OSDN Git Service
Eric Anholt [Thu, 10 Dec 2009 17:16:37 +0000 (09:16 -0800)]
intel: Attempt to fix up after "Update vertex texture code."
The MaxCombinedTextureImageUnits is the total number of samplers that can
be bound between vertex, geometry, and fragment, not 0. This should report
the correct value on 965 now. Other DRI drivers may also need updating if
their MaxVertexTextureImageUnits != 0 (for example, if using the sw vertex
pipeline).
It's not clear to me if there's going to be a valid value for this
limit other than MaxTextureImageUnits + MaxVertexTextureImageUnits (+
MaxGeometryTextureImageUnits eventually). If not, then we should probably
just move this into the core at Get time.
Bug #25518 (wine regression). Fixes piglit vp-combined-image-units.
Roland Scheidegger [Thu, 10 Dec 2009 02:51:35 +0000 (03:51 +0100)]
winsys/intel: fix dereferencing of opaque type due to pipe_reference changes
Marek Olšák [Tue, 8 Dec 2009 23:45:18 +0000 (00:45 +0100)]
r300g: fix routing of vertex streams if TCL is bypassed
Generating mipmaps finally works, among other things. Yay!
Marek Olšák [Tue, 8 Dec 2009 20:53:19 +0000 (21:53 +0100)]
r300g: always disable unused colorbuffers
Marek Olšák [Tue, 8 Dec 2009 03:55:32 +0000 (04:55 +0100)]
r300g: make pow(0,0) return 1 instead of NaN in the R500 fragment shader
Unfortunately we can't fix this easily in the R300 fragment shader,
and it's probably not worth the effort.
Marek Olšák [Tue, 8 Dec 2009 23:55:51 +0000 (00:55 +0100)]
r300g: clean up r300_emit_aos
Richard Li [Wed, 9 Dec 2009 20:39:16 +0000 (15:39 -0500)]
r600 : add pre-compile mesa shader calling interface, in order to handle
complex built-in shader instructions.
José Fonseca [Wed, 9 Dec 2009 10:59:38 +0000 (10:59 +0000)]
util: Document the meaning of util_format_layout.
The util_format_layout name was unfortunate and there
are as been a lot of confusion due to this. Hopefully
this will shed some light on what it was meant for.
Bottom line is: do not rely on these values unless
you're automatically code generating pixel
packing/unpacking routines.
Suggestions for better names than util_format_layout
are welcome!
Roland Scheidegger [Tue, 8 Dec 2009 16:51:19 +0000 (17:51 +0100)]
nouveau: use boolean instead of bool
Roland Scheidegger [Tue, 8 Dec 2009 16:44:51 +0000 (17:44 +0100)]
cell: use boolean instead of bool
Roland Scheidegger [Tue, 8 Dec 2009 16:42:49 +0000 (17:42 +0100)]
gallium: use boolean instead of bool in p_refcnt.h
all code in gallium should use boolean not bool
Roland Scheidegger [Tue, 8 Dec 2009 18:26:20 +0000 (19:26 +0100)]
Merge branch 'gallium-strict-aliasing'
Roland Scheidegger [Tue, 8 Dec 2009 18:13:48 +0000 (19:13 +0100)]
gallium: fix more potential strict aliasing issues
In particular, gcc man page warns that
union a_union {
int i;
double d;
};
int f() {
double d = 3.0;
return ((union a_union *) &d)->i;
}
"might" not be ok (why not?), even though it doesn't seem to generate
any warnings. Hence don't use this and do the extra step to actually use
assignment to get the values in/out of the union.
This changes parts of
3456f9149b3009fcfce80054759d05883d3c4ee5.
Brian Paul [Mon, 7 Dec 2009 21:12:28 +0000 (14:12 -0700)]
tgsi: fix some off-by-one errors in shader length, instruction length
The ureg and/or tgsi-simplification work introduced some inconsistencies
between the ureg and traditional TGSI construction code.
Now the tgsi_instruction::NrTokens field is consistant and the
tgsi_header::BodySize field isn't off by one.
Fixes bug 25455.
Andre Maasikas [Mon, 7 Dec 2009 15:22:03 +0000 (17:22 +0200)]
r600: and finally fix SCS
Andre Maasikas [Mon, 7 Dec 2009 14:27:05 +0000 (16:27 +0200)]
r600: remove (now) dead code
Andre Maasikas [Mon, 7 Dec 2009 14:23:07 +0000 (16:23 +0200)]
r600: fix SIN also
Andre Maasikas [Mon, 7 Dec 2009 14:09:10 +0000 (16:09 +0200)]
r600: use the new inline constants feature to fix COS
Andre Maasikas [Mon, 7 Dec 2009 13:23:40 +0000 (15:23 +0200)]
r600: add assembler support for literal(inline) constants
and use it in cubemap instruction sequence for testing
Andre Maasikas [Mon, 7 Dec 2009 11:04:32 +0000 (13:04 +0200)]
r600: merge alu_instruction/alu_instruction2
Andre Maasikas [Mon, 7 Dec 2009 09:51:36 +0000 (11:51 +0200)]
r600: add ABS support for source regs to assembler
use it in tex cube instruction sequence
Andre Maasikas [Fri, 4 Dec 2009 14:36:41 +0000 (16:36 +0200)]
r600: glsl - allow specifying texture sampler via uniforms
looks kinda hackish, should rethink later
Andre Maasikas [Fri, 4 Dec 2009 11:53:44 +0000 (13:53 +0200)]
r600: implement FRAG_ATTRIB_FACE, glsl/twoside works
Andre Maasikas [Fri, 4 Dec 2009 10:58:36 +0000 (12:58 +0200)]
r600: quick hack to get KIL_NV working - does condition TR only for now
Andre Maasikas [Fri, 4 Dec 2009 09:37:15 +0000 (11:37 +0200)]
r600: wip glsl - refactor conditional instructions a bit
remember the dst register which is used for cond updates
when it's time to use the cond codes issue a separate PRED instruction
Andre Maasikas [Thu, 3 Dec 2009 16:12:45 +0000 (18:12 +0200)]
r600: execute SET funtions on all channels
seems assemble_LOGIC was meant for non-condition-code instructions
so execute in for all components as previously
Alan Hourihane [Tue, 8 Dec 2009 13:11:09 +0000 (13:11 +0000)]
move assert to avoid crash in debug build.
Michel Dänzer [Tue, 8 Dec 2009 11:25:34 +0000 (12:25 +0100)]
Merge branch 'mesa_7_7_branch'
Michel Dänzer [Tue, 8 Dec 2009 10:45:19 +0000 (11:45 +0100)]
vmware/xorg: Fix SCons build.
Not sure how vmw_screen.c could build at all though...
Michel Dänzer [Tue, 8 Dec 2009 10:43:22 +0000 (11:43 +0100)]
vmware/xorg: Avoid warning about HAVE_STDINT_H being redefined.
Andre Maasikas [Tue, 8 Dec 2009 09:57:24 +0000 (11:57 +0200)]
Merge branch 'mesa_7_7_branch'
Conflicts:
src/mesa/drivers/dri/r600/r700_assembler.c
src/mesa/main/version.h
Andre Maasikas [Tue, 8 Dec 2009 08:16:14 +0000 (10:16 +0200)]
Merge branch 'mesa_7_6_branch' into mesa_7_7_branch
Andre Maasikas [Thu, 3 Dec 2009 10:26:44 +0000 (12:26 +0200)]
r600: add support for TXB instruction
makes testing other things easier - does not hang the card
TODO: enable TEX dependency tracking in vertex programs
Andre Maasikas [Wed, 4 Nov 2009 08:00:47 +0000 (10:00 +0200)]
r600: reorder state for render_target and blend
First time around render targets are not enabled yet (done in
r700SendRenderTargetState) so blend state is not emitted for any targets.
Affects first glClear in some mesa tests.
As a quick fix reorder state emit so that target is set first
Vinson Lee [Tue, 8 Dec 2009 03:04:07 +0000 (19:04 -0800)]
llvmpipe: Initialize variables in emit_instruction.
Vinson Lee [Tue, 8 Dec 2009 02:40:37 +0000 (18:40 -0800)]
progs/test: Initialize variable in prog_parameter.
Silences uninitialized variable compiler warning.
Ian Romanick [Tue, 8 Dec 2009 01:18:56 +0000 (17:18 -0800)]
mesa: set version string to 7.7-rc1
Also modify the Makefile to use the correct version for the tarballs.
Ian Romanick [Tue, 8 Dec 2009 01:16:10 +0000 (17:16 -0800)]
Merge branch 'mesa_7_6_branch' into mesa_7_7_branch
Conflicts:
progs/demos/projtex.c
progs/xdemos/glxinfo.c
src/mesa/main/version.h
To fix the confilicts in projtex.c and glxinfo.c I just took the code from
mesa_7_6_branch. The conflicts seem to have occured from cherry-picks from
mesa_7_7_branch to mesa_7_6_branch followed by commmits just to
mesa_7_6_branch.
Ian Romanick [Tue, 8 Dec 2009 01:06:07 +0000 (17:06 -0800)]
Revert "intel: Make RGB renderbuffers use XRGB8888 like we do for RGB system buffers."
This reverts commit
4598942b1b88a2a7d5af7febae7e79eedf00e385.
XRGB8888 doesn't work as intended. Revert this for now, and we'll revisit it
for 7.8 or something.
Ian Romanick [Tue, 8 Dec 2009 00:59:59 +0000 (16:59 -0800)]
mesa: set version string to 7.6.1-rc3
Maciej Cencora [Mon, 7 Dec 2009 21:24:41 +0000 (22:24 +0100)]
radeon: fix cases when only first image where put directly into miptree.
Make sure that minimal width, height and depth of texture image is 1.
Maciej Cencora [Mon, 7 Dec 2009 21:22:57 +0000 (22:22 +0100)]
radeon: fix image migration for small compressed textures
memcpy would give incorrect results if src rowstride != dst rowstride
Roland Scheidegger [Mon, 7 Dec 2009 20:47:49 +0000 (21:47 +0100)]
mesa: fix shader prog_execute strict aliasing violations
use unions instead of pointer casts.
Vinson Lee [Mon, 7 Dec 2009 20:31:08 +0000 (12:31 -0800)]
progs/demos: Fix memory leak in projtex.c.
Roland Scheidegger [Mon, 7 Dec 2009 19:35:42 +0000 (20:35 +0100)]
gallium/util: fix util_color_[un]pack[-ub] to be strict aliasing safe
use pointer to union instead of void pointer.
gcc complained a lot, depending what the pointer originally actually was.
Looks like it's in fact maybe legal to cast for instance uint pointers to
union pointers as long as union contains a uint type, hence use this with some
callers, other just use union util_color in the first place.
Roland Scheidegger [Mon, 7 Dec 2009 19:11:46 +0000 (20:11 +0100)]
mesa: fix strict aliasing issues in half-to-float/float-to-half conversions
use union instead of casts
Michal Krol [Mon, 7 Dec 2009 17:12:05 +0000 (18:12 +0100)]
util/tile: Support R8G8B8A8_UNORM format.
Brian Paul [Mon, 7 Dec 2009 16:05:40 +0000 (09:05 -0700)]
st/mesa: remove debug code
Brian Paul [Mon, 7 Dec 2009 16:03:39 +0000 (09:03 -0700)]
progs/glsl: updates to deriv.c test
Pass window size into the shader so scaling the window doesn't effect
the colors.
Brian Paul [Mon, 7 Dec 2009 16:00:57 +0000 (09:00 -0700)]
st/mesa: negate DDY to match GL semantics
This fixes the regression from commit
884007546c98b1779bf266ec5111b1e7e2b68b2e
Fixes bug 25456 (piglit derivs regression).
Brian Paul [Mon, 7 Dec 2009 15:59:38 +0000 (08:59 -0700)]
st/mesa: fix up comment
Brian Paul [Thu, 3 Dec 2009 16:49:27 +0000 (09:49 -0700)]
progs/glsl: added texture mapping to bump demo
Press 't' to toggle texture map.
Corbin Simpson [Mon, 7 Dec 2009 07:55:58 +0000 (23:55 -0800)]
radeong: Automatically softpipe for non-r3xx.
Well, technically non-r[345]xx. At any rate...
$ glxgears
libGL: OpenDriver: trying /home/simpson/mesa/lib/gallium/r600_dri.so
131 frames in 5.0 seconds = 26.107 FPS
I'm sure you can see where this is going. :3
Corbin Simpson [Mon, 7 Dec 2009 07:50:31 +0000 (23:50 -0800)]
radeong: Call softpipe_create directly.
Allows us to finally remove radeon_winsys_softpipe.
Corbin Simpson [Mon, 7 Dec 2009 07:49:02 +0000 (23:49 -0800)]
softpipe: sp_winsys.h should define/include what it needs.
Corbin Simpson [Mon, 7 Dec 2009 07:33:41 +0000 (23:33 -0800)]
r300g: Clean up previous commit.
If *I* can't read it, there's a strong possibility others can't,
either.
Marek Olšák [Sat, 5 Dec 2009 00:27:59 +0000 (01:27 +0100)]
r300g: enhance ZTOP conditions
Younes Manton [Sun, 6 Dec 2009 17:26:55 +0000 (12:26 -0500)]
nouveau: Work around nv04-nv40 miptrees not matching nouveau_miptree.
Thanks to Bob Gleitsmann for the patch.
I'll clean this up in a better way later if noone else beats me to it.
Vinson Lee [Sat, 21 Nov 2009 00:33:25 +0000 (16:33 -0800)]
st/egl: Fix memory leak in egl_tracker.c.
(cherry picked from commit
052b127842af3372fd768eae8e29b240a696a12a)
Marek Olšák [Sat, 5 Dec 2009 19:39:11 +0000 (20:39 +0100)]
r300g: don't render if everything is culled by scissoring
Otherwise a CS is refused by kernel 2.6.31 (and maybe all later
versions, not sure).
Marek Olšák [Sat, 5 Dec 2009 18:17:20 +0000 (19:17 +0100)]
r300g: remove redundant code and clean up
Vinson Lee [Sun, 29 Nov 2009 04:22:31 +0000 (23:22 -0500)]
dri: Fix potential null pointer dereference in driBindContext.
(cherry picked from commit
919898e92fa23ff71a59d86a46ff0886a6f34e4d)
Vinson Lee [Tue, 24 Nov 2009 05:57:55 +0000 (00:57 -0500)]
dri: Fix potential null pointer deference in dri_put_drawable.
(cherry picked from commit
364070b1f2b08d43fb205ec198894a35bec6b2f3)
Vinson Lee [Mon, 23 Nov 2009 06:30:32 +0000 (01:30 -0500)]
glx: Prevent potential null pointer deference in driCreateContext.
(cherry picked from commit
4b0b250aae6ae7d48cd24f9d91d05ab58086c4b2)
Vinson Lee [Fri, 9 Oct 2009 13:24:04 +0000 (07:24 -0600)]
progs/tests: Add tests to Makefile.
(cherry picked from commit
76d2ec3a0a047a65ffca70f53848241a3225dad3)
Alex Deucher [Sat, 5 Dec 2009 23:24:41 +0000 (18:24 -0500)]
radeon/r200/r600: fix drivers for changes in
433f0a82f5a4696e6b0c4061f645485ec8079bb4
Vinson Lee [Sat, 21 Nov 2009 00:03:48 +0000 (16:03 -0800)]
draw: Initialize variable in draw_pt.c.
(cherry picked from commit
ea98e9820d7117f7a187f355445796b1ef5d9e0c)
Vinson Lee [Sun, 29 Nov 2009 05:50:48 +0000 (00:50 -0500)]
mesa: Fix array out-of-bounds access by _mesa_TexGenf.
_mesa_TexGenf calls _mesa_TexGenfv, which uses the params argument
as an array.
(cherry picked from commit
ca5a7aadb4361e7d053aea8687372cd44cbd8795)
Vinson Lee [Sun, 29 Nov 2009 04:47:23 +0000 (23:47 -0500)]
mesa: Fix array out-of-bounds access by _mesa_TexGeni.
_mesa_TexGeni calls _mesa_TexGeniv, which uses the params argument
as an array.
(cherry picked from commit
d55fb7c835b56951f05a058083e7eda264ba192e)
Vinson Lee [Mon, 30 Nov 2009 02:17:44 +0000 (21:17 -0500)]
mesa: Fix array out-of-bounds access by _mesa_TexParameterf.
_mesa_TexParameterf calls set_tex_parameterf, which uses the params
argument as an array.
(cherry picked from commit
270d36da146b899d39e08f830fe34b63833a3731)
Vinson Lee [Sun, 29 Nov 2009 23:18:23 +0000 (18:18 -0500)]
mesa: Fix array out-of-bounds access by _mesa_TexParameterf.
_mesa_TexParameterf calls set_tex_parameteri, which uses the params
argument as an array.
(cherry picked from commit
a201dfb6bf28b89d6f511c2ec9ae0d81ef18511d)
Younes Manton [Sat, 5 Dec 2009 18:19:54 +0000 (13:19 -0500)]
st/xvmc: No more pf_get_block().
Michel Dänzer [Sat, 5 Dec 2009 16:48:00 +0000 (17:48 +0100)]
r300g: Need to emit a hardware scissor rectangle even if scissor is disabled.
Just make it cover the whole framebuffer in that case. Otherwise the kernel CS
checker may complain, e.g. running progs/demos/gearbox. That runs fast now
here, but doesn't look right yet.
Michel Dänzer [Sat, 5 Dec 2009 16:42:50 +0000 (17:42 +0100)]
st/mesa: Prefer alpha-less formats for RGB textures.
This can e.g. increase the chance of being able to accelerate
glCopyTex(Sub)Image from an alpha-less renderbuffer.
Michel Dänzer [Sat, 5 Dec 2009 16:20:03 +0000 (17:20 +0100)]
Add 'texture leak' test.
Michel Dänzer [Sat, 5 Dec 2009 16:20:03 +0000 (17:20 +0100)]
radeon: Only get DRI2 front buffer information for glXBindTexImageEXT.
Vinson Lee [Sat, 5 Dec 2009 09:43:29 +0000 (01:43 -0800)]
progs/samples: Fix memory leak if malloc fails in rgbtoppm.c.
Vinson Lee [Sat, 5 Dec 2009 09:38:14 +0000 (01:38 -0800)]
progs/samples: Fix memory leak if fopen fails in rgbtoppm.c.
Vinson Lee [Sat, 5 Dec 2009 09:28:47 +0000 (01:28 -0800)]
progs/samples: Add rgbtoppm to Makefile.
Vinson Lee [Sat, 5 Dec 2009 09:22:34 +0000 (01:22 -0800)]
progs/tests: Removed unused variable from texdown.c.
Vinson Lee [Sat, 5 Dec 2009 09:11:26 +0000 (01:11 -0800)]
progs/vpglsl: Assign glGetUniformLocationARB return value to GLint.
The return type of glGetUniformLocationARB is GLint, not GLuint.
Vinson Lee [Sat, 5 Dec 2009 07:31:39 +0000 (23:31 -0800)]
progs/fp: Redraw upon keypress.
José Fonseca [Sat, 5 Dec 2009 06:34:59 +0000 (06:34 +0000)]
svga: Dump SVGA_3D_CMD_BLIT_SURFACE_TO_SCREEN commands.
José Fonseca [Sat, 5 Dec 2009 06:05:56 +0000 (06:05 +0000)]
svga: Use _debug_printf, so that output may be dumped in release builds too.
The dump calls should be wrapped in #ifdef DEBUG .. #endif.
José Fonseca [Sat, 5 Dec 2009 05:43:53 +0000 (05:43 +0000)]
llvmpipe: Tweak disassembly to match gdb.
Helps verifying udis86 output.
José Fonseca [Sat, 5 Dec 2009 05:43:10 +0000 (05:43 +0000)]
llvmpipe: Stop disassembling when an unsupported opcode is found.
Otherwise the terminal gets full of garbage.
Vinson Lee [Sun, 29 Nov 2009 03:04:06 +0000 (22:04 -0500)]
mesa: Fix array out-of-bounds access in _mesa_TexEnvf.
_mesa_TexEnvf calls _mesa_TexEnvfv, which uses the param argument
as an array.
(cherry picked from commit
a11d60d14caf8efc07f70af63b57b33273f8cf9b)
Vinson Lee [Tue, 17 Nov 2009 20:00:22 +0000 (12:00 -0800)]
progs/util: Fix memory leak if fread returns 0 in CompileShaderFile.
(cherry picked from commit
11905da8836822f7dd60c84b5eefc72e46c94b50)
Vinson Lee [Tue, 3 Nov 2009 21:41:08 +0000 (14:41 -0700)]
progs/util: Fix memory leak if fail to load/compile shader
Signed-off-by: Brian Paul <brianp@vmware.com>
(cherry picked from commit
c475079ef2d901ba4506ebd53e19419cd46793ab)
Vinson Lee [Sat, 5 Dec 2009 01:29:53 +0000 (17:29 -0800)]
progs/fp: Add tri-inv.c to Makefile.
José Fonseca [Fri, 4 Dec 2009 21:25:40 +0000 (21:25 +0000)]
llvmpipe: Port vertex sampler support from softpipe.
Just enough boilerplate code to avoid segfaulting.
José Fonseca [Fri, 4 Dec 2009 21:16:14 +0000 (21:16 +0000)]
llvmpipe: Ensure transfers are mapped.
This shouldn't happen but it does by some misterious reason. Fail the
assertion but at least do not segfault on release builds.
José Fonseca [Fri, 4 Dec 2009 18:59:24 +0000 (18:59 +0000)]
llvmpipe: Remove debug printf.
José Fonseca [Fri, 4 Dec 2009 12:02:22 +0000 (12:02 +0000)]
gallium: Disable force_align_arg_pointer attribute on x86_64.
Apparently not only unnecessary but also causes gcc to complain.
Roland Scheidegger [Fri, 4 Dec 2009 17:34:52 +0000 (18:34 +0100)]
identity: fix copy&paste error
Marek Olšák [Fri, 4 Dec 2009 14:54:29 +0000 (15:54 +0100)]
radeong: flush CS if a buffer being mapped is referenced by it
Also, overlapping occlusion queries seems to work now.
Marek Olšák [Wed, 2 Dec 2009 16:15:27 +0000 (17:15 +0100)]
r300g, radeong: fix the CS overflow
Marek Olšák [Fri, 4 Dec 2009 15:06:16 +0000 (16:06 +0100)]
r300g: fix warnings
Roland Scheidegger [Fri, 4 Dec 2009 17:23:35 +0000 (18:23 +0100)]
svga: fix another pipe_reference strict aliasing violation
Coleman Kane [Fri, 4 Dec 2009 15:44:57 +0000 (08:44 -0700)]
r300g: use $(MAKE) variable
Fixes bug 24501