mesa.git
13 years agoutil: add a way to store translated indices to a user memory in u_index_modify
Marek Olšák [Wed, 29 Dec 2010 03:52:55 +0000 (04:52 +0100)]
util: add a way to store translated indices to a user memory in u_index_modify

I am about to use the upload buffer in r300g instead.

13 years agor300g: support user buffers as constant buffers
Marek Olšák [Wed, 29 Dec 2010 03:04:32 +0000 (04:04 +0100)]
r300g: support user buffers as constant buffers

13 years agoi965: Do lowering of array indexing of a vector in the FS.
Eric Anholt [Wed, 29 Dec 2010 01:19:10 +0000 (17:19 -0800)]
i965: Do lowering of array indexing of a vector in the FS.

Fixes a regression in ember since switching to the native FS backend,
and the new piglit tests glsl-fs-vec4-indexing-{2,3} for catching this.

13 years agoi965: Fix regression in FS comparisons on original gen4 due to gen6 changes.
Eric Anholt [Tue, 28 Dec 2010 21:55:14 +0000 (13:55 -0800)]
i965: Fix regression in FS comparisons on original gen4 due to gen6 changes.

Fixes 26 piglit cases on my GM965.

13 years agoi965: Factor out the ir comparision to BRW_CONDITIONAL_* code.
Eric Anholt [Tue, 28 Dec 2010 05:08:59 +0000 (21:08 -0800)]
i965: Factor out the ir comparision to BRW_CONDITIONAL_* code.

13 years agoglcpp: Add negative tests for redefintions with valueless macros.
Vinson Lee [Tue, 28 Dec 2010 07:20:35 +0000 (23:20 -0800)]
glcpp: Add negative tests for redefintions with valueless macros.

13 years agotgsi_dump: fix assert due to missing property name.
Dave Airlie [Tue, 28 Dec 2010 06:52:19 +0000 (16:52 +1000)]
tgsi_dump: fix assert due to missing property name.

13 years agor300g: rename aos to vertex arrays
Marek Olšák [Tue, 28 Dec 2010 03:52:36 +0000 (04:52 +0100)]
r300g: rename aos to vertex arrays

13 years agor300g: mark vertex arrays as dirty after a buffer_offset change
Marek Olšák [Mon, 27 Dec 2010 15:15:24 +0000 (16:15 +0100)]
r300g: mark vertex arrays as dirty after a buffer_offset change

We shouldn't hit this bug in theory.

NOTE: This is a candidate for the 7.10 branch.

13 years agoi965: Fix occlusion query on sandybridge
Zhenyu Wang [Tue, 28 Dec 2010 03:03:45 +0000 (11:03 +0800)]
i965: Fix occlusion query on sandybridge

Clear target query buffer fixed occlusion query on sandybridge.

https://bugs.freedesktop.org/show_bug.cgi?id=32167

13 years agoRevert "i965: upload multisample state for fragment program change"
Zhenyu Wang [Tue, 28 Dec 2010 01:21:20 +0000 (09:21 +0800)]
Revert "i965: upload multisample state for fragment program change"

This reverts commit de6fd527a545f8344e074312544517d05573fb72.

Revert this workaround as it seems the real trouble is caused by
lineloop, which doesn't require GS convert on sandybridge actually.

13 years agoi965: Internally enable GL_NV_blend_square on ES2.
Kenneth Graunke [Mon, 27 Dec 2010 23:44:52 +0000 (15:44 -0800)]
i965: Internally enable GL_NV_blend_square on ES2.

Hopefully should fix bug #32520.

13 years agoi965: don't spawn GS thread for LINELOOP on Sandybridge
Xiang, Haihao [Mon, 27 Dec 2010 08:39:08 +0000 (16:39 +0800)]
i965: don't spawn GS thread for LINELOOP on Sandybridge

LINELOOP is converted to LINESTRIP at the beginning of the 3D pipeline.
This fixes https://bugs.freedesktop.org/show_bug.cgi?id=32596

13 years agoi965: Flatten if-statements beyond depth 16 on pre-gen6.
Kenneth Graunke [Mon, 27 Dec 2010 08:40:26 +0000 (00:40 -0800)]
i965: Flatten if-statements beyond depth 16 on pre-gen6.

Gen4 and Gen5 hardware can have a maximum supported nesting depth of 16.
Previously, shaders with control flow nested 17 levels deep would
cause a driver assertion or segmentation fault.

Gen6 (Sandybridge) hardware no longer has this restriction.

Fixes fd.o bug #31967.

13 years agoglsl: Support if-flattening beyond a given maximum nesting depth.
Kenneth Graunke [Mon, 27 Dec 2010 08:22:38 +0000 (00:22 -0800)]
glsl: Support if-flattening beyond a given maximum nesting depth.

This adds a new optional max_depth parameter (defaulting to 0) to
lower_if_to_cond_assign, and makes the pass only flatten if-statements
nested deeper than that.

By default, all if-statements will be flattened, just like before.

This patch also renames do_if_to_cond_assign to lower_if_to_cond_assign,
to match the new naming conventions.

13 years agosvga, glhd: Remove incorrect assert and add note
Jakob Bornecrantz [Wed, 15 Dec 2010 11:17:26 +0000 (12:17 +0100)]
svga, glhd: Remove incorrect assert and add note

Stride can be lower then the size of the attribute.
But should probably be aligned to component size atleast for floats.

13 years agosvga: Minor debug text fix
Jakob Bornecrantz [Thu, 16 Dec 2010 03:05:59 +0000 (04:05 +0100)]
svga: Minor debug text fix

13 years agosvga: Remove debug print in winsys
Jakob Bornecrantz [Thu, 16 Dec 2010 02:00:16 +0000 (03:00 +0100)]
svga: Remove debug print in winsys

13 years agosvga: Correct spelling in swtnl backend
Jakob Bornecrantz [Thu, 16 Dec 2010 02:42:56 +0000 (03:42 +0100)]
svga: Correct spelling in swtnl backend

13 years agosvga: Fix newline at EOF
Jakob Bornecrantz [Wed, 15 Dec 2010 11:14:33 +0000 (12:14 +0100)]
svga: Fix newline at EOF

13 years agosvga: Add Galahad and Softpipe to scons build
Jakob Bornecrantz [Wed, 15 Dec 2010 11:13:54 +0000 (12:13 +0100)]
svga: Add Galahad and Softpipe to scons build

13 years agowrapper: Flush pipe on unmap
Jakob Bornecrantz [Thu, 16 Dec 2010 06:24:32 +0000 (07:24 +0100)]
wrapper: Flush pipe on unmap

For drivers that does DMA transfers instead of mapping directly

13 years agowrapper: Fix width and height given to map and remove uneeded fields
Jakob Bornecrantz [Wed, 15 Dec 2010 11:17:06 +0000 (12:17 +0100)]
wrapper: Fix width and height given to map and remove uneeded fields

13 years agoi915g: Ignore color0 writes all cbufs tgsi property
Jakob Bornecrantz [Sat, 25 Dec 2010 14:01:09 +0000 (14:01 +0000)]
i915g: Ignore color0 writes all cbufs tgsi property

13 years agost/egl: Fix eglChooseConfig when configs is NULL.
Chia-I Wu [Sun, 26 Dec 2010 15:34:57 +0000 (23:34 +0800)]
st/egl: Fix eglChooseConfig when configs is NULL.

When configs is NULL, the app wants to know the number of matching
configs.

13 years agoswrast: Clean up header file inclusion in ss_vb.h.
Vinson Lee [Sun, 26 Dec 2010 04:53:27 +0000 (20:53 -0800)]
swrast: Clean up header file inclusion in ss_vb.h.

13 years agoswrast: Clean up header file inclusion in ss_triangle.h.
Vinson Lee [Sun, 26 Dec 2010 04:48:29 +0000 (20:48 -0800)]
swrast: Clean up header file inclusion in ss_triangle.h.

13 years agoswrast: Clean up header file inclusion in s_texfilter.h.
Vinson Lee [Sun, 26 Dec 2010 04:28:17 +0000 (20:28 -0800)]
swrast: Clean up header file inclusion in s_texfilter.h.

13 years agoswrast: Clean up header file inclusion in s_texcombine.h.
Vinson Lee [Sun, 26 Dec 2010 04:12:06 +0000 (20:12 -0800)]
swrast: Clean up header file inclusion in s_texcombine.h.

13 years agoswrast: Clean up header file inclusion in s_masking.h.
Vinson Lee [Sun, 26 Dec 2010 04:03:33 +0000 (20:03 -0800)]
swrast: Clean up header file inclusion in s_masking.h.

13 years agonvfx: Remove unused variable.
Vinson Lee [Sun, 26 Dec 2010 03:09:54 +0000 (19:09 -0800)]
nvfx: Remove unused variable.

Fixes this GCC warning.
nvfx_vbo.c: In function 'nvfx_idxbuf_emit':
nvfx_vbo.c:410: warning: unused variable 'eng3d'

13 years agonvfx: restore BEGIN_RING usage
Xavier Chantry [Sat, 25 Dec 2010 15:39:01 +0000 (16:39 +0100)]
nvfx: restore BEGIN_RING usage

Michel Hermier reported libdrm segfault (and kernel crash) on nv40 using
gallium :
http://www.mail-archive.com/nouveau@lists.freedesktop.org/msg06563.html

It turns out these were caused by some missing WAIT_RING (or wrong
computation of the WAIT_RING sizes). Unlike all other libdrm_nouveau users,
nvfx gallium tried to use a mininum calls of WAIT_RING, one WAIT_RING could
apply to many methods for different code paths and spread across several
functions. This made it too tricky to find out what the missing or wrong
WAIT_RING was.

By restoring BEGIN_RING, we force one WAIT_RING per method, and it's much
easier to check if the free size required in the pushbuffer is correct.  As
curro said, "let's keep it simple for the maintainers until the big
bottlenecks are gone"

Benchmarked on nv35 with openarena, nexuiz and ut2004 and no performance
regression.

The core of this patch was made with Coccinelle, with minor manual fixes
made on top.

Tested-by: Michel Hermier <hermier@frugalware.org>
Signed-off-by: Francisco Jerez <currojerez@riseup.net>
13 years agointel: Only do frame throttling at glFlush time when using frontbuffer.
Eric Anholt [Sat, 25 Dec 2010 16:57:22 +0000 (08:57 -0800)]
intel: Only do frame throttling at glFlush time when using frontbuffer.

This is the hack for input interactivity of frontbuffer rendering
(like we do for backbuffer at intelDRI2Flush()) by waiting for the n-2
frame to complete before starting a new one.  However, for an
application doing multiple contexts or regular rebinding of a single
context, this would end up lockstepping the CPU to the GPU because
every unbind was considered the end of a frame.

Improves WOW performance on my Ironlake by 48.8% (+/- 2.3%, n=5)

13 years agor300g: simplify buffer_transfer_inline_write
Marek Olšák [Sat, 25 Dec 2010 13:48:34 +0000 (14:48 +0100)]
r300g: simplify buffer_transfer_inline_write

13 years agor300g: simplify the code for buffer uploads
Marek Olšák [Sat, 25 Dec 2010 13:46:45 +0000 (14:46 +0100)]
r300g: simplify the code for buffer uploads

13 years agor300g: user index buffers are always aligned
Marek Olšák [Sat, 25 Dec 2010 13:39:35 +0000 (14:39 +0100)]
r300g: user index buffers are always aligned

13 years agor300g: increase the size of upload buffers
Marek Olšák [Sat, 25 Dec 2010 13:39:07 +0000 (14:39 +0100)]
r300g: increase the size of upload buffers

13 years agoswrast: Clean up header file inclusion in s_logic.h.
Vinson Lee [Sat, 25 Dec 2010 04:43:35 +0000 (20:43 -0800)]
swrast: Clean up header file inclusion in s_logic.h.

13 years agoswrast: Clean up header file inclusion in s_fragprog.h.
Vinson Lee [Sat, 25 Dec 2010 04:34:53 +0000 (20:34 -0800)]
swrast: Clean up header file inclusion in s_fragprog.h.

13 years agoswrast: Clean up header file inclusion in s_span.h.
Vinson Lee [Sat, 25 Dec 2010 04:17:18 +0000 (20:17 -0800)]
swrast: Clean up header file inclusion in s_span.h.

13 years agoswrast: Clean up header file inclusion in s_fog.h.
Vinson Lee [Sat, 25 Dec 2010 04:11:44 +0000 (20:11 -0800)]
swrast: Clean up header file inclusion in s_fog.h.

13 years agoswrast: Clean up header file inclusion in s_depth.h.
Vinson Lee [Sat, 25 Dec 2010 04:06:11 +0000 (20:06 -0800)]
swrast: Clean up header file inclusion in s_depth.h.

13 years agoswrast: Clean up header file inclusion in s_blend.h.
Vinson Lee [Sat, 25 Dec 2010 03:55:42 +0000 (19:55 -0800)]
swrast: Clean up header file inclusion in s_blend.h.

13 years agoswrast: Clean up header file inclusion in s_atifragshader.h.
Vinson Lee [Sat, 25 Dec 2010 03:47:54 +0000 (19:47 -0800)]
swrast: Clean up header file inclusion in s_atifragshader.h.

13 years agoswrast: Clean up header file inclusion in s_alpha.h.
Vinson Lee [Sat, 25 Dec 2010 03:30:38 +0000 (19:30 -0800)]
swrast: Clean up header file inclusion in s_alpha.h.

13 years agoswrast: Clean up header file inclusion in s_accum.h.
Vinson Lee [Sat, 25 Dec 2010 03:25:30 +0000 (19:25 -0800)]
swrast: Clean up header file inclusion in s_accum.h.

13 years agoswrast: Clean up header file inclusion in s_aatriangle.h.
Vinson Lee [Sat, 25 Dec 2010 02:48:00 +0000 (18:48 -0800)]
swrast: Clean up header file inclusion in s_aatriangle.h.

13 years agoswrast: Clean up header file inclusion in s_aaline.h.
Vinson Lee [Sat, 25 Dec 2010 02:35:10 +0000 (18:35 -0800)]
swrast: Clean up header file inclusion in s_aaline.h.

13 years agost/mesa: Clean up header file inclusion in st_mesa_to_tgsi.h.
Vinson Lee [Sat, 25 Dec 2010 02:27:55 +0000 (18:27 -0800)]
st/mesa: Clean up header file inclusion in st_mesa_to_tgsi.h.

13 years agost/mesa: Clean up header file inclusion in st_gen_mipmap.h.
Vinson Lee [Sat, 25 Dec 2010 02:06:20 +0000 (18:06 -0800)]
st/mesa: Clean up header file inclusion in st_gen_mipmap.h.

13 years agodocs/egl: Update egl.html.
Chia-I Wu [Fri, 24 Dec 2010 18:45:13 +0000 (02:45 +0800)]
docs/egl: Update egl.html.

Various updates and a new section about packaging.

13 years agor300g/swtcl: re-enable LLVM
Marek Olšák [Fri, 24 Dec 2010 17:16:16 +0000 (18:16 +0100)]
r300g/swtcl: re-enable LLVM

Based on a patch from Drill <drill87@gmail.com>.

NOTE: This is a candidate for the 7.10 branch.

13 years agor600g: r600_blit_uncompress_depth() can't fail.
Henri Verbeet [Fri, 24 Dec 2010 10:36:50 +0000 (11:36 +0100)]
r600g: r600_blit_uncompress_depth() can't fail.

13 years agor600g: Get rid of r600_blit_uncompress_depth_ptr.
Henri Verbeet [Fri, 24 Dec 2010 10:36:50 +0000 (11:36 +0100)]
r600g: Get rid of r600_blit_uncompress_depth_ptr.

13 years agomapi: Move mapi_func typedef to entry.h.
Chia-I Wu [Fri, 24 Dec 2010 09:26:08 +0000 (17:26 +0800)]
mapi: Move mapi_func typedef to entry.h.

Make it clear that entry.h does not depend on stub.h.

13 years agomapi: Define MAPI_TMP_DEFINES only when needed.
Chia-I Wu [Fri, 24 Dec 2010 08:04:14 +0000 (16:04 +0800)]
mapi: Define MAPI_TMP_DEFINES only when needed.

Since struct mapi_table is opaque, MAPI_TMP_DEFINES is not needed in
table.h.

13 years agomapi: Add and use entry_get_public.
Chia-I Wu [Fri, 24 Dec 2010 07:06:41 +0000 (15:06 +0800)]
mapi: Add and use entry_get_public.

Given a dispatch slot, entry_get_public returns the address of the
corresponding public entry point.  There may be more than one of them.
But since they are all equivalent, it is fine to return any one of them.

With entry_get_public, the address of any public entry point can be
calculated at runtime when an assembly dispatcher is used.  There is no
need to have a mapping table in such case.  This omits the unnecessary
relocations from the binary.

13 years agomapi: Make struct mapi_stub opaque.
Chia-I Wu [Fri, 24 Dec 2010 05:49:01 +0000 (13:49 +0800)]
mapi: Make struct mapi_stub opaque.

Add accessors for struct mapi_stub and make it opaque.

13 years agomapi: Allow blocks to be disabled from the output.
Chia-I Wu [Thu, 16 Dec 2010 17:31:43 +0000 (01:31 +0800)]
mapi: Allow blocks to be disabled from the output.

For example, a printer may ask not to output noop dispatch table.

13 years agomapi: Fix hidden entries.
Chia-I Wu [Fri, 10 Dec 2010 16:26:32 +0000 (00:26 +0800)]
mapi: Fix hidden entries.

Hidden entries are just like normal entries except that they are not
exported.  Since it is not always possible to hide them, and two hidden
aliases can share the same entry, the name of hidden aliases are mangled
to '_dispatch_stub_<slot>'.

13 years agomapi: Add "handcode" attribute to the script.
Chia-I Wu [Tue, 7 Dec 2010 05:33:11 +0000 (13:33 +0800)]
mapi: Add "handcode" attribute to the script.

Entries with handcode attribute will be handled normally, except no
entry point will be generated for them.

13 years agomapi: Minor ABIPrinter refactoring.
Chia-I Wu [Fri, 10 Dec 2010 15:57:36 +0000 (23:57 +0800)]
mapi: Minor ABIPrinter refactoring.

Split out function name generation from _c_decl to _c_function, and use
it everywhere.  Add an optional 'export' argument to _cdecl.  It is
prepended to the returned string.

13 years agomapi: Store alias entry instead of alias name.
Chia-I Wu [Tue, 14 Dec 2010 19:33:30 +0000 (03:33 +0800)]
mapi: Store alias entry instead of alias name.

An entry can hold more info than plain name.

13 years agor600g: hack around property unknown issues.
Dave Airlie [Fri, 24 Dec 2010 07:33:41 +0000 (17:33 +1000)]
r600g: hack around property unknown issues.

should fix https://bugs.freedesktop.org/show_bug.cgi?id=32619

Need to add proper support for properties later.

Signed-off-by: Dave Airlie <airlied@redhat.com>
13 years agor300g: turn back on rv530 hiz.
Dave Airlie [Fri, 24 Dec 2010 05:44:36 +0000 (15:44 +1000)]
r300g: turn back on rv530 hiz.

still needs RADEON_HYPERZ=y env var.

Signed-off-by: Dave Airlie <airlied@redhat.com>
13 years agor300g: hyperz fixing typo.
Dave Airlie [Fri, 24 Dec 2010 05:41:30 +0000 (15:41 +1000)]
r300g: hyperz fixing typo.

Really no idea why I didn't see this before, but these values were opposite
the register spec.

this seems to fix rv530 HiZ on my laptop, will reenable in next commit.

Signed-off-by: Dave Airlie <airlied@redhat.com>
13 years agomesa: Assert format is not MESA_FORMAT_COUNT in _mesa_format_to_type_and_comps.
Vinson Lee [Fri, 24 Dec 2010 02:16:59 +0000 (18:16 -0800)]
mesa: Assert format is not MESA_FORMAT_COUNT in _mesa_format_to_type_and_comps.

The case of format being MESA_FORMAT_COUNT should never occur.

13 years agoi965: use align1 access mode for instructions with execSize=1 in VS
Xiang, Haihao [Fri, 24 Dec 2010 01:34:50 +0000 (09:34 +0800)]
i965: use align1 access mode for instructions with execSize=1 in VS

All operands must be 16-bytes aligned in aligh16 mode. This fixes l_xxx.c
in oglconform.

13 years agoi965: fix register region description
Xiang, Haihao [Fri, 24 Dec 2010 01:24:08 +0000 (09:24 +0800)]
i965: fix register region description

This fixes
 brw_eu_emit.c:179: validate_reg: Assertion `width == 1' failed.

13 years agor600g: Rearrange print order of outputs of R600_ERR.
Vinson Lee [Fri, 24 Dec 2010 01:25:32 +0000 (17:25 -0800)]
r600g: Rearrange print order of outputs of R600_ERR.

13 years agomesa: Assert _mesa_DeleteFragmentShaderATI doesn't ever free static DummyShader.
Vinson Lee [Fri, 24 Dec 2010 00:43:15 +0000 (16:43 -0800)]
mesa: Assert _mesa_DeleteFragmentShaderATI doesn't ever free static DummyShader.

13 years agost/egl: Remove unnecessary header.
Vinson Lee [Fri, 24 Dec 2010 00:23:53 +0000 (16:23 -0800)]
st/egl: Remove unnecessary header.

13 years agolibgl-xlib: Remove unnecessary header.
Vinson Lee [Fri, 24 Dec 2010 00:19:11 +0000 (16:19 -0800)]
libgl-xlib: Remove unnecessary header.

13 years agor300g: Remove unnecessary header.
Vinson Lee [Fri, 24 Dec 2010 00:05:28 +0000 (16:05 -0800)]
r300g: Remove unnecessary header.

13 years agotgsi_text: just parse as an integer (value is a boolean).
Dave Airlie [Thu, 23 Dec 2010 23:28:14 +0000 (09:28 +1000)]
tgsi_text: just parse as an integer (value is a boolean).

fixes warning reported by vlee on irc.

Signed-off-by: Dave Airlie <airlied@redhat.com>
13 years agomapi: Remove unnecessary header.
Vinson Lee [Thu, 23 Dec 2010 23:25:38 +0000 (15:25 -0800)]
mapi: Remove unnecessary header.

13 years agointel: Remove unnecessary headers.
Vinson Lee [Thu, 23 Dec 2010 23:08:53 +0000 (15:08 -0800)]
intel: Remove unnecessary headers.

13 years agor300g: add support for color0 writes to all bound color buffers.
Dave Airlie [Sat, 18 Dec 2010 00:40:33 +0000 (10:40 +1000)]
r300g: add support for color0 writes to all bound color buffers.

Thanks to Marek Olšák for making my initial attempt actually work.

Signed-off-by: Dave Airlie <airlied@redhat.com>
13 years agomesa/st: set the color write cbuf property for fragColor writes
Dave Airlie [Sat, 18 Dec 2010 00:36:55 +0000 (10:36 +1000)]
mesa/st: set the color write cbuf property for fragColor writes

13 years agosoftpipe: add support for color writes all color bufs property
Dave Airlie [Sat, 18 Dec 2010 00:36:31 +0000 (10:36 +1000)]
softpipe: add support for color writes all color bufs property

13 years agogallium: add fragment shader property for color writes to all buffers. (v2)
Dave Airlie [Sat, 18 Dec 2010 00:34:35 +0000 (10:34 +1000)]
gallium: add fragment shader property for color writes to all buffers. (v2)

For GL fragColor semantics we need to tell the pipe drivers that the fragment
shader color result is to be replicated to all bound color buffers, this
adds the basic TGSI + documentation.

v2: fix missing comma pointed out by Tilman on mesa-dev.

Signed-off-by: Dave Airlie <airlied@redhat.com>
13 years agoi965: Remove unnecessary headers.
Vinson Lee [Thu, 23 Dec 2010 19:54:29 +0000 (11:54 -0800)]
i965: Remove unnecessary headers.

13 years agomesa: Fix #ifdef typo in _mesa_format_to_type_and_comps.
Vinson Lee [Thu, 23 Dec 2010 19:32:16 +0000 (11:32 -0800)]
mesa: Fix #ifdef typo in _mesa_format_to_type_and_comps.

According to the comment, the warning should be for debug builds.

13 years agor300g: use a simpler fallback for misaligned ushort indices with triangles
Marek Olšák [Wed, 22 Dec 2010 23:22:38 +0000 (00:22 +0100)]
r300g: use a simpler fallback for misaligned ushort indices with triangles

If 'start' is odd, render the first triangle with indices embedded
in the command stream, which adds 3 to 'start' and makes it even.
Then continue with the fast path.

13 years agor300g: add support for B2G3R3 texturing
Marek Olšák [Wed, 22 Dec 2010 12:28:55 +0000 (13:28 +0100)]
r300g: add support for B2G3R3 texturing

13 years agomesa: fix texel store functions for some float formats
Marek Olšák [Mon, 29 Nov 2010 04:24:18 +0000 (05:24 +0100)]
mesa: fix texel store functions for some float formats

These are copy-paste errors obviously.

13 years agost/mesa: do not require all texture formats to be renderable
Marek Olšák [Wed, 22 Dec 2010 00:58:52 +0000 (01:58 +0100)]
st/mesa: do not require all texture formats to be renderable

This is a bandaid on the problem that if some formats were not renderable
(like luminance_alpha), st/mesa fell back to some RGBA format, so basically
some non-renderable formats were actually not used at all. This is only
a problem with hardware drivers, softpipe can render to anything.

Instead, require only RGB8/RGBA8 to be renderable.

13 years agost/mesa: use the formats RGB233, ARGB2101010, AL44, AL1616, A16, L16, I16
Marek Olšák [Wed, 22 Dec 2010 00:44:30 +0000 (01:44 +0100)]
st/mesa: use the formats RGB233, ARGB2101010, AL44, AL1616, A16, L16, I16

13 years agogallium: add new formats L16A16_UNORM, A16_UNORM, I16_UNORM, B2G3R3_UNORM
Marek Olšák [Tue, 21 Dec 2010 04:43:52 +0000 (05:43 +0100)]
gallium: add new formats L16A16_UNORM, A16_UNORM, I16_UNORM, B2G3R3_UNORM

13 years agomesa: implement new texture format I16
Marek Olšák [Wed, 22 Dec 2010 00:12:20 +0000 (01:12 +0100)]
mesa: implement new texture format I16

13 years agomesa: implement new texture format L16
Marek Olšák [Wed, 22 Dec 2010 00:11:13 +0000 (01:11 +0100)]
mesa: implement new texture format L16

13 years agomesa: implement new texture format A16
Marek Olšák [Tue, 21 Dec 2010 23:32:33 +0000 (00:32 +0100)]
mesa: implement new texture format A16

13 years agomesa: implement new texture format AL44
Marek Olšák [Tue, 21 Dec 2010 22:46:32 +0000 (23:46 +0100)]
mesa: implement new texture format AL44

Radeon GPUs can do this. R600 can even do render-to-texture.
Packing and extracting aren't implemented, but we shouldn't hit them (I think).
Tested with swrast, softpipe, and r300g.

13 years agomesa: implement new texture format ARGB2101010
Marek Olšák [Tue, 21 Dec 2010 17:54:50 +0000 (18:54 +0100)]
mesa: implement new texture format ARGB2101010

Radeon GPUs do support GL_RGB10_A2.

13 years agost/mesa: if Z32 is unsupported, prefer Z24 to Z16
Marek Olšák [Tue, 21 Dec 2010 04:50:58 +0000 (05:50 +0100)]
st/mesa: if Z32 is unsupported, prefer Z24 to Z16

13 years agost/mesa: use RGBA16 for RGB12 and RGB16
Marek Olšák [Tue, 21 Dec 2010 04:45:46 +0000 (05:45 +0100)]
st/mesa: use RGBA16 for RGB12 and RGB16

To provide enough precision if a user wants it.

13 years agost/mesa: use DXT SRGB formats for COMPRESSED_SRGB
Marek Olšák [Tue, 21 Dec 2010 05:41:59 +0000 (06:41 +0100)]
st/mesa: use DXT SRGB formats for COMPRESSED_SRGB

And also check if the formats are supported to return something meaningful
if compression cannot be used.

13 years agoi965: Keep around a copy of the VS constant surface dumping code.
Eric Anholt [Thu, 23 Dec 2010 06:48:24 +0000 (22:48 -0800)]
i965: Keep around a copy of the VS constant surface dumping code.

Just like everywhere else, I never trust my constant uploads to
correctly put constants in the right places, even though that's so
rarely where the issue is.

13 years agoi965: Correct the dp_read message descriptor setup on g4x.
Eric Anholt [Thu, 23 Dec 2010 06:46:12 +0000 (22:46 -0800)]
i965: Correct the dp_read message descriptor setup on g4x.

It's mostly like gen4 message descriptor setup, except that the sizes
of type/control changed to be like gen5.  Fixes 21 piglit cases on
gm45, including the regressions in bug #32311 from increased VS
constant buffer usage.

13 years agoi965: upload multisample state for fragment program change
Zhenyu Wang [Thu, 23 Dec 2010 08:30:20 +0000 (16:30 +0800)]
i965: upload multisample state for fragment program change

This makes conformance tests stable on sandybridge D0 to track
multisample state before SF/WM state.