mesa.git
5 years agomesa: add EXT_dsa Generate*MipmapEXT functions
Pierre-Eric Pelloux-Prayer [Wed, 11 Sep 2019 08:01:24 +0000 (10:01 +0200)]
mesa: add EXT_dsa Generate*MipmapEXT functions

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
5 years agomesa: refactor GenerateTextureMipmap handling
Pierre-Eric Pelloux-Prayer [Wed, 11 Sep 2019 07:58:47 +0000 (09:58 +0200)]
mesa: refactor GenerateTextureMipmap handling

Rework _mesa_GenerateTextureMipmap to allow code sharing with EXT_dsa functions.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
5 years agomesa: add EXT_dsa glGetFloati_vEXT/glGetDoublei_vEXT
Pierre-Eric Pelloux-Prayer [Wed, 11 Sep 2019 07:30:14 +0000 (09:30 +0200)]
mesa: add EXT_dsa glGetFloati_vEXT/glGetDoublei_vEXT

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
5 years agomesa: add EXT_dsa + EXT_gpu_program_parameters functions
Pierre-Eric Pelloux-Prayer [Mon, 9 Sep 2019 15:26:30 +0000 (17:26 +0200)]
mesa: add EXT_dsa + EXT_gpu_program_parameters functions

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
5 years agomesa: add EXT_dsa + EXT_gpu_shader4 functions
Pierre-Eric Pelloux-Prayer [Mon, 9 Sep 2019 15:14:18 +0000 (17:14 +0200)]
mesa: add EXT_dsa + EXT_gpu_shader4 functions

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
5 years agomesa: add EXT_dsa + EXT_texture_integer functions
Pierre-Eric Pelloux-Prayer [Mon, 9 Sep 2019 14:44:11 +0000 (16:44 +0200)]
mesa: add EXT_dsa + EXT_texture_integer functions

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
5 years agomesa: add EXT_dsa + EXT_texture_buffer_object functions
Pierre-Eric Pelloux-Prayer [Mon, 9 Sep 2019 14:22:29 +0000 (16:22 +0200)]
mesa: add EXT_dsa + EXT_texture_buffer_object functions

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
5 years agomesa: add EXT_dsa glProgramUniform*EXT functions
Pierre-Eric Pelloux-Prayer [Mon, 9 Sep 2019 13:53:00 +0000 (15:53 +0200)]
mesa: add EXT_dsa glProgramUniform*EXT functions

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
5 years agomesa: add EXT_dsa NamedProgram functions
Pierre-Eric Pelloux-Prayer [Tue, 28 May 2019 15:06:00 +0000 (17:06 +0200)]
mesa: add EXT_dsa NamedProgram functions

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
5 years agomesa: add EXT_dsa glClientAttribDefaultEXT / glPushClientAttribDefaultEXT
Pierre-Eric Pelloux-Prayer [Tue, 28 May 2019 08:27:52 +0000 (10:27 +0200)]
mesa: add EXT_dsa glClientAttribDefaultEXT / glPushClientAttribDefaultEXT

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
5 years agomesa: add EXT_dsa glNamedRenderbufferStorageEXT and glGetNamedRenderbufferParameterivEXT
Pierre-Eric Pelloux-Prayer [Thu, 23 May 2019 14:34:16 +0000 (16:34 +0200)]
mesa: add EXT_dsa glNamedRenderbufferStorageEXT and glGetNamedRenderbufferParameterivEXT

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
5 years agopanfrost: Respect offset for imported resources
Daniel Stone [Thu, 17 Oct 2019 11:49:54 +0000 (13:49 +0200)]
panfrost: Respect offset for imported resources

When we import a resource through Gallium, we need to take account of
the offset parameter passed.

Fixes a failure seen with the VIVID V4L2 driver, which would create NV12
resources within the same BO, with an offset. Sample pipeline to
reproduce (replace videoN with your actual VIVID device node):
    gst-launch-1.0 v4l2src device=/dev/videoN ! video/x-raw,format=NV12 ! glimagesink

Signed-off-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reported-by: Nicolas Dufresne <nicolas.dufresne@collabora.com>
Tested-by: Nicolas Dufresne <nicolas.dufresne@collabora.com>
5 years agoiris/resource: Use isl surface alignment during bo allocation
Jordan Justen [Fri, 31 May 2019 22:50:53 +0000 (15:50 -0700)]
iris/resource: Use isl surface alignment during bo allocation

Reworks:
 * Change subject from "iris: Align main surface allocation to 64k on gen12+"
 * Make use of isl surf alignment. (Nanley)

Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
5 years agointel/isl: Add isl_aux_usage_has_ccs
Jason Ekstrand [Fri, 4 May 2018 16:43:01 +0000 (09:43 -0700)]
intel/isl: Add isl_aux_usage_has_ccs

Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
5 years agointel/isl: Add R10G10B10_FLOAT_A2_UNORM format
Jordan Justen [Thu, 12 Apr 2018 05:48:33 +0000 (22:48 -0700)]
intel/isl: Add R10G10B10_FLOAT_A2_UNORM format

Reworks:
 * Fill out the format's entry in the ISL format table. (Nanley)
 * Support CCS_E-enabled BLORP copies with the format. (Nanley)

Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Sagar Ghuge <sagar.ghuge@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
5 years agointel/compiler: Report the number of non-spill/fill SEND messages
Kenneth Graunke [Tue, 10 Sep 2019 01:31:41 +0000 (18:31 -0700)]
intel/compiler: Report the number of non-spill/fill SEND messages

This can be useful to measure whether memory access optimizations are
having the desired effect.  For example, we might see a reduction in
image loads/stores, or constant buffer loads.  We can already see this
in cycle estimates to some extent, but this is a more direct approach,
minus a lot of the noise of random scheduler shuffling.

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
5 years agost/mesa: don't call variables "tgsi" when they can reference NIR
Marek Olšák [Thu, 17 Oct 2019 00:21:11 +0000 (20:21 -0400)]
st/mesa: don't call variables "tgsi" when they can reference NIR

Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
5 years agost/mesa: merge st_fragment_program into st_common_program
Marek Olšák [Wed, 16 Oct 2019 20:46:19 +0000 (16:46 -0400)]
st/mesa: merge st_fragment_program into st_common_program

Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
5 years agost/mesa: remove redundant function st_reference_compprog
Marek Olšák [Thu, 17 Oct 2019 18:51:23 +0000 (14:51 -0400)]
st/mesa: remove redundant function st_reference_compprog

Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
5 years agost/mesa: remove unused st_xxx_program::sha1
Marek Olšák [Wed, 16 Oct 2019 20:22:43 +0000 (16:22 -0400)]
st/mesa: remove unused st_xxx_program::sha1

Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
5 years agost/mesa: remove st_vp_variant_key in favor of st_common_variant_key
Marek Olšák [Wed, 16 Oct 2019 20:19:09 +0000 (16:19 -0400)]
st/mesa: remove st_vp_variant_key in favor of st_common_variant_key

Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
5 years agost/mesa: remove num_tgsi_tokens from st_xx_program
Marek Olšák [Wed, 16 Oct 2019 20:10:43 +0000 (16:10 -0400)]
st/mesa: remove num_tgsi_tokens from st_xx_program

Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
5 years agost/mesa: rename basic -> common for st_common_program
Marek Olšák [Wed, 16 Oct 2019 20:05:10 +0000 (16:05 -0400)]
st/mesa: rename basic -> common for st_common_program

Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
5 years agost/mesa: rename st_xxx_program::tgsi to state
Marek Olšák [Wed, 16 Oct 2019 19:59:41 +0000 (15:59 -0400)]
st/mesa: rename st_xxx_program::tgsi to state

Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
5 years agost/mesa: lower doubles for NIR after linking
Marek Olšák [Tue, 15 Oct 2019 20:49:06 +0000 (16:49 -0400)]
st/mesa: lower doubles for NIR after linking

This allows dropping 1 call to st_nir_opts, because shaders are always
optimized after linking.

Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
5 years agost/mesa: call st_nir_opts for linked shaders only once
Marek Olšák [Tue, 8 Oct 2019 02:15:01 +0000 (22:15 -0400)]
st/mesa: call st_nir_opts for linked shaders only once

The removed st_nir_opts calls are mostly redundant.

There is an improvement with shader-db on radeonsi:

Before:
    real 1m54.047s
    user 28m37.857s
    sys  0m7.573s

After:
    real 1m52.012s
    user 28m3.412s
    sys  0m7.808s

Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
5 years agointel/vec4: Don't try both sources as immediates for DPH
Ian Romanick [Mon, 15 Jul 2019 18:38:55 +0000 (11:38 -0700)]
intel/vec4: Don't try both sources as immediates for DPH

DPH isn't actually commutative, so this doesn't work.  If the immediate
in src0 would be a VF candidate, we could do better. *shrug*

No shader-db changes on any Intel platform.

Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Fixes: b04beaf41d2 ("intel/vec4: Try both sources as candidates for being immediates")
5 years agonir/search: Fix possible NULL dereference in is_fsign
Ian Romanick [Mon, 15 Jul 2019 22:18:47 +0000 (15:18 -0700)]
nir/search: Fix possible NULL dereference in is_fsign

Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Fixes: 09705747d72 ("nir/algebraic: Reassociate fadd into fmul in DPH-like pattern")
5 years agoiris: Let isl decide the supported tiling in more situations
Jordan Justen [Fri, 19 Apr 2019 23:28:01 +0000 (16:28 -0700)]
iris: Let isl decide the supported tiling in more situations

Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Suggested-by: Nanley Chery <nanley.g.chery@intel.com>
Reviewed-by: Nanley Chery <nanley.g.chery@intel.com>
5 years agointel/isl: Add gen12 depth/stencil surface alignments
Jordan Justen [Fri, 19 Jan 2018 08:48:33 +0000 (00:48 -0800)]
intel/isl: Add gen12 depth/stencil surface alignments

Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Nanley Chery <nanley.g.chery@intel.com>
5 years agointel/isl: Select Y-tiling for stencil on gen12
Jason Ekstrand [Tue, 9 Jan 2018 00:28:46 +0000 (16:28 -0800)]
intel/isl: Select Y-tiling for stencil on gen12

Rework:
 * Disallow linear 1D stencil buffers (Nanley)
 * Force Y for gen12 stencil rather than ~W (Nanley)

Co-authored-by: Nanley Chery <nanley.g.chery@intel.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Nanley Chery <nanley.g.chery@intel.com>
5 years agointel/genxml: Remove W-tiling on gen12
Jason Ekstrand [Tue, 9 Jan 2018 00:27:45 +0000 (16:27 -0800)]
intel/genxml: Remove W-tiling on gen12

It's no longer supported by the hardware

Reviewed-by: Nanley Chery <nanley.g.chery@intel.com>
5 years agointel/genxml,isl: Add gen12 stencil buffer changes
Jordan Justen [Wed, 16 Aug 2017 23:45:47 +0000 (16:45 -0700)]
intel/genxml,isl: Add gen12 stencil buffer changes

Rework:
 * NULL stencil buffer path (Jason)
 * genxml fixes (Nanley)

Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Nanley Chery <nanley.g.chery@intel.com>
5 years agointel/genxml,isl: Add gen12 depth buffer changes
Jordan Justen [Wed, 16 Aug 2017 23:45:47 +0000 (16:45 -0700)]
intel/genxml,isl: Add gen12 depth buffer changes

Reworks:
 * Fix 3DSTATE_DEPTH_BUFFER "Surface Format" end in xml (Jason)
 * Remove WM_HZ_OP changes (Nanley)

Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Nanley Chery <nanley.g.chery@intel.com>
5 years agointel/genxml,isl: Add gen12 render surface state changes
Jordan Justen [Wed, 16 Aug 2017 23:45:47 +0000 (16:45 -0700)]
intel/genxml,isl: Add gen12 render surface state changes

Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Nanley Chery <nanley.g.chery@intel.com>
5 years agomesa: Refactor the entirety of _mesa_format_matches_format_and_type().
Eric Anholt [Mon, 19 Aug 2019 22:17:58 +0000 (15:17 -0700)]
mesa: Refactor the entirety of _mesa_format_matches_format_and_type().

This function was difficult to implement for new formats due to the
combination of endianness and swapbytes support.  Since it's mostly
used for fast paths, bugs in it were often missed during testing.

Just reimplement it on top of the recent
_mesa_format_from_format_and_type() which can give us a canonical
MESA_FORMAT for a format and type enum (while respecting endianness).

Fixes:
- R4G4B4A4_UNORM, B4G4R4_UINT, R4G4B4A4_UINT incorrectly matched with
  swapBytes (you can't just reverse the channels if the channels
  aren't bytes)
- A4R4G4B4_UNORM and A4R4G4B4_UINT missing BGRA/4444_REV matches
- failing to match RGB/BGR unorm8 array formats on BE
2101010 formats incorrectly matching with swapBytes set.
- UINT/SINT byte formats failed to match with swapBytes set.

This deletes the part of tests/mesa_formats.cpp that called
_mesa_format_matches_format_and_type() to make sure it didn't
assertion fail, as it now would assertion fail due to the fact that we
were passing an invalid format (GL_RG) for most types.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
5 years agomesa: Add support for array formats of depth and stencil.
Eric Anholt [Wed, 18 Sep 2019 21:01:29 +0000 (14:01 -0700)]
mesa: Add support for array formats of depth and stencil.

In desktop GL, you can specify things like GL_DEPTH_COMPONENT/GL_BYTE as a
ReadPixels format, and we need to be able to represent that to see if we
have proper MESA_FORMATs for them.  That's exactly what the
mesa_array_format enum is for.

v2: Drop _mesa from static fn.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
5 years agomesa: Add format/type matching for DEPTH/UINT_24_8.
Eric Anholt [Wed, 18 Sep 2019 21:27:51 +0000 (14:27 -0700)]
mesa: Add format/type matching for DEPTH/UINT_24_8.

We had missed this case where GLES3 allows glReadPixels(DEPTH, UINT_24_8),
and just got lucky by the readpixels path never asking for the matching
format from this function.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
5 years agomesa: Fix depth/stencil ordering in _mesa_format_from_format_and_type().
Eric Anholt [Tue, 17 Sep 2019 21:54:48 +0000 (14:54 -0700)]
mesa: Fix depth/stencil ordering in _mesa_format_from_format_and_type().

The GL spec says the 24-bit component is in the high bits, and
format_unpack.c looks at the high 24 bits in the S8Z24 case, not
Z24SS8.

Avoids a regression in the next commit.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
5 years agomesa: Add debug info to _mesa_format_from_format_and_type() error path.
Eric Anholt [Wed, 18 Sep 2019 21:46:49 +0000 (14:46 -0700)]
mesa: Add debug info to _mesa_format_from_format_and_type() error path.

The unreachable() that follows isn't very useful for debug, and by adding
this here we get a nice description of the failure in debug builds.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
5 years agofreedreno/a6xx: Turn on geometry shaders
Kristian H. Kristensen [Fri, 11 Oct 2019 03:54:28 +0000 (20:54 -0700)]
freedreno/a6xx: Turn on geometry shaders

Signed-off-by: Kristian H. Kristensen <hoegsberg@google.com>
5 years agofreedreno/ci: Add failing tests to skip list
Kristian H. Kristensen [Mon, 14 Oct 2019 02:32:37 +0000 (19:32 -0700)]
freedreno/ci: Add failing tests to skip list

Some queries are still failing and layered rending needs more work.

Signed-off-by: Kristian H. Kristensen <hoegsberg@google.com>
5 years agofreedreno/a6xx: Implement PIPE_QUERY_PRIMITIVES_GENERATED for GS
Kristian H. Kristensen [Wed, 16 Oct 2019 19:08:19 +0000 (12:08 -0700)]
freedreno/a6xx: Implement PIPE_QUERY_PRIMITIVES_GENERATED for GS

When we don't have streamout enabled, we have to read this register to
get the number of primitives emitted.

Signed-off-by: Kristian H. Kristensen <hoegsberg@google.com>
5 years agofreedreno/blitter: Save GS state
Kristian H. Kristensen [Fri, 11 Oct 2019 22:15:23 +0000 (15:15 -0700)]
freedreno/blitter: Save GS state

We have GS state now.

Signed-off-by: Kristian H. Kristensen <hoegsberg@google.com>
5 years agost/mesa: Also enable GS when ESSLVersion > 320
Kristian H. Kristensen [Thu, 10 Oct 2019 18:27:47 +0000 (11:27 -0700)]
st/mesa: Also enable GS when ESSLVersion > 320

Signed-off-by: Kristian H. Kristensen <hoegsberg@google.com>
5 years agofreedreno/a6xx: Support layered render targets
Kristian H. Kristensen [Fri, 11 Oct 2019 20:56:20 +0000 (13:56 -0700)]
freedreno/a6xx: Support layered render targets

Signed-off-by: Kristian H. Kristensen <hoegsberg@google.com>
5 years agofreedreno/a6xx: Emit program state for GS
Kristian H. Kristensen [Fri, 11 Oct 2019 20:43:53 +0000 (13:43 -0700)]
freedreno/a6xx: Emit program state for GS

Signed-off-by: Kristian H. Kristensen <hoegsberg@google.com>
5 years agofreedreno/ir3: End VS with CHMASK and CHSH in GS pipelines
Kristian H. Kristensen [Fri, 11 Oct 2019 19:37:38 +0000 (12:37 -0700)]
freedreno/ir3: End VS with CHMASK and CHSH in GS pipelines

When used in a GS pipeline, the VS doesn't end with the END
instruction. Instead it chains to the GS, which continues running with
the same register allocation.  The intended use cases seems to be that
you can compile a regular VS (ie outputs in registers and ending with
END) but then tack on link-time generated code past the END to write
the outputs using STLW, in case the VS is used with GS.

Signed-off-by: Kristian H. Kristensen <hoegsberg@google.com>
5 years agofreedreno/ir3: Start GS with (ss) and (sy)
Kristian H. Kristensen [Fri, 11 Oct 2019 19:36:49 +0000 (12:36 -0700)]
freedreno/ir3: Start GS with (ss) and (sy)

We don't know what kind of loads we might have to wait on when coming
in from chsh in the VS so set both sync flags.

Signed-off-by: Kristian H. Kristensen <hoegsberg@google.com>
5 years agofreedreno/ir3: Pre-color GS header and primitive ID
Kristian H. Kristensen [Fri, 11 Oct 2019 19:34:54 +0000 (12:34 -0700)]
freedreno/ir3: Pre-color GS header and primitive ID

These sysvals have to be unclobbered by VS and in the same registers
in both VS and GS, since the chsh from VS to GS doesn't reload the
values. We use the pre-color argument to ir3_ra() to always place
these values in r0.x and r0.y.

Signed-off-by: Kristian H. Kristensen <hoegsberg@google.com>
5 years agofreedreno/ir3: Setup ir3 inputs and outputs for GS
Kristian H. Kristensen [Fri, 11 Oct 2019 19:24:12 +0000 (12:24 -0700)]
freedreno/ir3: Setup ir3 inputs and outputs for GS

Inputs are the GS header, which contains vertex ID, local primitive ID
and thread ID as well as primitive ID. The setup is a little different
from other sysvals, since we always have to receive them in the VS so
that it can pass them on into the GS.

The vertex flag outputs from GS is set up as a proper nir output in
the lowering pass and doesn't need special handling here.

Signed-off-by: Kristian H. Kristensen <hoegsberg@google.com>
5 years agofreedreno/ir3: Implement primitive layout intrinsics
Kristian H. Kristensen [Fri, 11 Oct 2019 04:02:45 +0000 (21:02 -0700)]
freedreno/ir3: Implement primitive layout intrinsics

This implements the load_vs_primitive_stride_ir3,
load_vs_vertex_stride_ir3 and load_primitive_location_ir3 intrinsics,
used for getting the primitive layout strides and locations.

Signed-off-by: Kristian H. Kristensen <hoegsberg@google.com>
5 years agofreedreno/ir3: Implement lowering passes for VS and GS
Kristian H. Kristensen [Fri, 11 Oct 2019 00:17:10 +0000 (17:17 -0700)]
freedreno/ir3: Implement lowering passes for VS and GS

This introduces two new lowering passes. One to lower VS to explicit
outputs using STLW and one to lower GS to load input using LDLW and
implement the GS specific functionality.

Signed-off-by: Kristian H. Kristensen <hoegsberg@google.com>
5 years agofreedreno/ir3: Add has_gs flag to shader key
Kristian H. Kristensen [Thu, 10 Oct 2019 22:37:19 +0000 (15:37 -0700)]
freedreno/ir3: Add has_gs flag to shader key

Since the presence of GS changes how the VS operates we need to track
that in the shader key.

Signed-off-by: Kristian H. Kristensen <hoegsberg@google.com>
5 years agofreedreno/a6xx: Add missing adjacency primitives to table
Kristian H. Kristensen [Thu, 10 Oct 2019 22:24:10 +0000 (15:24 -0700)]
freedreno/a6xx: Add missing adjacency primitives to table

Signed-off-by: Kristian H. Kristensen <hoegsberg@google.com>
5 years agofreedreno/ir3: Add intrinsics that map to LDLW/STLW
Kristian H. Kristensen [Thu, 10 Oct 2019 22:15:37 +0000 (15:15 -0700)]
freedreno/ir3: Add intrinsics that map to LDLW/STLW

These intrinsics will let us do all the offset calculations in nir,
which is nicer to work with and lets nir_opt_algebraic eat it all up.

Signed-off-by: Kristian H. Kristensen <hoegsberg@google.com>
5 years agofreedreno/ir3: Add new LDLW/STLW instructions
Kristian H. Kristensen [Thu, 10 Oct 2019 22:09:49 +0000 (15:09 -0700)]
freedreno/ir3: Add new LDLW/STLW instructions

These access memory used for passing data between geometry stages.

Signed-off-by: Kristian H. Kristensen <hoegsberg@google.com>
5 years agofreedreno/ir3: Extend RA with mechanism for pre-coloring registers
Kristian H. Kristensen [Thu, 10 Oct 2019 21:43:03 +0000 (14:43 -0700)]
freedreno/ir3: Extend RA with mechanism for pre-coloring registers

We'll need to pre-color certain input registers betwee VS and GS
shaders.

Signed-off-by: Kristian H. Kristensen <hoegsberg@google.com>
5 years agofreedreno/ir3: Use third register for offset for LDL and LDLV
Kristian H. Kristensen [Thu, 10 Oct 2019 20:44:14 +0000 (13:44 -0700)]
freedreno/ir3: Use third register for offset for LDL and LDLV

Before, offset held the offset, which can be either immediate or a
register.  Use a third register to hold the offset so that we can use
a register.

Signed-off-by: Kristian H. Kristensen <hoegsberg@google.com>
5 years agofreedreno/ir3: Add support for CHSH and CHMASK instructions
Kristian H. Kristensen [Thu, 10 Oct 2019 20:21:25 +0000 (13:21 -0700)]
freedreno/ir3: Add support for CHSH and CHMASK instructions

Just add the constructors for now and special case similar to END so
we don't remove them.

Signed-off-by: Kristian H. Kristensen <hoegsberg@google.com>
5 years agofreedreno/a6xx: Trim a few regs from fd6_emit_restore()
Kristian H. Kristensen [Fri, 11 Oct 2019 19:17:54 +0000 (12:17 -0700)]
freedreno/a6xx: Trim a few regs from fd6_emit_restore()

We know what these do an either write them in the program stateobj or
don't need to write them.

Signed-off-by: Kristian H. Kristensen <hoegsberg@google.com>
5 years agofreedreno/registers: Update with GS, HS and DS registers
Kristian H. Kristensen [Thu, 10 Oct 2019 22:21:25 +0000 (15:21 -0700)]
freedreno/registers: Update with GS, HS and DS registers

Signed-off-by: Kristian H. Kristensen <hoegsberg@google.com>
5 years agofreedreno/ci: Ban texsubimage2d_pbo.r16ui_2d, due to two flakes reported.
Eric Anholt [Sun, 6 Oct 2019 02:46:38 +0000 (19:46 -0700)]
freedreno/ci: Ban texsubimage2d_pbo.r16ui_2d, due to two flakes reported.

Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Daniel Stone <daniels@collabora.com>
5 years agost/mesa: silence a warning in st_nir_lower_tex_src_plane
Marek Olšák [Thu, 17 Oct 2019 20:07:26 +0000 (16:07 -0400)]
st/mesa: silence a warning in st_nir_lower_tex_src_plane

trivial

5 years agogallium/u_blitter: remove an unused variable
Marek Olšák [Thu, 17 Oct 2019 20:07:02 +0000 (16:07 -0400)]
gallium/u_blitter: remove an unused variable

trivial

5 years agoradeonsi: recreate aux_context after a GPU reset
Marek Olšák [Wed, 16 Oct 2019 21:22:20 +0000 (17:22 -0400)]
radeonsi: recreate aux_context after a GPU reset

Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
5 years agoradeonsi: call the reset callback if get_device_reset_status returns a failure
Marek Olšák [Wed, 16 Oct 2019 21:09:29 +0000 (17:09 -0400)]
radeonsi: call the reset callback if get_device_reset_status returns a failure

Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
5 years agost/mesa: call the reset callback if glGetGraphicsResetStatus returns a failure
Marek Olšák [Wed, 16 Oct 2019 21:12:43 +0000 (17:12 -0400)]
st/mesa: call the reset callback if glGetGraphicsResetStatus returns a failure

so that we immediately set the no-op dispatch

Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
5 years agointel/fs/gen12: Add tests for scoreboard pass
Caio Marcelo de Oliveira Filho [Fri, 4 Oct 2019 19:30:08 +0000 (12:30 -0700)]
intel/fs/gen12: Add tests for scoreboard pass

Tests the combinations of cases of RAW, WAW and WAR hazards involving
both inorder and outoforder instructions.  Also tests that
dependencies combine and propagate correctly through control
flow (loops and conditionals).

v2: Add an extra test illustrating that the non-logical CFG edge
    between then-block and else-block is being taking into
    account.  (Curro)

Reviewed-by: Francisco Jerez <currojerez@riseup.net>
5 years agoaco: don't combine minmax3 if there is a neg or abs modifier in between
Daniel Schürmann [Thu, 17 Oct 2019 13:06:48 +0000 (15:06 +0200)]
aco: don't combine minmax3 if there is a neg or abs modifier in between

This fixes a graphical corruption in HotS.
No pipelinedb changes other than that.

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
5 years agogallivm: Fix saturated signed psub/padd intrinsics on llvm 8
Roland Scheidegger [Thu, 17 Oct 2019 02:14:28 +0000 (04:14 +0200)]
gallivm: Fix saturated signed psub/padd intrinsics on llvm 8

LLVM 8 did remove both the signed and unsigned sse2/avx intrinsics in
the end, and provide arch-independent llvm intrinsics instead.
Fixes a crash when using snorm framebuffers (tested with piglit
arb_color_buffer_float-render GL_RGBA8_SNORM -auto).

Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Reviewed-by: Dave Airlie <airlied@redhat.com>
CC: <mesa-stable@lists.freedesktop.org>
5 years agoradv: fix DCC fast clear code for intensity formats (correctly)
Samuel Pitoiset [Wed, 16 Oct 2019 20:51:08 +0000 (22:51 +0200)]
radv: fix DCC fast clear code for intensity formats (correctly)

Previous fix was pretty bogus.

This fixes a rendering regression with Nier (minimap too large).

Closes: https://gitlab.freedesktop.org/mesa/mesa/issues/1943
Closes: https://gitlab.freedesktop.org/mesa/mesa/issues/1952
Fixes: ea92273cea8 ("radv: fix DCC fast clear code for intensity formats")
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
5 years agopanfrost: Keep track of active BOs
Tomeu Vizoso [Wed, 9 Oct 2019 08:10:44 +0000 (10:10 +0200)]
panfrost: Keep track of active BOs

If two jobs use the same GEM object at the same time, the job that
finishes first will (previous to this commit) close the GEM object, even
if there's a job still referencing it.

To prevent this, have all jobs use the same panfrost_bo for a given GEM
object, so it's only closed once the last job is done with it.

Signed-off-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Rohan Garg <rohan.garg@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
5 years agonv50/ir: remove DUMMY edge type
Karol Herbst [Mon, 14 Oct 2019 20:50:58 +0000 (22:50 +0200)]
nv50/ir: remove DUMMY edge type

it was never used

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
5 years agogallium: do not increase ref count of the new throttle fence
James Xiong [Wed, 16 Oct 2019 16:22:22 +0000 (09:22 -0700)]
gallium: do not increase ref count of the new throttle fence

A new throttle fence was initialized to 1, and increased by 1
again when it's put in drawable->throttle_fence; the ref was
decreased by 1 when it's removed from drawable->throttle_fence,
and never reached to 0, caused leak.

Fixes: ff77bf5cbf7 ("gallium: simplify throttle implementation")
Closes: https://gitlab.freedesktop.org/mesa/mesa/issues/1949
Signed-off-by: James Xiong <james.xiong@intel.com>
Reported-by: Florian Wesch <fw@info-beamer.com>
Reviewed-by: Jose Maria Casanova Crespo <jmcasanova@igalia.com>
5 years agonir: drop unused alpha_ref_float
Erik Faye-Lund [Mon, 7 Oct 2019 10:19:15 +0000 (12:19 +0200)]
nir: drop unused alpha_ref_float

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
5 years agonir: drop support for using load_alpha_ref_float
Erik Faye-Lund [Mon, 7 Oct 2019 10:08:55 +0000 (12:08 +0200)]
nir: drop support for using load_alpha_ref_float

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
5 years agov3d: do not report alpha-test as supported
Erik Faye-Lund [Mon, 7 Oct 2019 10:18:09 +0000 (12:18 +0200)]
v3d: do not report alpha-test as supported

This triggers lowering in the state-tracker, which makes things a bit
simpler.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
5 years agovc4: do not report alpha-test as supported
Erik Faye-Lund [Mon, 7 Oct 2019 10:07:47 +0000 (12:07 +0200)]
vc4: do not report alpha-test as supported

This triggers lowering in the state-tracker, which makes things a bit
simpler.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
5 years agopanfrost: do not report alpha-test as supported
Erik Faye-Lund [Mon, 7 Oct 2019 10:07:20 +0000 (12:07 +0200)]
panfrost: do not report alpha-test as supported

This triggers lowering in the state-tracker, which makes things a bit
simpler.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
5 years agomesa/st: support lowering user-clip-planes automatically
Erik Faye-Lund [Thu, 25 Jul 2019 12:06:33 +0000 (14:06 +0200)]
mesa/st: support lowering user-clip-planes automatically

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
5 years agomesa/program: support referencing the clip-space clip-plane state
Erik Faye-Lund [Thu, 3 Oct 2019 20:53:47 +0000 (16:53 -0400)]
mesa/program: support referencing the clip-space clip-plane state

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
5 years agonir: support feeding state to nir_lower_clip_[vg]s
Erik Faye-Lund [Wed, 2 Oct 2019 20:30:45 +0000 (16:30 -0400)]
nir: support feeding state to nir_lower_clip_[vg]s

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
5 years agonir: support lowering clipdist to arrays
Erik Faye-Lund [Wed, 2 Oct 2019 20:19:08 +0000 (16:19 -0400)]
nir: support lowering clipdist to arrays

This allows us to make sure clipdist is emitted as a scalar array rather
than two vec4s. This matches SPIR-V semantics, and will be useful for
Zink.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
5 years agomesa/gallium: automatically lower two-sided lighting
Erik Faye-Lund [Thu, 3 Oct 2019 20:51:19 +0000 (16:51 -0400)]
mesa/gallium: automatically lower two-sided lighting

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
5 years agonir: support derefs in two-sided lighting lowering
Erik Faye-Lund [Thu, 3 Oct 2019 20:51:03 +0000 (16:51 -0400)]
nir: support derefs in two-sided lighting lowering

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
5 years agomesa/gallium: automatically lower point-size
Erik Faye-Lund [Thu, 3 Oct 2019 20:49:15 +0000 (16:49 -0400)]
mesa/gallium: automatically lower point-size

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
5 years agonir: add lowering-pass for point-size mov
Erik Faye-Lund [Thu, 3 Oct 2019 20:44:29 +0000 (16:44 -0400)]
nir: add lowering-pass for point-size mov

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
5 years agost/mesa: move point_size_per_vertex-logic to helper
Erik Faye-Lund [Thu, 18 Jul 2019 14:29:27 +0000 (16:29 +0200)]
st/mesa: move point_size_per_vertex-logic to helper

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
5 years agomesa/gallium: automatically lower alpha-testing
Erik Faye-Lund [Thu, 3 Oct 2019 20:35:23 +0000 (16:35 -0400)]
mesa/gallium: automatically lower alpha-testing

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
5 years agonir: allow passing alpha-ref state to lowering-code
Erik Faye-Lund [Thu, 3 Oct 2019 20:22:58 +0000 (16:22 -0400)]
nir: allow passing alpha-ref state to lowering-code

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
5 years agomesa: expose alpha-ref as a state-variable
Erik Faye-Lund [Thu, 4 Jul 2019 08:07:27 +0000 (10:07 +0200)]
mesa: expose alpha-ref as a state-variable

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
5 years agost/mesa: handling lower flatshading for NIR drivers.
Dave Airlie [Thu, 24 Jan 2019 03:12:10 +0000 (13:12 +1000)]
st/mesa: handling lower flatshading for NIR drivers.

This uses the NIR pass to lower flatshading when the driver
requests it.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
5 years agogallium: add flatshade lowering capability
Dave Airlie [Thu, 24 Jan 2019 03:11:22 +0000 (13:11 +1000)]
gallium: add flatshade lowering capability

This allows the driver to request flatshade lowering.
(NIR drivers only so far).

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
5 years agonir: add a pass to lower flat shading.
Dave Airlie [Thu, 24 Jan 2019 03:07:42 +0000 (13:07 +1000)]
nir: add a pass to lower flat shading.

This takes any color or backcolor that has unspecified
shading and converts it to flat shading.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
5 years agogallium/u_blitter: set a more sane viewport-state
Erik Faye-Lund [Mon, 15 Jul 2019 12:44:48 +0000 (14:44 +0200)]
gallium/u_blitter: set a more sane viewport-state

This actually corresponds to legal GL depth-ranges, because depth-clear
values are always in the 0..1 range in OpenGL.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
5 years agost/mesa: reorder and document code in st_translate_vertex_program
Marek Olšák [Wed, 16 Oct 2019 02:20:17 +0000 (22:20 -0400)]
st/mesa: reorder and document code in st_translate_vertex_program

move the TGSI code after the ARB_vp code

Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
5 years agost/mesa: call prog_to_nir sooner for ARB_fp
Marek Olšák [Wed, 16 Oct 2019 02:19:42 +0000 (22:19 -0400)]
st/mesa: call prog_to_nir sooner for ARB_fp

Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
5 years agost/mesa: don't call translate_*_program functions for NIR
Marek Olšák [Fri, 27 Sep 2019 21:43:42 +0000 (17:43 -0400)]
st/mesa: don't call translate_*_program functions for NIR

move the initializaton to st_link_nir

Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
5 years agost/mesa: finalize NIR after shader variant passes for TCS/TES/GS/CS
Marek Olšák [Wed, 16 Oct 2019 01:08:35 +0000 (21:08 -0400)]
st/mesa: finalize NIR after shader variant passes for TCS/TES/GS/CS

Same as VS and FS.

This might fix vertex color clamping.

Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>