 +0000 (17:56 -0400)]
r600: fix dst reg indexing for real
This fixes segfaults in apps like teapot and tunnel
Alex Deucher [Tue, 21 Jul 2009 21:44:36 +0000 (17:44 -0400)]
Revert "r600: fix dst reg indexing"
This reverts commit
cc893d9a98255d3c26df7123ba5cc02e478c9328.
Richard has the proper fix.
Alex Deucher [Tue, 21 Jul 2009 21:07:17 +0000 (17:07 -0400)]
r600: add stencil support
Alex Deucher [Tue, 21 Jul 2009 18:06:47 +0000 (14:06 -0400)]
r600: use state functions to set default state
Alex Deucher [Tue, 21 Jul 2009 17:46:15 +0000 (13:46 -0400)]
r600: fill in point functions
Alex Deucher [Tue, 21 Jul 2009 17:28:12 +0000 (13:28 -0400)]
r600: set provoking vertex to last vertex for OGL
Alex Deucher [Tue, 21 Jul 2009 17:19:46 +0000 (13:19 -0400)]
r600: fill in r700UpdateViewportOffset
Alex Deucher [Tue, 21 Jul 2009 17:10:26 +0000 (13:10 -0400)]
r600: first pass at polyoffset support
not working yet
Jerome Glisse [Tue, 21 Jul 2009 19:12:40 +0000 (21:12 +0200)]
radeon: fix colorbuffer pitch emission regarding tiling in KMS/CS case
We need to emit a relocation for pitch register so that kernel can
check and properly setup tiling on the color buffer.
Alex Deucher [Tue, 21 Jul 2009 15:09:05 +0000 (11:09 -0400)]
R600: fix up some build problems
Alex Deucher [Tue, 21 Jul 2009 05:58:05 +0000 (01:58 -0400)]
r600: add alpha test support
Michel Dänzer [Tue, 21 Jul 2009 08:46:29 +0000 (10:46 +0200)]
Track Radeon driver symlinks in Git.
Peter Hutterer [Mon, 20 Jul 2009 06:11:26 +0000 (16:11 +1000)]
Add missing X11_INCLUDES to egl/drivers/demo and egl/main.
Compiling mesa on a system with no X headers installed in the default
include paths fails due to missing X11 includes. The header includes are
picked up by configure but not applied.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Eric Anholt [Tue, 21 Jul 2009 00:58:12 +0000 (17:58 -0700)]
i965: Don't clip everything if FRONT_AND_BACK culling while culling disabled.
Fixes everything-black with meta_clear_tris on quake4-mpdemo and doom3-demo.
Bug #18844, 22077.
Alex Deucher [Mon, 20 Jul 2009 23:33:05 +0000 (19:33 -0400)]
r600: fix typo in blend code
Kevin DeKorte [Mon, 20 Jul 2009 22:56:47 +0000 (18:56 -0400)]
r600: fix dst reg indexing
This fixes segfaults in apps like teapot and tunnel
Alex Deucher [Mon, 20 Jul 2009 22:50:59 +0000 (18:50 -0400)]
r600: add blending support
Alex Deucher [Mon, 20 Jul 2009 21:22:59 +0000 (17:22 -0400)]
r600: add user clip plane support
Alex Deucher [Mon, 20 Jul 2009 19:27:28 +0000 (15:27 -0400)]
r600: add logicop support
Keith Whitwell [Mon, 20 Jul 2009 18:45:24 +0000 (19:45 +0100)]
tgsi: get texturing working in vertex shader sse2 path
Missing file from previous commit.
Michel Dänzer [Mon, 20 Jul 2009 18:41:11 +0000 (20:41 +0200)]
Merge branch 'mesa_7_5_branch'
Michel Dänzer [Mon, 20 Jul 2009 18:26:28 +0000 (20:26 +0200)]
radeon: With DRI1, if we have HW stencil, only expose fbconfigs with stencil.
Otherwise simple apps like glxgears pick up a DirectColor visual since the X
server mixes the depth 32 visual in with the other GLX visuals, and this seems
to result in a (mostly) black screen due to a bad ColorMap for a lot of people.
The bad ColorMap may be a bug in the apps, the X server or X driver, and
regardless of that I think the X server should ideally make the depth 32 GLX
visual separate from the rest again, but in the meantime this makes us cope.
(depth_bits is either 16 or 24, never 0)
Keith Whitwell [Mon, 20 Jul 2009 18:29:22 +0000 (19:29 +0100)]
tgsi: get texturing working in vertex shader sse2 path
Keith Whitwell [Mon, 20 Jul 2009 18:28:34 +0000 (19:28 +0100)]
tgsi: fix regression in indexed const lookups
This function was calling get_input_base() and get_output_base() to
get the names of a couple of register to use as temps. Those
functions no longer return registers, so adjust it to get the
registers elsewhere.
This change doesn't address the issue that it's a fairly poor way to
grab a register name by calling a function with an apparently
unrelated meaning.
Kevin DeKorte [Mon, 20 Jul 2009 13:27:17 +0000 (09:27 -0400)]
r600: Fix compilation
Cooper Yuan [Mon, 20 Jul 2009 09:42:47 +0000 (17:42 +0800)]
R6xx/r7xx: Fix line stipple and width issue
Michel Dänzer [Mon, 20 Jul 2009 00:11:17 +0000 (02:11 +0200)]
Add common_ppc.c to MESA_GALLIUM_SOURCES.
Michel Dänzer [Mon, 20 Jul 2009 00:06:37 +0000 (02:06 +0200)]
r300g: Fix SCons build.
Michel Dänzer [Mon, 20 Jul 2009 00:05:38 +0000 (02:05 +0200)]
gallium/trace: Fix SCons build.
Michel Dänzer [Sun, 19 Jul 2009 23:53:15 +0000 (01:53 +0200)]
r300g: Guard R500 register writes by is_r500 check.
Flagged by the DRM command stream checker. This allows the driver to work on
non-R500 cards.
Dave Airlie [Sat, 18 Jul 2009 07:44:44 +0000 (17:44 +1000)]
gallium: make g3dvl build again
Brian Paul [Sat, 18 Jul 2009 04:00:47 +0000 (22:00 -0600)]
Merge branch 'mesa_7_5_branch'
Conflicts:
Makefile
progs/glsl/multitex.c
src/mesa/main/enums.c
src/mesa/main/state.c
src/mesa/main/texenvprogram.c
src/mesa/main/version.h
Alex Deucher [Fri, 17 Jul 2009 23:04:19 +0000 (19:04 -0400)]
R6xx/R7xx: add fine grained syncing support
Alex Deucher [Fri, 17 Jul 2009 21:08:40 +0000 (17:08 -0400)]
R6xx/r7xx: send depth state in it's own function
Andrew Randrianasulu [Fri, 17 Jul 2009 22:55:12 +0000 (00:55 +0200)]
st/egl: Fix broken build after EGL thread changes
Alex Deucher [Fri, 17 Jul 2009 22:10:48 +0000 (18:10 -0400)]
R200: fix build when RADEON_DEBUG_BO is set
Dave Airlie [Fri, 17 Jul 2009 22:00:23 +0000 (08:00 +1000)]
radeon: disable BO debug
Brian Paul [Fri, 17 Jul 2009 20:43:29 +0000 (14:43 -0600)]
progs/util: remove extfuncs.h (we use GLEW instead)
Brian Paul [Fri, 17 Jul 2009 19:36:06 +0000 (13:36 -0600)]
egl: commit missing eglcurrent.[ch] files
Not sure how these got left out from earlier commit.
Brian Paul [Fri, 17 Jul 2009 19:23:11 +0000 (13:23 -0600)]
progs/glsl: finish conversion to GLEW
This is a follow-on to
b799af91d5ffbee1481161fec29eb4c92b161272.
Remove _func function suffix and unneeded #includes.
Alex Deucher [Fri, 17 Jul 2009 18:46:40 +0000 (14:46 -0400)]
Merge branch 'master' of git+ssh://agd5f@git.freedesktop.org/git/mesa/mesa into r6xx-rewrite
Kevin DeKorte [Fri, 17 Jul 2009 18:43:42 +0000 (14:43 -0400)]
R6xx/r7xx: warning fixes
patch from Kevin DeKorte with some minor fixes from me.
Brian Paul [Fri, 17 Jul 2009 18:40:48 +0000 (12:40 -0600)]
mesa: set version to 7.5
Brian Paul [Fri, 17 Jul 2009 18:40:44 +0000 (12:40 -0600)]
mesa: set version to 7.5
Brian Paul [Fri, 17 Jul 2009 18:40:24 +0000 (12:40 -0600)]
docs: set date for 7.5 release
Chia-I Wu [Fri, 17 Jul 2009 04:21:59 +0000 (21:21 -0700)]
xeglthreads: Check current context for EGL per-thread support.
Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
Chia-I Wu [Fri, 17 Jul 2009 17:56:00 +0000 (11:56 -0600)]
egl: Remove redundant DeletePending flag.
A context or surface that is neither linked to a display nor current to
a thread should be destroyed. Therefore, an unlinked context or surface
implies a pending delete automatically.
Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
Chia-I Wu [Fri, 17 Jul 2009 04:21:57 +0000 (21:21 -0700)]
egl: Destroy display's resources upon termination.
eglTerminate should destroy the contexts and surfaces of the display.
Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
Chia-I Wu [Fri, 17 Jul 2009 04:21:56 +0000 (21:21 -0700)]
egl: Return the same EGL Display for the same native display.
The latest revision of the spec explicitly requires the same handle to
be returned for the same native display.
Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
Chia-I Wu [Fri, 17 Jul 2009 17:53:03 +0000 (11:53 -0600)]
egl: Use the link functions to manage resources.
This commit uses the newly introduced link functions to manage EGL
contexts and surfaces. As a result of this, the API for drivers are
changed. All drivers are updated for the change.
Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
Chia-I Wu [Fri, 17 Jul 2009 17:48:27 +0000 (11:48 -0600)]
egl: Add funtions to link contexts and surfaces to displays.
EGL contexts and surfaces are resources of displays. They should be
managed by displays. This commit adds a bunch of functions to
egldisplay.c to help establish the links between contexts/surfaces and
displays. How links are established is considered opaque outside
display. Functions like _eglGetSurfaceHandle or _eglLookupSurface are
therefore moved to egldisplay.c, with some small modifications.
The idea is also extended to display. That is, displays need to link to
themselves to be looked up.
This commit only adds the functions. A commit to use them should
follow.
Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
Chia-I Wu [Fri, 17 Jul 2009 17:42:04 +0000 (11:42 -0600)]
egl: Extend per-thread info for multiple current contexts.
EGL allows multiple current contexts, as long as they are bound to
different client APIs.
Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
Chia-I Wu [Fri, 17 Jul 2009 17:41:02 +0000 (11:41 -0600)]
egl: Support per-thread info.
This commit introduces a "current" system to manage per-thread info. It
uses TLS, if GLX_USE_TLS is defined, or pthread, if PTHREADS is defined.
If none of them are defined, it uses a dummy implementation that is just
like before.
Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
Chia-I Wu [Fri, 17 Jul 2009 04:21:51 +0000 (21:21 -0700)]
egl: Add eglcompiler.h for compiler features.
Only INLINE (from mesa/main/compiler.h) is defined now. It may be used
to deal with symbol visibility and int/pointer conversion in the future.
Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
Signed-off-by: Brian Paul <brianp@vmware.com>
Alex Deucher [Fri, 17 Jul 2009 16:21:33 +0000 (12:21 -0400)]
R6xx/r7xx: disable depth/stencil compression for now
Alex Deucher [Fri, 17 Jul 2009 16:07:15 +0000 (12:07 -0400)]
R6xx/r7xx: add some missing state regs
Brian Paul [Fri, 17 Jul 2009 15:40:45 +0000 (09:40 -0600)]
docs: news entry for 7.5 release
Brian Paul [Fri, 17 Jul 2009 15:40:28 +0000 (09:40 -0600)]
docs: update links to Gallium wiki page on freedesktop.org
Eric Anholt [Fri, 17 Jul 2009 01:41:03 +0000 (18:41 -0700)]
texenv: Calculate whether we need to do secondary color on our own.
The _TriangleCaps bit is deprecated, not updated when we require, and
is set based on state that hasn't been updated at that point in
_mesa_update_state_locked().
Fixes incorrect clear color in glsl/twoside.c with meta_clear_tris.
Eric Anholt [Thu, 16 Jul 2009 22:57:22 +0000 (15:57 -0700)]
i965: Add missing state dependency of sf_unit on _NEW_BUFFERS.
Alex Deucher [Thu, 16 Jul 2009 21:35:44 +0000 (17:35 -0400)]
R6xx: fix rendering on r6xx/rs780 chips
You always need to emit a fetch shader (fs) even if you
aren't using it. For now, just emit the fs with the
vs address to make the kernel happy.
José Fonseca [Thu, 16 Jul 2009 18:34:44 +0000 (19:34 +0100)]
python/retrace: Process the call no passed to --to option inclusively.
José Fonseca [Thu, 16 Jul 2009 18:32:40 +0000 (19:32 +0100)]
python/retrace: Dump the surface copy contents.
José Fonseca [Thu, 16 Jul 2009 18:31:36 +0000 (19:31 +0100)]
python/retrace: Flush stdout before calling the pipe driver.
So that messages are in sync with stderr.
José Fonseca [Thu, 16 Jul 2009 18:20:25 +0000 (19:20 +0100)]
mesa: Fix logbase2.
It was providing 1 too many for non power two values.