profile/ivi/mesa.git
14 years agogallivm: Clamp mipmap level and zero mip weight simultaneously.
José Fonseca [Fri, 8 Oct 2010 13:05:50 +0000 (14:05 +0100)]
gallivm: Clamp mipmap level and zero mip weight simultaneously.

14 years agogallivm: Use lp_build_ifloor_fract for lod computation.
José Fonseca [Fri, 8 Oct 2010 12:36:18 +0000 (13:36 +0100)]
gallivm: Use lp_build_ifloor_fract for lod computation.

Forgot this one before.

14 years agogallivm: Don't compute the second mipmap level when frac(lod) == 0
José Fonseca [Fri, 8 Oct 2010 12:26:37 +0000 (13:26 +0100)]
gallivm: Don't compute the second mipmap level when frac(lod) == 0

14 years agogallivm: Simplify lp_build_mipmap_level_sizes' interface.
José Fonseca [Fri, 8 Oct 2010 12:24:39 +0000 (13:24 +0100)]
gallivm: Simplify lp_build_mipmap_level_sizes' interface.

14 years agogallivm: Do not do mipfiltering when magnifying.
José Fonseca [Fri, 8 Oct 2010 09:54:23 +0000 (10:54 +0100)]
gallivm: Do not do mipfiltering when magnifying.

If lod < 0, then invariably follows that ilevel0 == ilevel1 == 0.

14 years agor600g: Remove unnecessary header.
Vinson Lee [Fri, 8 Oct 2010 11:56:49 +0000 (04:56 -0700)]
r600g: Remove unnecessary header.

14 years agor600g: drop width/height per level storage.
Dave Airlie [Fri, 8 Oct 2010 09:55:05 +0000 (19:55 +1000)]
r600g: drop width/height per level storage.

these aren't used anywhere, so just waste memory.

14 years agoi965: Normalize cubemap coordinates like is done in the Mesa IR path.
Eric Anholt [Wed, 6 Oct 2010 05:30:42 +0000 (22:30 -0700)]
i965: Normalize cubemap coordinates like is done in the Mesa IR path.

Fixes glsl-fs-texturecube-2-*

14 years agoi965: Disable emitting if () statements on gen6 until we really fix them.
Eric Anholt [Thu, 7 Oct 2010 16:13:09 +0000 (09:13 -0700)]
i965: Disable emitting if () statements on gen6 until we really fix them.

14 years agor600g: add some RG texture format support.
Dave Airlie [Thu, 7 Oct 2010 23:35:17 +0000 (09:35 +1000)]
r600g: add some RG texture format support.

14 years agogles2: Add GL_EXT_texture_format_BGRA8888 support
Kristian Høgsberg [Thu, 7 Oct 2010 21:03:53 +0000 (17:03 -0400)]
gles2: Add GL_EXT_texture_format_BGRA8888 support

14 years agogallivm: Vectorize the rho computation.
José Fonseca [Thu, 7 Oct 2010 21:03:59 +0000 (22:03 +0100)]
gallivm: Vectorize the rho computation.

14 years agor600g: fix Z export enable bits.
Dave Airlie [Thu, 7 Oct 2010 05:32:05 +0000 (15:32 +1000)]
r600g: fix Z export enable bits.

we should be checking output array not input to decide.

Signed-off-by: Dave Airlie <airlied@redhat.com>
14 years agor600g: use format from the sampler view not from the texture.
Dave Airlie [Thu, 7 Oct 2010 05:13:09 +0000 (15:13 +1000)]
r600g: use format from the sampler view not from the texture.

we want to use the format from the sampler view which isn't always the
same as the texture format when creating sampler views.

14 years agor600g: fix evergreen interpolation setup
Andre Maasikas [Wed, 6 Oct 2010 18:14:15 +0000 (21:14 +0300)]
r600g: fix evergreen interpolation setup

interp data is stored in gpr0 so first interp overwrote it
and subsequent ones got wrong values

reserve register 0 so it's not used for attribs.
alternative is to interpolate attrib0 last (reverse, as r600c does)

14 years agost/vega: Fix version check in context creation.
Chia-I Wu [Thu, 7 Oct 2010 04:14:38 +0000 (12:14 +0800)]
st/vega: Fix version check in context creation.

This fixes a regression since 4531356817ec8383ac35932903773de67af92e37.

14 years agotargets/egl: Fix linking with libdrm.
Chia-I Wu [Thu, 7 Oct 2010 04:06:07 +0000 (12:06 +0800)]
targets/egl: Fix linking with libdrm.

14 years agoi965: Fix gen6 pointsize handling to match pre-gen6.
Eric Anholt [Thu, 7 Oct 2010 00:29:29 +0000 (17:29 -0700)]
i965: Fix gen6 pointsize handling to match pre-gen6.

Fixes point-line-no-cull.
Bug #30532

14 years agoi965: Don't assume that WPOS is always provided on gen6 in the new FS.
Eric Anholt [Wed, 6 Oct 2010 19:10:31 +0000 (12:10 -0700)]
i965: Don't assume that WPOS is always provided on gen6 in the new FS.

We sensibly only provide it if the FS asks for it.  We could actually
skip WPOS unless the FS needed WPOS.zw, but that's something for
later.

Fixes: glsl-texture2d and probably many others.

14 years agoi965: Add support for gl_FrontFacing on gen6.
Eric Anholt [Wed, 6 Oct 2010 18:19:48 +0000 (11:19 -0700)]
i965: Add support for gl_FrontFacing on gen6.

Fixes glsl1-gl_FrontFacing var (2) with new FS.

14 years agoi965: Refactor gl_FrontFacing setup out of general variable setup.
Eric Anholt [Wed, 6 Oct 2010 18:13:22 +0000 (11:13 -0700)]
i965: Refactor gl_FrontFacing setup out of general variable setup.

14 years agoi965: Gen6's sampler messages are the same as Ironlake.
Eric Anholt [Wed, 6 Oct 2010 18:04:02 +0000 (11:04 -0700)]
i965: Gen6's sampler messages are the same as Ironlake.

This should fix texturing in the new FS backend.

14 years agoi965: Don't do 1/w multiplication in new FS for gen6
Eric Anholt [Wed, 6 Oct 2010 18:00:31 +0000 (11:00 -0700)]
i965: Don't do 1/w multiplication in new FS for gen6

Not needed now that we're doing barycentric.

14 years agoi965: Add some clarification of the WECtrl field.
Eric Anholt [Wed, 6 Oct 2010 17:22:22 +0000 (10:22 -0700)]
i965: Add some clarification of the WECtrl field.

14 years agoi965: Fix botch in the header_present case in the new FS.
Eric Anholt [Wed, 6 Oct 2010 18:25:05 +0000 (11:25 -0700)]
i965: Fix botch in the header_present case in the new FS.

I only set it on the color_regions == 0 case, missing the important
case, causing GPU hangs on pre-gen6.

14 years agollvmpipe: Cleanup depth-stencil clears.
José Fonseca [Wed, 6 Oct 2010 09:11:15 +0000 (10:11 +0100)]
llvmpipe: Cleanup depth-stencil clears.

Only cosmetic changes. No actual practical difference.

14 years agoutil: Cleanup util_pack_z_stencil and friends.
José Fonseca [Wed, 6 Oct 2010 09:09:37 +0000 (10:09 +0100)]
util: Cleanup util_pack_z_stencil and friends.

- Handle PIPE_FORMAT_Z32_FLOAT packing correctly.

- In the integer version z shouldn't be passed as as double.

- Make it clear that the integer versions should only be used for masks.

- Make integer type sizes explicit (uint32_t for now, although
  uint64_t will be necessary later to encode f32_s8_x24).

14 years agogallivm: Compute lod as integer whenever possible.
José Fonseca [Wed, 6 Oct 2010 17:44:51 +0000 (18:44 +0100)]
gallivm: Compute lod as integer whenever possible.

More accurate/faster results for PIPE_TEX_MIPFILTER_NEAREST. Less
FP <-> SI conversion overall.

14 years agogallivm: Only apply min/max_lod when necessary.
José Fonseca [Wed, 6 Oct 2010 13:53:19 +0000 (14:53 +0100)]
gallivm: Only apply min/max_lod when necessary.

14 years agogallivm: don't apply zero lod_bias
Keith Whitwell [Thu, 30 Sep 2010 15:43:56 +0000 (16:43 +0100)]
gallivm: don't apply zero lod_bias

14 years agogallivm: Combined ifloor & fract helper.
José Fonseca [Wed, 6 Oct 2010 17:31:36 +0000 (18:31 +0100)]
gallivm: Combined ifloor & fract helper.

The only way to ensure we don't do redundant FP <-> SI conversions.

14 years agogallivm: Fast implementation of iround(log2(x))
José Fonseca [Wed, 6 Oct 2010 16:44:05 +0000 (17:44 +0100)]
gallivm: Fast implementation of iround(log2(x))

Not tested yet, but should be correct.

14 years agogallivm: Use a faster (and less accurate) log2 in lod computation.
José Fonseca [Wed, 6 Oct 2010 13:06:14 +0000 (14:06 +0100)]
gallivm: Use a faster (and less accurate) log2 in lod computation.

14 years agogallivm: Take the type signedness in consideration in round/ceil/floor.
José Fonseca [Wed, 6 Oct 2010 11:09:32 +0000 (12:09 +0100)]
gallivm: Take the type signedness in consideration in round/ceil/floor.

14 years agoi965: Fix up IF/ELSE/ENDIF for gen6.
Eric Anholt [Mon, 4 Oct 2010 22:08:03 +0000 (15:08 -0700)]
i965: Fix up IF/ELSE/ENDIF for gen6.

The jump delta is now in the part of the instruction where the
destination fields used to be, and the src args are ignored (or not,
for the new non-predicated IF that we don't use yet).

14 years agoi965: Gen6 no longer has the IFF instruction; always use IF.
Eric Anholt [Mon, 4 Oct 2010 22:09:18 +0000 (15:09 -0700)]
i965: Gen6 no longer has the IFF instruction; always use IF.

14 years agoi965: Add back gen6 headerless FB writes to the new FS backend.
Eric Anholt [Wed, 6 Oct 2010 16:57:55 +0000 (09:57 -0700)]
i965: Add back gen6 headerless FB writes to the new FS backend.

It's not that hard to detect when we need the header.

14 years agor600g: fix dirty state handling
Jerome Glisse [Wed, 6 Oct 2010 16:56:53 +0000 (12:56 -0400)]
r600g: fix dirty state handling

Avoid having object ending up in dead list of dirty object.

Signed-off-by: Jerome Glisse <jglisse@redhat.com>
14 years agoi965: Also do constant propagation for the second operand of CMP.
Eric Anholt [Tue, 5 Oct 2010 17:25:22 +0000 (10:25 -0700)]
i965: Also do constant propagation for the second operand of CMP.

We could do the first operand as well by flipping the comparison, but
this covered several CMPs in code I was looking at.

14 years agoi965: Enable the constant propagation code.
Eric Anholt [Tue, 5 Oct 2010 17:20:16 +0000 (10:20 -0700)]
i965: Enable the constant propagation code.

A debug disable had slipped in.

14 years agor600g: avoid segfault due to unintialized list pointer
Jerome Glisse [Wed, 6 Oct 2010 13:40:27 +0000 (09:40 -0400)]
r600g: avoid segfault due to unintialized list pointer

Signed-off-by: Jerome Glisse <jglisse@redhat.com>
14 years agollvmpipe: Fix sprite coord perspective interpolation of Q.
José Fonseca [Wed, 6 Oct 2010 08:40:51 +0000 (09:40 +0100)]
llvmpipe: Fix sprite coord perspective interpolation of Q.

Q coordinate's coefficients also need to be multiplied by w, otherwise
it will have 1/w, causing problems with TXP.

14 years agollvmpipe: Fix perspective interpolation for point sprites.
José Fonseca [Sun, 26 Sep 2010 10:22:20 +0000 (11:22 +0100)]
llvmpipe: Fix perspective interpolation for point sprites.

Once a fragment is generated with LP_INTERP_PERSPECTIVE set for an input,
it will do a divide by w for that input. Therefore it's not OK to treat LP_INTERP_PERSPECTIVE as
LP_INTERP_LINEAR or vice-versa, even if the attribute is known to not
vary.

A better strategy would be to take the primitive in consideration when
generating the fragment shader key, and therefore avoid the per-fragment
perspective divide.

14 years agollvmpipe: Dump a few missing shader key flags.
José Fonseca [Tue, 5 Oct 2010 09:50:02 +0000 (10:50 +0100)]
llvmpipe: Dump a few missing shader key flags.

14 years agollvmpipe: make debug_fs_variant respect variant->nr_samplers
Keith Whitwell [Sun, 3 Oct 2010 10:39:02 +0000 (11:39 +0100)]
llvmpipe: make debug_fs_variant respect variant->nr_samplers

14 years agoretrace: Handle clear_render_target and clear_depth_stencil.
José Fonseca [Mon, 4 Oct 2010 16:05:03 +0000 (17:05 +0100)]
retrace: Handle clear_render_target and clear_depth_stencil.

14 years agor600g: add evergreen stencil support.
Dave Airlie [Tue, 5 Oct 2010 23:21:16 +0000 (09:21 +1000)]
r600g: add evergreen stencil support.

this sets the stencil up for evergreen properly.

14 years agor600g: userspace fence to avoid kernel call for testing bo busy status
Jerome Glisse [Tue, 5 Oct 2010 20:14:11 +0000 (16:14 -0400)]
r600g: userspace fence to avoid kernel call for testing bo busy status

Signed-off-by: Jerome Glisse <jglisse@redhat.com>
14 years agost/mesa: replace assertion w/ conditional in framebuffer invalidation
Brian Paul [Tue, 5 Oct 2010 20:32:59 +0000 (14:32 -0600)]
st/mesa: replace assertion w/ conditional in framebuffer invalidation

https://bugs.freedesktop.org/show_bug.cgi?id=30632

NOTE: this is a candidate for the 7.9 branch.

14 years agor600g: simplify block relocation
Jerome Glisse [Tue, 5 Oct 2010 19:23:07 +0000 (15:23 -0400)]
r600g: simplify block relocation

Since flush rework there could be only one relocation per
register in a block.

Signed-off-by: Jerome Glisse <jglisse@redhat.com>
14 years agor600g: use dirty list to track dirty blocks
Bas Nieuwenhuizen [Tue, 5 Oct 2010 19:01:43 +0000 (21:01 +0200)]
r600g: use dirty list to track dirty blocks

Got a speed up by tracking the dirty blocks in a seperate list instead of looping through all blocks. This version should work with block that get their dirty state disabled again and I added a dirty check during the flush as some blocks were already dirty.

14 years agodocs: added news item for 7.9 release
Ian Romanick [Tue, 5 Oct 2010 17:07:16 +0000 (10:07 -0700)]
docs: added news item for 7.9 release

Also fix link to release notes in 7.9-rc1 news item.

14 years agodocs: Import news updates from 7.9 branch
Ian Romanick [Mon, 27 Sep 2010 17:17:11 +0000 (10:17 -0700)]
docs: Import news updates from 7.9 branch

Partially cherry-picked from commit 61653b488da76ee1ca4f77363e222d3b717dd865

14 years agodocs: Update mailing lines from sf.net to freedesktop.org
Ian Romanick [Wed, 16 Jun 2010 21:28:08 +0000 (14:28 -0700)]
docs: Update mailing lines from sf.net to freedesktop.org
(cherry picked from commit c19bc5de961fe5e1f8a17131bcfae3dbcccaca29)

14 years agodocs: download.html does not need to be updated for each release
Ian Romanick [Wed, 16 Jun 2010 21:24:46 +0000 (14:24 -0700)]
docs: download.html does not need to be updated for each release
(cherry picked from commit 41e371e351cc4c77b2b20a545af2dfa2dab253d7)

14 years agodocs: Import 7.8.x release notes from 7.8 branch.
Ian Romanick [Tue, 5 Oct 2010 16:55:54 +0000 (09:55 -0700)]
docs: Import 7.8.x release notes from 7.8 branch.

14 years agodocs: Import 7.9 release notes from 7.9 branch.
Ian Romanick [Tue, 5 Oct 2010 16:54:09 +0000 (09:54 -0700)]
docs: Import 7.9 release notes from 7.9 branch.

14 years agonv50: fix always true conditional in shader optimization
Nicolas Kaiser [Tue, 5 Oct 2010 09:26:43 +0000 (11:26 +0200)]
nv50: fix always true conditional in shader optimization

14 years agor600g: improve bo flushing
Jerome Glisse [Tue, 5 Oct 2010 14:29:30 +0000 (10:29 -0400)]
r600g: improve bo flushing

Flush read cache before writting register. Track flushing inside
of a same cs and avoid reflushing same bo if not necessary. Allmost
properly force flush if bo rendered too and then use as a texture
in same cs (missing pipeline flush dunno if it's needed or not).

Signed-off-by: Jerome Glisse <jglisse@redhat.com>
14 years agor600g: store reloc information in bo structure
Jerome Glisse [Tue, 5 Oct 2010 12:42:42 +0000 (08:42 -0400)]
r600g: store reloc information in bo structure

Allow fast lookup of relocation information & id which
was a CPU time consumming operation.

Signed-off-by: Jerome Glisse <jglisse@redhat.com>
14 years agopb: fix numDelayed accounting
Dave Airlie [Tue, 5 Oct 2010 09:08:41 +0000 (19:08 +1000)]
pb: fix numDelayed accounting

we weren't decreasing when removing from the list.

14 years agor600g: avoid unneeded bo wait
Dave Airlie [Tue, 5 Oct 2010 06:00:48 +0000 (16:00 +1000)]
r600g: avoid unneeded bo wait

if we know the bo has gone not busy, no need to add another bo wait

thanks to Andre (taiu) on irc for pointing this out.

14 years agor600g: drop use_mem_constant.
Dave Airlie [Tue, 5 Oct 2010 06:00:23 +0000 (16:00 +1000)]
r600g: drop use_mem_constant.

since we plan on using dx10 constant buffers everywhere.

14 years agor600g: drop mman allocator
Dave Airlie [Tue, 5 Oct 2010 05:57:57 +0000 (15:57 +1000)]
r600g: drop mman allocator

we don't use this since constant buffers are now being used on all gpus.

14 years agor600g: add bo busy backoff.
Dave Airlie [Tue, 5 Oct 2010 05:51:38 +0000 (15:51 +1000)]
r600g: add bo busy backoff.

When we go to do a lot of bos in one draw like constant bufs we need
to avoid bouncing off the busy ioctl, this mitigates by backing off
on busy bos for a short amount of times.

14 years agopb: don't keep checking buffers after first busy
Dave Airlie [Tue, 5 Oct 2010 05:50:58 +0000 (15:50 +1000)]
pb: don't keep checking buffers after first busy

If we assume busy buffers are added to the list in order its unlikely
we'd fine one after the first busy one that isn't busy.

14 years agor600g: add bo fenced list.
Dave Airlie [Tue, 5 Oct 2010 05:35:52 +0000 (15:35 +1000)]
r600g: add bo fenced list.

this just keeps a list of bos submitted together, and uses them to decide
bo busy state for the whole group.

14 years agoswrast: fix choose_depth_texture_level() to respect mipmap filtering state
Brian Paul [Tue, 5 Oct 2010 01:59:23 +0000 (19:59 -0600)]
swrast: fix choose_depth_texture_level() to respect mipmap filtering state

NOTE: this is a candidate for the 7.9 branch.

14 years agor300g: fix microtiling for 16-bits-per-channel formats
Marek Olšák [Mon, 4 Oct 2010 19:19:27 +0000 (21:19 +0200)]
r300g: fix microtiling for 16-bits-per-channel formats

These texture formats (like R16G16B16A16_UNORM) were untested until now
because st/mesa doesn't use them. I am testing this with a hacked st/mesa
here.

14 years agoupdate release notes for Gallium
Marek Olšák [Tue, 5 Oct 2010 00:56:14 +0000 (02:56 +0200)]
update release notes for Gallium

I am trying to be exhaustive, but still I might have missed tons of other
changes to Gallium.
(cherry picked from commit 968a9ec76eadf55e8b58171884e1175d7b8cf59a)

Conflicts:

docs/relnotes-7.9.html

14 years agodocs: Add list of bugs fixed in 7.9
Ian Romanick [Mon, 4 Oct 2010 23:35:09 +0000 (16:35 -0700)]
docs: Add list of bugs fixed in 7.9

14 years agoi965: Add support for gen6 FB writes to the new FS.
Eric Anholt [Mon, 4 Oct 2010 22:07:17 +0000 (15:07 -0700)]
i965: Add support for gen6 FB writes to the new FS.

This uses message headers for now, since we'll need it for MRT.  We
can cut out the header later.

14 years agoi965: In disasm, gen6 fb writes don't put msg reg # in destreg_conditionalmod.
Eric Anholt [Mon, 4 Oct 2010 22:03:32 +0000 (15:03 -0700)]
i965: In disasm, gen6 fb writes don't put msg reg # in destreg_conditionalmod.

It instead sensibly appears in the src0 slot.

14 years agoi965: Add initial folding of constants into operand immediate slots.
Eric Anholt [Mon, 4 Oct 2010 18:48:04 +0000 (11:48 -0700)]
i965: Add initial folding of constants into operand immediate slots.

We could try to detect this in expression handling and do it
proactively there, but it seems like less logic to do it in one
optional pass at the end.

14 years agoi965: Add trivial dead code elimination in the new FS backend.
Eric Anholt [Sun, 3 Oct 2010 22:15:18 +0000 (15:15 -0700)]
i965: Add trivial dead code elimination in the new FS backend.

The glsl core should be handling most dead code issues for us, but we
generate some things in codegen that may not get used, like the 1/w
value or pixel deltas.  It seems a lot easier this way than trying to
work out up front whether we're going to use those values or not.

14 years agoi965: Be more conservative on live interval calculation.
Eric Anholt [Sun, 3 Oct 2010 22:01:20 +0000 (15:01 -0700)]
i965: Be more conservative on live interval calculation.

This also means that our intervals now highlight dead code.

14 years agor600g: Fix SCons build.
Vinson Lee [Mon, 4 Oct 2010 22:56:55 +0000 (15:56 -0700)]
r600g: Fix SCons build.

14 years agor600g: remove dead label & fix indentation
Jerome Glisse [Mon, 4 Oct 2010 14:40:07 +0000 (10:40 -0400)]
r600g: remove dead label & fix indentation

Signed-off-by: Jerome Glisse <jglisse@redhat.com>
14 years agor600g: rename radeon_ws_bo to r600_bo
Jerome Glisse [Mon, 4 Oct 2010 14:38:50 +0000 (10:38 -0400)]
r600g: rename radeon_ws_bo to r600_bo

Signed-off-by: Jerome Glisse <jglisse@redhat.com>
14 years agor600g: use r600_bo for relocation argument, simplify code
Jerome Glisse [Mon, 4 Oct 2010 14:37:32 +0000 (10:37 -0400)]
r600g: use r600_bo for relocation argument, simplify code

Signed-off-by: Jerome Glisse <jglisse@redhat.com>
14 years agor600g: allow r600_bo to be a sub allocation of a big bo
Jerome Glisse [Mon, 4 Oct 2010 14:25:23 +0000 (10:25 -0400)]
r600g: allow r600_bo to be a sub allocation of a big bo

Add bo offset everywhere needed if r600_bo is ever a sub bo
of a bigger bo.

Signed-off-by: Jerome Glisse <jglisse@redhat.com>
14 years agor600g: rename radeon_ws_bo to r600_bo
Jerome Glisse [Mon, 4 Oct 2010 14:06:13 +0000 (10:06 -0400)]
r600g: rename radeon_ws_bo to r600_bo

Signed-off-by: Jerome Glisse <jglisse@redhat.com>
14 years agonvfx: Pair os_malloc_aligned() with os_free_aligned().
Krzysztof Smiechowicz [Mon, 4 Oct 2010 18:43:29 +0000 (11:43 -0700)]
nvfx: Pair os_malloc_aligned() with os_free_aligned().

From AROS.

14 years agor600g: TODO domain management
Dave Airlie [Mon, 4 Oct 2010 06:41:49 +0000 (16:41 +1000)]
r600g: TODO domain management

no wonder it was slow, the code is deliberately forcing stuff into GTT,
we used to have domain management but it seems to have disappeared.

14 years agor600g: fix wwarning in bo_map function
Dave Airlie [Mon, 4 Oct 2010 06:26:46 +0000 (16:26 +1000)]
r600g: fix wwarning in bo_map function

14 years agor600g: the code to check whether a new vertex shader is needed was wrong
Dave Airlie [Mon, 4 Oct 2010 06:24:59 +0000 (16:24 +1000)]
r600g: the code to check whether a new vertex shader is needed was wrong

this code was memcmp'ing two structs, but refcounting one of them afterwards,
so any subsequent memcmp was never going to work.

again this stops unnecessary uploads of vertex program,

14 years agor600g: break out of search for reloc bo after finding it.
Dave Airlie [Mon, 4 Oct 2010 05:58:39 +0000 (15:58 +1000)]
r600g: break out of search for reloc bo after finding it.

this function was taking quite a lot of pointless CPU.

14 years agoi965: Fix glean/texSwizzle regression in previous commit.
Eric Anholt [Sun, 3 Oct 2010 07:24:09 +0000 (00:24 -0700)]
i965: Fix glean/texSwizzle regression in previous commit.

Easy enough patch, who needs a full test run.  Oh, that's right.  Me.

14 years agoi965: Set up swizzling of shadow compare results for GL_DEPTH_TEXTURE_MODE.
Eric Anholt [Sun, 3 Oct 2010 06:27:31 +0000 (23:27 -0700)]
i965: Set up swizzling of shadow compare results for GL_DEPTH_TEXTURE_MODE.

The brw_wm_surface_state.c handling of GL_DEPTH_TEXTURE_MODE doesn't
apply to shadow compares, which always return an intensity value.  The
texture swizzles can do the job for us.

Fixes:
glsl1-shadow2D(): 1
glsl1-shadow2D(): 3

14 years agoi965: Add support for EXT_texture_swizzle to the new FS backend.
Eric Anholt [Sun, 3 Oct 2010 06:44:29 +0000 (23:44 -0700)]
i965: Add support for EXT_texture_swizzle to the new FS backend.

14 years agor300g: add support for L8A8 colorbuffers
Marek Olšák [Sat, 2 Oct 2010 21:13:12 +0000 (23:13 +0200)]
r300g: add support for L8A8 colorbuffers

Blending with DST_ALPHA is undefined. SRC_ALPHA works, though.
I bet some other formats have similar limitations too.

14 years agor300g: add support for R8G8 colorbuffers
Marek Olšák [Sat, 2 Oct 2010 19:42:22 +0000 (21:42 +0200)]
r300g: add support for R8G8 colorbuffers

The hw swizzles have been obtained by a brute force approach,
and only C0 and C2 are stored in UV88, the other channels are
ignored.

R16G16 is going to be a lot trickier.

14 years agomesa/st: initial attempt at RG support for gallium drivers
Dave Airlie [Wed, 11 Aug 2010 09:04:05 +0000 (19:04 +1000)]
mesa/st: initial attempt at RG support for gallium drivers

passes all piglit RG tests with softpipe.

14 years agoi965: Fix incorrect batchbuffer size in gen6 clip state command.
Kenneth Graunke [Sat, 2 Oct 2010 02:53:24 +0000 (19:53 -0700)]
i965: Fix incorrect batchbuffer size in gen6 clip state command.

FORCE_ZERO_RTAINDEX should be in the fourth (and final) dword.

14 years agoi965: Don't try to emit code if we failed register allocation.
Eric Anholt [Sat, 2 Oct 2010 00:18:07 +0000 (17:18 -0700)]
i965: Don't try to emit code if we failed register allocation.

14 years agoi965: Fix off-by-ones in handling the last members of register classes.
Eric Anholt [Fri, 1 Oct 2010 23:36:17 +0000 (16:36 -0700)]
i965: Fix off-by-ones in handling the last members of register classes.

Luckily, one of them would result in failing out register allocation
when the other bugs were encountered.  Applies to
glsl-fs-vec4-indexing-temp-dst-in-nested-loop-combined, which still
fails register allocation, but now legitimately.

14 years agoi965: Add a sanity check for register allocation sizes.
Eric Anholt [Fri, 1 Oct 2010 23:35:55 +0000 (16:35 -0700)]
i965: Add a sanity check for register allocation sizes.

14 years agoi965: When producing a single channel swizzle, don't make a temporary.
Eric Anholt [Fri, 1 Oct 2010 22:42:39 +0000 (15:42 -0700)]
i965: When producing a single channel swizzle, don't make a temporary.

This quickly cuts 8% of the instructions in my glsl demo.

14 years agoi965: Restore the forcing of aligned pairs for delta_xy on chips with PLN.
Eric Anholt [Fri, 1 Oct 2010 22:11:09 +0000 (15:11 -0700)]
i965: Restore the forcing of aligned pairs for delta_xy on chips with PLN.

By doing so using the register allocator now, we avoid wasting a
register to make the alignment happen.

14 years agor600c: fix segfault in evergreen stencil code
Alex Deucher [Sat, 2 Oct 2010 00:13:36 +0000 (20:13 -0400)]
r600c: fix segfault in evergreen stencil code

Fixes:
https://bugs.freedesktop.org/show_bug.cgi?id=30551