mesa.git
7 years agoAndroid: remove remaining explicit libcxx includes
Rob Herring [Wed, 3 May 2017 19:35:15 +0000 (14:35 -0500)]
Android: remove remaining explicit libcxx includes

Explicitly including libcxx includes is not necessary at least on
Android M and later. It appears that libc++ was made the default in
commit "Make libc++ the default STL." in Android build system post L.
However, if L support is still needed, using "LOCAL_CXX_STL=libc++" is
the preferred way.

Signed-off-by: Rob Herring <robh@kernel.org>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
7 years agoAndroid: define required __STDC* macros as cflags
Mauro Rossi [Wed, 3 May 2017 19:35:14 +0000 (14:35 -0500)]
Android: define required __STDC* macros as cflags

Necessary to fix the following radeonsi building errors:

In file included from external/mesa/src/gallium/drivers/radeonsi/si_blit.c:24:
In file included from external/mesa/src/gallium/drivers/radeonsi/si_pipe.h:29:
In file included from external/mesa/src/gallium/drivers/radeonsi/si_shader.h:71:
In file included from external/llvm/include/llvm-c/Core.h:18:
In file included from external/llvm/include/llvm-c/ErrorHandling.h:17:
In file included from external/llvm/include/llvm-c/Types.h:17:
external/llvm/include/llvm/Support/DataTypes.h:49:3: error: "Must #define __STDC_LIMIT_MACROS before #including Support/DataTypes.h"
  ^
external/llvm/include/llvm/Support/DataTypes.h:53:3: error: "Must #define __STDC_CONSTANT_MACROS before "         "#including Support/DataTypes.h"
  ^
2 errors generated.

[Emil Velikov: add inline comment about the defines]
Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
7 years agoAndroid: drop static linking of R600 LLVM libraries
Mauro Rossi [Wed, 3 May 2017 19:35:13 +0000 (14:35 -0500)]
Android: drop static linking of R600 LLVM libraries

Inspired by Chih-Wei Huang and Zhen Wu similar patches

Linking against llvm with both static and shared may be avoided,
provided that libLLVM shared library for device supports
whole static R600/AMDGPU libraries, necessary for radeonsi/amdgpu.

Complementary changes, limited to android external/llvm project
are necessary to correclty build libLLVM

Tested with marshmallow-x86 and nougat-x86 builds

Reviewed-by: Chih-Wei Huang <cwhuang@linux.org.tw>
Signed-off-by: Rob Herring <robh@kernel.org>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
7 years agoconfigure.ac: Fix help string for --disable-pwr8 configure option
Philipp Zabel [Wed, 10 May 2017 15:44:29 +0000 (17:44 +0200)]
configure.ac: Fix help string for --disable-pwr8 configure option

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Reviewed-by: Eric Engestrom <eric.engestrom@imgtec.com>
7 years agomesa: remove _CurrentFragmentProgram from gl_pipeline_object
Timothy Arceri [Wed, 10 May 2017 00:37:35 +0000 (10:37 +1000)]
mesa: remove _CurrentFragmentProgram from gl_pipeline_object

This was added in b527dd65c830a as a work around because fixed function
fragment shaders were tracked in ctx->FragmentProgram._Current as
a gl_program rather than gl_shader_program.

However after my refactoring of the program and shader structs
at the end of 2016 which culminated in c505d6d85222, we no longer
need gl_shader_program to track the current program making
_CurrentFragmentProgram obsolete.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
7 years agomesa: add KHR_no_error support for FramebufferTexture*D functions
Timothy Arceri [Mon, 8 May 2017 01:10:58 +0000 (11:10 +1000)]
mesa: add KHR_no_error support for FramebufferTexture*D functions

Reviewed-by: Eric Anholt <eric@anholt.net>
7 years agomesa: add no error version of framebuffer_texture_with_dims()
Timothy Arceri [Mon, 8 May 2017 00:59:15 +0000 (10:59 +1000)]
mesa: add no error version of framebuffer_texture_with_dims()

Reviewed-by: Eric Anholt <eric@anholt.net>
7 years agomesa: add error version of get_texture_for_framebuffer()
Timothy Arceri [Fri, 5 May 2017 06:25:11 +0000 (16:25 +1000)]
mesa: add error version of get_texture_for_framebuffer()

This is a step towards KHR_no_error support.

Reviewed-by: Eric Anholt <eric@anholt.net>
7 years agomesa: pass rb attachment to _mesa_framebuffer_texture()
Timothy Arceri [Fri, 5 May 2017 07:00:34 +0000 (17:00 +1000)]
mesa: pass rb attachment to _mesa_framebuffer_texture()

This change will help us add KHR_no_error support to the caller.

Reviewed-by: Eric Anholt <eric@anholt.net>
7 years agomesa: add _mesa_get_and_validate_attachment() helper
Timothy Arceri [Fri, 5 May 2017 06:46:03 +0000 (16:46 +1000)]
mesa: add _mesa_get_and_validate_attachment() helper

Will be used to add KHR_no_error support. We make this available
external so it can be called from meta.

Reviewed-by: Eric Anholt <eric@anholt.net>
7 years agomesa: remove _mesa_problem() from a few locations
Timothy Arceri [Fri, 5 May 2017 05:39:15 +0000 (15:39 +1000)]
mesa: remove _mesa_problem() from a few locations

_mesa_problem() is still useful in some places such as is if a backend
compile fails, but for the majority of cases we should be able to
remove it.

OpenGL test suites are becoming very mature, we should place more
trust in debug builds picking up missed cases.

Reviewed-by: Eric Anholt <eric@anholt.net>
7 years agomesa: make _mesa_get_framebuffer_attachment_parameter() static
Timothy Arceri [Fri, 5 May 2017 05:21:22 +0000 (15:21 +1000)]
mesa: make _mesa_get_framebuffer_attachment_parameter() static

Reviewed-by: Eric Anholt <eric@anholt.net>
7 years agomesa: fix indentation
Timothy Arceri [Fri, 5 May 2017 05:09:37 +0000 (15:09 +1000)]
mesa: fix indentation

Reviewed-by: Eric Anholt <eric@anholt.net>
7 years agomesa: remove _mesa from static framebuffer object function
Timothy Arceri [Fri, 5 May 2017 05:02:20 +0000 (15:02 +1000)]
mesa: remove _mesa from static framebuffer object function

Reviewed-by: Eric Anholt <eric@anholt.net>
7 years agogallivm: Fix build against LLVM SVN >= r302589
Michel Dänzer [Wed, 10 May 2017 08:26:07 +0000 (17:26 +0900)]
gallivm: Fix build against LLVM SVN >= r302589

deregisterEHFrames doesn't take any parameters anymore.

Reviewed-by: Vedran Miletić <vedran@miletic.net>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
7 years agomesa: small _mesa_UseProgram() tidy up
Timothy Arceri [Wed, 3 May 2017 05:56:03 +0000 (15:56 +1000)]
mesa: small _mesa_UseProgram() tidy up

Makes the code easier to follow.

Reviewed-by: Eric Anholt <eric@anholt.net>
7 years agomesa: add KHR_no_error support for glBindProgramPipeline()
Timothy Arceri [Wed, 3 May 2017 05:34:52 +0000 (15:34 +1000)]
mesa: add KHR_no_error support for glBindProgramPipeline()

Reviewed-by: Eric Anholt <eric@anholt.net>
7 years agomesa: add KHR_no_error support for glActiveShaderProgram()
Timothy Arceri [Wed, 3 May 2017 05:26:22 +0000 (15:26 +1000)]
mesa: add KHR_no_error support for glActiveShaderProgram()

Reviewed-by: Eric Anholt <eric@anholt.net>
7 years agomesa: add KHR_no_error support for glUseProgramStages()
Timothy Arceri [Wed, 3 May 2017 04:20:26 +0000 (14:20 +1000)]
mesa: add KHR_no_error support for glUseProgramStages()

Reviewed-by: Eric Anholt <eric@anholt.net>
7 years agomesa: create use_program_stages() helper
Timothy Arceri [Wed, 3 May 2017 03:50:21 +0000 (13:50 +1000)]
mesa: create use_program_stages() helper

This will be used to create a KHR_no_error version of
glUseProgramStages().

Reviewed-by: Eric Anholt <eric@anholt.net>
7 years agoradv: handle fragment shader srgb resolve pass better
Dave Airlie [Thu, 11 May 2017 00:05:58 +0000 (10:05 +1000)]
radv: handle fragment shader srgb resolve pass better

Bas pointed out the fs key doesn't take srgb into account,
since there is just one srgb variant, just create a separate
pipeline for it. This also uses dest format to be more consistent
on when srgb matters.

Fixes: 69136f4e633 "radv/meta: add resolve pass using fragment/vertex shaders"
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Signed-off-by: Dave Airlie <airlied@redhat.com>
7 years agoi965: Make INTEL_DEBUG=bat decode VS/CLIP/GS/SF/WM/CC_STATE on Gen4-5.
Kenneth Graunke [Thu, 4 May 2017 05:07:16 +0000 (22:07 -0700)]
i965: Make INTEL_DEBUG=bat decode VS/CLIP/GS/SF/WM/CC_STATE on Gen4-5.

This is something the original decoder did, but I didn't bother with
until now.  I recently had to debug an Ironlake issue, and wanted to
inspect VS_STATE.  So, now it's back.

The other packets in the switch statement are all Gen6/7+, where we
use offsets from dynamic state base address, so we don't need the
gtt_offset subtraction introduced here.  We might want to make a
helper for this hack at some point - perhaps when we introduce the
next occurance.

Acked-by: Jason Ekstrand <jason@jlekstrand.net>
7 years agoi965: Switch BRW_NEW_CURBE_OFFSETS to BRW_NEW_PUSH_CONSTANT_ALLOCATION.
Kenneth Graunke [Mon, 8 May 2017 05:50:20 +0000 (22:50 -0700)]
i965: Switch BRW_NEW_CURBE_OFFSETS to BRW_NEW_PUSH_CONSTANT_ALLOCATION.

The BRW_NEW_CURBE_OFFSETS dirty bit is signalled when changing the
partitioning of the Constant Buffer URB section between the various
shader stages, on Gen4-5.

BRW_NEW_PUSH_CONSTANT_ALLOCATION is basically the same thing on Gen7+.

So, save a bit, and use the new name.

Reviewed-by: Topi Pohjolainen <topi.pohjolainen@intel.com>
7 years agoi965: Drop BRW_NEW_PUSH_CONSTANT_ALLOCATION from Gen6 code.
Kenneth Graunke [Mon, 8 May 2017 05:59:38 +0000 (22:59 -0700)]
i965: Drop BRW_NEW_PUSH_CONSTANT_ALLOCATION from Gen6 code.

Gen6 doesn't have a configurable push constant region.  This is only
used on Gen7+.

Reviewed-by: Topi Pohjolainen <topi.pohjolainen@intel.com>
7 years agoi965: Only #if...#endif a single function or related section at a time.
Kenneth Graunke [Mon, 1 May 2017 22:51:41 +0000 (15:51 -0700)]
i965: Only #if...#endif a single function or related section at a time.

Previously we guarded large swathes of code with #if GEN ... #endif
blocks.  This made it difficult to see which generations include what.

This patch splits up the #if..#endif sections so they surround a small
section of code - usually a single function/atom, or sometimes a group
of related functions.  It should make the code easier to work on.

Reviewed-by: Rafael Antognolli <rafael.antognolli@intel.com>
7 years agoi965: Turn brw_get_line_width_float() into brw_get_line_width().
Kenneth Graunke [Tue, 9 May 2017 05:39:25 +0000 (22:39 -0700)]
i965: Turn brw_get_line_width_float() into brw_get_line_width().

Drop the old brw_get_line_width() helper which return the unsigned
fixed-point encoding of the line width - it's been dead since the
conversion to GENXML (which does the encoding for us).

Then rename brw_get_line_width_float() to the shorter name.

Reviewed-by: Rafael Antognolli <rafael.antognolli@intel.com>
7 years agoi965: Drop INTEL_DEBUG=stats.
Kenneth Graunke [Tue, 9 May 2017 06:53:36 +0000 (23:53 -0700)]
i965: Drop INTEL_DEBUG=stats.

For whatever reason, we had an INTEL_DEBUG=stats option that enabled
various statistics counters on Gen4-5 systems.  It's been around
forever, though I can't think of a single time that it's been useful.

On Gen6+, we enable statistics all the time because they're necessary
to support various query object targets.  Turning them off would break
those queries.

Gen4-5 don't support those queries, so the statistics counters generally
aren't useful; we disabled them by default.  This patch disables them
altogether.

Reviewed-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com>
7 years agoi965: Disable ARB_pipeline_statistics_query on Gen4-5.
Kenneth Graunke [Tue, 9 May 2017 06:41:31 +0000 (23:41 -0700)]
i965: Disable ARB_pipeline_statistics_query on Gen4-5.

We apparently enabled this on all platforms in Mesa 10.6.  However, it
was only ever implemented for Gen6+.  The Gen4-5 query code goes up in
flames with an "Unrecognized query target" unreachable() error if you
even attempt to use any of the new functionality.

This wasn't caught because the Piglit tests require OpenGL 3.0, which
Gen4-5 cannot support.  The extension spec does say 3.0 is required,
though I'm not sure why - it seems like 2.1 would work fine.

We could implement it anyway, but it's a little bit of a pain due to the
lack of hardware contexts (so we have to snapshot around batches).

Given that it's been 100% broken for two years and I haven't seen a bug
report about it, I'm not terribly inclined to care.  So, let it go.

Reviewed-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com>
7 years agoradeonsi: add new vega10 pci ids
Alex Deucher [Wed, 10 May 2017 15:40:01 +0000 (11:40 -0400)]
radeonsi: add new vega10 pci ids

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Cc: 17.1 <mesa-stable@lists.freedesktop.org>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
7 years agost/mesa: move the logic of all_varyings_in_vbos into st_update_array
Marek Olšák [Mon, 8 May 2017 00:15:08 +0000 (02:15 +0200)]
st/mesa: move the logic of all_varyings_in_vbos into st_update_array

The function was pretty slow. This brings a substantial decrease in draw
call overhead when min/max index bounds are not needed:

Before:  DrawElements (1 VBO) w/ no state change:          5.75 million
After:   DrawElements (1 VBO) w/ no state change:          7.03 million

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
7 years agost/mesa: unify common code in st_draw_vbo functions
Marek Olšák [Mon, 8 May 2017 00:11:31 +0000 (02:11 +0200)]
st/mesa: unify common code in st_draw_vbo functions

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
7 years agost/mesa: make st_draw_vbo static
Marek Olšák [Mon, 8 May 2017 00:08:17 +0000 (02:08 +0200)]
st/mesa: make st_draw_vbo static

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
7 years agoradeonsi: remove upload code for zero-stride vertex attribs
Marek Olšák [Sat, 29 Apr 2017 23:24:29 +0000 (01:24 +0200)]
radeonsi: remove upload code for zero-stride vertex attribs

st/mesa takes care of it now.

Tested-by: Edmondo Tommasina <edmondo.tommasina@gmail.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
7 years agost/mesa: upload zero-stride vertex attributes here
Marek Olšák [Sat, 29 Apr 2017 23:20:19 +0000 (01:20 +0200)]
st/mesa: upload zero-stride vertex attributes here

This is the best place to do it. Now drivers without u_vbuf don't have to
do it.

v2: use correct upload size and optimal alignment

Tested-by: Edmondo Tommasina <edmondo.tommasina@gmail.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
7 years agogallium: add PIPE_CAP_CAN_BIND_CONST_BUFFER_AS_VERTEX
Marek Olšák [Sat, 29 Apr 2017 23:18:43 +0000 (01:18 +0200)]
gallium: add PIPE_CAP_CAN_BIND_CONST_BUFFER_AS_VERTEX

The next patch will use it. This is really for svga and GL2-level drivers.

Tested-by: Edmondo Tommasina <edmondo.tommasina@gmail.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
7 years agost/mesa: simplify the signature of get_client_array
Marek Olšák [Sun, 30 Apr 2017 12:57:35 +0000 (14:57 +0200)]
st/mesa: simplify the signature of get_client_array

Tested-by: Edmondo Tommasina <edmondo.tommasina@gmail.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
7 years agost/mesa: remove vpv->num_inputs dereferences in st_update_array
Marek Olšák [Sun, 30 Apr 2017 12:42:45 +0000 (14:42 +0200)]
st/mesa: remove vpv->num_inputs dereferences in st_update_array

Tested-by: Edmondo Tommasina <edmondo.tommasina@gmail.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
7 years agost/mesa: fold error handling into setup_(non_)interleaved_attribs
Marek Olšák [Sun, 30 Apr 2017 12:42:45 +0000 (14:42 +0200)]
st/mesa: fold error handling into setup_(non_)interleaved_attribs

Tested-by: Edmondo Tommasina <edmondo.tommasina@gmail.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
7 years agost/mesa: fold cso calls into setup_(non_)interleaved_attribs
Marek Olšák [Sun, 30 Apr 2017 12:42:45 +0000 (14:42 +0200)]
st/mesa: fold cso calls into setup_(non_)interleaved_attribs

Tested-by: Edmondo Tommasina <edmondo.tommasina@gmail.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
7 years agost/mesa: don't call util_draw_init_info in st_draw_vbo
Marek Olšák [Sun, 9 Apr 2017 11:17:54 +0000 (13:17 +0200)]
st/mesa: don't call util_draw_init_info in st_draw_vbo

7 years agogallium: remove pipe_index_buffer and set_index_buffer
Marek Olšák [Sun, 2 Apr 2017 14:24:39 +0000 (16:24 +0200)]
gallium: remove pipe_index_buffer and set_index_buffer

pipe_draw_info::indexed is replaced with index_size. index_size == 0 means
non-indexed.

Instead of pipe_index_buffer::offset, pipe_draw_info::start is used.
For indexed indirect draws, pipe_draw_info::start is added to the indirect
start. This is the only case when "start" affects indirect draws.

pipe_draw_info::index is a union. Use either index::resource or
index::user depending on the value of pipe_draw_info::has_user_indices.

v2: fixes for nine, svga

7 years agogallium: separate indirect stuff from pipe_draw_info - 80 -> 56 bytes
Marek Olšák [Sun, 2 Apr 2017 13:24:19 +0000 (15:24 +0200)]
gallium: separate indirect stuff from pipe_draw_info - 80 -> 56 bytes

For faster initialization of non-indirect draws.

7 years agogallium: decrease the size of pipe_vertex_buffer - 24 -> 16 bytes
Marek Olšák [Sun, 2 Apr 2017 12:30:16 +0000 (14:30 +0200)]
gallium: decrease the size of pipe_vertex_buffer - 24 -> 16 bytes

7 years agodocs: add news item and link release notes for 17.1.0
Emil Velikov [Wed, 10 May 2017 14:24:03 +0000 (15:24 +0100)]
docs: add news item and link release notes for 17.1.0

Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
7 years agodocs: add sha256 checksums for 17.1.0
Emil Velikov [Wed, 10 May 2017 14:20:37 +0000 (15:20 +0100)]
docs: add sha256 checksums for 17.1.0

Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
(cherry picked from commit 806f802e7b6a898a4f8c47bc56b97579a86a873e)

7 years agodocs: Update 17.1.0 release notes
Emil Velikov [Wed, 10 May 2017 11:11:03 +0000 (12:11 +0100)]
docs: Update 17.1.0 release notes

Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
(cherry picked from commit 15a38605fc3aa49c01943f9ec2be19d58fef4aac)

7 years agost/glsl_to_tgsi: make sure resource file for samplers is PROGRAM_SAMPLER
Samuel Pitoiset [Fri, 5 May 2017 13:36:27 +0000 (15:36 +0200)]
st/glsl_to_tgsi: make sure resource file for samplers is PROGRAM_SAMPLER

Similar to how image resources are handled. That way we are sure
that inst->resource.file is PROGRAM_SAMPLER for "bound" samplers.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
7 years agoradeonsi: silent a compiler warning
Samuel Pitoiset [Wed, 10 May 2017 10:20:41 +0000 (12:20 +0200)]
radeonsi: silent a compiler warning

This fixes:

si_shader.c: In function ‘si_shader_dump_stats’:
si_shader.c:6704:31: warning: passing argument 1 of ‘si_get_max_workgroup_size’ discards ‘const’ qualifier from pointer target type [-Wdiscarded-qualifiers]
     si_get_max_workgroup_size(shader);
                               ^~~~~~
si_shader.c:5832:17: note: expected ‘struct si_shader *’ but argument is of type ‘const struct si_shader *’
 static unsigned si_get_max_workgroup_size(struct si_shader *shader)

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
7 years agomesa: use u_bit_scan() in update_program_texture_state()
Samuel Pitoiset [Tue, 9 May 2017 21:33:27 +0000 (23:33 +0200)]
mesa: use u_bit_scan() in update_program_texture_state()

The check in update_single_program_texture() can also be
removed.

v2: - remove unused 's' variable

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
7 years agomesa: remove never used gl_shader_compiler_options::EmitNoFunctions
Samuel Pitoiset [Mon, 8 May 2017 20:54:06 +0000 (22:54 +0200)]
mesa: remove never used gl_shader_compiler_options::EmitNoFunctions

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Elie Tournier <elie.tournier@collabora.com>
7 years agoradeonsi: dump compute descriptor lists
Nicolai Hähnle [Mon, 8 May 2017 11:10:02 +0000 (13:10 +0200)]
radeonsi: dump compute descriptor lists

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
7 years agoradeonsi: dump both enabled and required descriptor slots
Nicolai Hähnle [Mon, 8 May 2017 11:10:35 +0000 (13:10 +0200)]
radeonsi: dump both enabled and required descriptor slots

This allows a meaningful dump with info == NULL (for compute shaders).

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
7 years agoradeonsi: dump compute shader as part of debug dump
Nicolai Hähnle [Mon, 8 May 2017 11:01:37 +0000 (13:01 +0200)]
radeonsi: dump compute shader as part of debug dump

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
7 years agoradeonsi: move struct si_compute into a header
Nicolai Hähnle [Mon, 8 May 2017 11:01:18 +0000 (13:01 +0200)]
radeonsi: move struct si_compute into a header

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
7 years agoradeonsi: split descriptor list dumping
Nicolai Hähnle [Mon, 8 May 2017 11:02:33 +0000 (13:02 +0200)]
radeonsi: split descriptor list dumping

Prepare for dumping CS descriptor list.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
7 years agoradeonsi: split shader dumping
Nicolai Hähnle [Mon, 8 May 2017 11:00:12 +0000 (13:00 +0200)]
radeonsi: split shader dumping

Prepare for dumping compute shaders.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
7 years agoradeonsi: more const qualifiers in shader dump functions
Nicolai Hähnle [Mon, 8 May 2017 10:59:26 +0000 (12:59 +0200)]
radeonsi: more const qualifiers in shader dump functions

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
7 years agoddebug: implement dd_dump_launch_grid
Nicolai Hähnle [Mon, 8 May 2017 11:29:50 +0000 (13:29 +0200)]
ddebug: implement dd_dump_launch_grid

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
7 years agoddebug: extract dd_dump_shader
Nicolai Hähnle [Mon, 8 May 2017 11:29:33 +0000 (13:29 +0200)]
ddebug: extract dd_dump_shader

Will be re-used for compute shaders.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
7 years agogallium/util: dump tokens in util_dump_shader_state only if type is TGSI
Nicolai Hähnle [Mon, 8 May 2017 15:29:03 +0000 (17:29 +0200)]
gallium/util: dump tokens in util_dump_shader_state only if type is TGSI

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
7 years agogallium/util: add util_dump_grid_info
Nicolai Hähnle [Mon, 8 May 2017 11:29:02 +0000 (13:29 +0200)]
gallium/util: add util_dump_grid_info

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
7 years agoradv: always free nir shaders from modules on stack
Grazvydas Ignotas [Tue, 2 May 2017 19:46:06 +0000 (22:46 +0300)]
radv: always free nir shaders from modules on stack

valgrind reports them as leaked, and I could not find anything making a
copy of the nir pointer. Also, radv_device_init_meta_blit_color() is
already freeing them unconditionally like this.

Signed-off-by: Grazvydas Ignotas <notasas@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
7 years agoanv: don't leak DRM devices
Grazvydas Ignotas [Tue, 2 May 2017 17:11:06 +0000 (20:11 +0300)]
anv: don't leak DRM devices

After successful drmGetDevices2() call, drmFreeDevices() needs to be
called.

Fixes: b1fb6e8d "anv: do not open random render node(s)"
Signed-off-by: Grazvydas Ignotas <notasas@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl> # radv version
7 years agoanv: fix possible stack corruption
Grazvydas Ignotas [Tue, 2 May 2017 16:26:17 +0000 (19:26 +0300)]
anv: fix possible stack corruption

drmGetDevices2 takes count and not size. Probably hasn't caused problems
yet in practice and was missed as setups with more than 8 DRM devices
are not very common.

Fixes: b1fb6e8d "anv: do not open random render node(s)"
Signed-off-by: Grazvydas Ignotas <notasas@gmail.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
7 years agoi965/vec4: Delete the system value infastructure
Jason Ekstrand [Fri, 5 May 2017 21:31:41 +0000 (14:31 -0700)]
i965/vec4: Delete the system value infastructure

The only thing still using it is INVOCATION_ID for geometry shaders.
That's easily enough inlined into the nir_intrinsic_load_invocation_id
handling code.

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
7 years agoi965/vec4: Use NIR to do GS input remapping
Jason Ekstrand [Thu, 4 May 2017 23:36:26 +0000 (16:36 -0700)]
i965/vec4: Use NIR to do GS input remapping

We're already doing this in the FS back-end.  This just does the same
thing in the vec4 back-end.

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
7 years agoi965/fs: Move remapping of gl_PointSize to the NIR level
Jason Ekstrand [Thu, 4 May 2017 23:33:32 +0000 (16:33 -0700)]
i965/fs: Move remapping of gl_PointSize to the NIR level

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
7 years agoi965/nir: Inline remap_inputs_with_vue_map
Jason Ekstrand [Thu, 4 May 2017 21:57:52 +0000 (14:57 -0700)]
i965/nir: Inline remap_inputs_with_vue_map

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
7 years agoi965/vec4: Use NIR remapping for VS attributes
Jason Ekstrand [Thu, 4 May 2017 00:40:54 +0000 (17:40 -0700)]
i965/vec4: Use NIR remapping for VS attributes

The NIR pass already handles remapping system values to attributes for
us so we delete the system value code as part of the conversion.

We also change nir_lower_vs_inputs to take an explicit inputs_read
bitmask and pass in the inputs_read from prog_data instead from pulling
it out of NIR.  This is because the version in prog_data may get
EDGEFLAG added to it on some old platforms.

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
7 years agointel/compiler/vs: Move inputs_read handling to generic code
Jason Ekstrand [Thu, 4 May 2017 21:50:20 +0000 (14:50 -0700)]
intel/compiler/vs: Move inputs_read handling to generic code

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
7 years agoi965/vec4: Set VERT_BIT_EDGEFLAG based on the VUE map
Jason Ekstrand [Tue, 9 May 2017 19:34:10 +0000 (12:34 -0700)]
i965/vec4: Set VERT_BIT_EDGEFLAG based on the VUE map

We also add a nice little comment to make it more clear exactly what
happens with the edge flag copy.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
7 years agoi965/fs: Lower gl_VertexID and friends to inputs at the NIR level
Jason Ekstrand [Wed, 3 May 2017 23:53:40 +0000 (16:53 -0700)]
i965/fs: Lower gl_VertexID and friends to inputs at the NIR level

NIR calls these system values but they come in from the VF unit as
vertex data.  It's terribly convenient to just be able to treat them as
such in the back-end.

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
7 years agoi965/vs: Set uses_vertexid and friends from brw_compile_vs
Jason Ekstrand [Thu, 4 May 2017 00:24:43 +0000 (17:24 -0700)]
i965/vs: Set uses_vertexid and friends from brw_compile_vs

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
7 years agoi965: Move multiply by 4 for VS ATTR setup into the scalar backend.
Jason Ekstrand [Wed, 3 May 2017 23:56:15 +0000 (16:56 -0700)]
i965: Move multiply by 4 for VS ATTR setup into the scalar backend.

The vec4 backend will want to count in units of vec4s, not scalar
components.  The simplest solution is to move the multiplication by 4
into the scalar backend.  This also improves consistency with how we
count varyings.

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
7 years agoi965/nir: Inline remap_vs_attrs
Jason Ekstrand [Wed, 3 May 2017 23:41:43 +0000 (16:41 -0700)]
i965/nir: Inline remap_vs_attrs

Now that we have nice block iterators, there's no good reason for this
to be off on it's own.  While we're here, we convert to using the NIR
const index getters/setters instead of whacking const_index values
directly.

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
7 years agonir: Embed the shader_info in the nir_shader again
Jason Ekstrand [Mon, 8 May 2017 16:20:21 +0000 (09:20 -0700)]
nir: Embed the shader_info in the nir_shader again

Commit e1af20f18a86f52a9640faf2d4ff8a71b0a4fa9b changed the shader_info
from being embedded into being just a pointer.  The idea was that
sharing the shader_info between NIR and GLSL would be easier if it were
a pointer pointing to the same shader_info struct.  This, however, has
caused a few problems:

 1) There are many things which generate NIR without GLSL.  This means
    we have to support both NIR shaders which come from GLSL and ones
    that don't and need to have an info elsewhere.

 2) The solution to (1) raises all sorts of ownership issues which have
    to be resolved with ralloc_parent checks.

 3) Ever since 00620782c92100d77c660f9783504c6d80fa1d58, we've been
    using nir_gather_info to fill out the final shader_info.  Thanks to
    cloning and the above ownership issues, the nir_shader::info may not
    point back to the gl_shader anymore and so we have to do a copy of
    the shader_info from NIR back to GLSL anyway.

All of these issues go away if we just embed the shader_info in the
nir_shader.  There's a little downside of having to copy it back after
calling nir_gather_info but, as explained above, we have to do that
anyway.

Acked-by: Timothy Arceri <tarceri@itsqueeze.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
7 years agomesa: Make _mesa_primitive_restart_index a static inline in the header.
Kenneth Graunke [Sat, 6 May 2017 23:11:03 +0000 (16:11 -0700)]
mesa: Make _mesa_primitive_restart_index a static inline in the header.

It's now basically a single expression, so it probably makes sense to
have it inlined into the callers.

Suggested by Marek.

Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
7 years agofreedreno: fix clang error in fd_get_compute_param
Rob Herring [Tue, 9 May 2017 16:56:49 +0000 (11:56 -0500)]
freedreno: fix clang error in fd_get_compute_param

With commit 10c17f23b752 ("freedreno: core compute state support"),
Android builds fail with the following error:

external/mesa3d/src/gallium/drivers/freedreno/freedreno_screen.c:610:17: error: format string is not a string literal (potentially insecure) [-Werror,-Wformat-security]
                        sprintf(ret, ir);
                                     ^~

Signed-off-by: Rob Herring <robh@kernel.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>
7 years agomesa/vbo: fix invalid min/max indexes
Rob Clark [Mon, 8 May 2017 20:31:56 +0000 (16:31 -0400)]
mesa/vbo: fix invalid min/max indexes

Fixes: c3f37e9b ("st/mesa: use min_index and max_index directly from vbo")
Signed-off-by: Rob Clark <robdclark@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Tested-by: Dieter Nützel <Dieter@nuetzel-hh.de>
7 years agointel: compiler: prevent integer overflow
Lionel Landwerlin [Mon, 8 May 2017 17:55:06 +0000 (18:55 +0100)]
intel: compiler: prevent integer overflow

CID: 13994771399478 (Integer handling issues)

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
7 years agointel: compiler: remove duplicated code
Lionel Landwerlin [Mon, 8 May 2017 17:50:53 +0000 (18:50 +0100)]
intel: compiler: remove duplicated code

CID: 1399470: (Control flow issues)

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
7 years agointel: gen decoder: don't check for size_t negative values
Lionel Landwerlin [Mon, 8 May 2017 17:47:06 +0000 (18:47 +0100)]
intel: gen decoder: don't check for size_t negative values

We should get either 0 or 1 here.

CID: 1373562 (Control flow issues)

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Acked-by: Matt Turner <mattst88@gmail.com>
7 years agobin/*py: honor editorconfig formatting
Andres Gomez [Fri, 5 May 2017 13:09:42 +0000 (16:09 +0300)]
bin/*py: honor editorconfig formatting

Replace the two stray tabs with respective space.

Signed-off-by: Andres Gomez <agomez@igalia.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
7 years agobin: use tabs for coding style on *.sh files
Andres Gomez [Fri, 5 May 2017 14:49:52 +0000 (17:49 +0300)]
bin: use tabs for coding style on *.sh files

v2: Instead of changing *.sh, adapt the editorconfig file (Emil).

Signed-off-by: Andres Gomez <agomez@igalia.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
7 years agoandroid: i965: add per-gen libmesa_i965_gen{4,45,5} static
Mauro Rossi [Sat, 6 May 2017 14:03:23 +0000 (16:03 +0200)]
android: i965: add per-gen libmesa_i965_gen{4,45,5} static

Needed to fix android building errors:

external/mesa/src/mesa/drivers/dri/i965/brw_state_upload.c:148: error: undefined reference to 'gen5_init_atoms'
external/mesa/src/mesa/drivers/dri/i965/brw_state_upload.c:150: error: undefined reference to 'gen45_init_atoms'
external/mesa/src/mesa/drivers/dri/i965/brw_state_upload.c:152: error: undefined reference to 'gen4_init_atoms'
clang++: error: linker command failed with exit code 1 (use -v to see invocation)

Fixes: 5a19d0b ("i965: Get real per-gen atom lists")
Acked-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
7 years agoswr: fix polygonmode for front==back
George Kyriazis [Wed, 19 Apr 2017 18:55:26 +0000 (13:55 -0500)]
swr: fix polygonmode for front==back

Rasterizer core only supports polygonmode front==back.  Add logic for
populating fillMode for the rasterizer only for that case correctly.
Provide enum conversion between mesa enums and core enums.

The core renders lines/points as tris. Previously, code would enable
stipple for polygonmode != FILL.  Modify stipple enable logic so that
this works correctly.

No regressions in vtk tests.
Fixes the following piglit tests:
pointsprite
gl-1.0-edgeflag-const

v2: remove cc stable, and remove "not implemented" assert
v3: modified commit message

Reviewed-by: Bruce Cherniak <bruce.cherniak@intel.com>
7 years agoswr/rast: support polygonmode point
George Kyriazis [Mon, 1 May 2017 19:02:51 +0000 (14:02 -0500)]
swr/rast: support polygonmode point

Add support for polygonmode point in the binner.  This is done by
splitting BinPostSetupPoints from BinPoints, so the earlier call can be
called from BinTriangles.  Setup has already been done at the time
BinPostSetupPoints needs to be called.

This checkin just adds support in the rasterizer.  A separate checkin
will add the appropriate driver support.

v2: remove cc stable
v3: modified commit message and subject line

Reviewed-by: Bruce Cherniak <bruce.cherniak@intel.com>
7 years agoutil: move ALWAYS_INLINE macro to util/macro.h
Timothy Arceri [Sun, 7 May 2017 22:01:05 +0000 (08:01 +1000)]
util: move ALWAYS_INLINE macro to util/macro.h

Also added clang check.

macro.h is include by p_compiler.h so no other change is needed.

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
7 years agoswr: move msaa resolve to generalized StoreTile
Bruce Cherniak [Fri, 5 May 2017 00:33:36 +0000 (19:33 -0500)]
swr: move msaa resolve to generalized StoreTile

v3: list piglit tests fixed by this patch. Fixed typo Tim pointed out.
v2: Reword commit message to more closely adhere to community
guidelines.

This patch moves msaa resolve down into core/StoreTiles where the
surface format conversion routines are available.  The previous
"experimental" resolve was limited to 8-bit unsigned render targets.

This fixes a number of piglit msaa tests by adding resolve support for
all the render target formats we support.

Specifically:
layered-rendering/gl-layer-render: fail->pass
layered-rendering/gl-layer-render-storage: fail->pass
multisample-formats *[2,4,8,16] gl_arb_texture_rg: crash->pass
multisample-formats *[2,4,8,16] gl_ext_texture_snorm: crash->pass
multisample-formats *[2,4,8,16] gl_arb_texture_float: fail->pass
multisample-formats *[2,4,8,16] gl_arb_texture_rg-float: fail->pass

MSAA is still disabled by default, but can be enabled with
"export SWR_MSAA_MAX_COUNT=4" (1,2,4,8,16 are options)
The default is 0, which is disabled.

This patch improves the number of multisample-formats supported by swr,
and fixes several crashes currently in the 17.1 branch.  Therefore, it
should be considered for inclusion in the 17.1 stable release.  Being
disabled by default, it poses no risk to most users of swr.

Reviewed-by: Tim Rowley <timothy.o.rowley@intel.com>
cc: mesa-stable@lists.freedesktop.org

7 years agoglsl: Don't allow redefining builtin functions on GLSL 1.00.
Eric Anholt [Mon, 1 May 2017 23:49:27 +0000 (16:49 -0700)]
glsl: Don't allow redefining builtin functions on GLSL 1.00.

The spec text cited above says you can't, but only the GLSL 3.00 (redefine
or overload) case was implemented.

Fixes dEQP scoping.invalid.redefine_builtin_fragment/vertex.

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Tested-by: Matt Turner <mattst88@gmail.com>
7 years agoglsl: Restrict func redeclarations (not just redefinitions) on GLSL 1.00.
Eric Anholt [Mon, 1 May 2017 23:42:03 +0000 (16:42 -0700)]
glsl: Restrict func redeclarations (not just redefinitions) on GLSL 1.00.

Fixes DEQP's scoping.invalid.redeclare_function_fragment/vertex.

v2: Fix accidental rejection of prototype+decl.

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com> (v1)
Tested-by: Matt Turner <mattst88@gmail.com>
7 years agoglsl: Ban #undefining __LINE__ and friends on GLES2.
Eric Anholt [Mon, 1 May 2017 23:35:34 +0000 (16:35 -0700)]
glsl: Ban #undefining __LINE__ and friends on GLES2.

Fixes deqp_gles2 undefine_invalid_object_* failures.

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Tested-by: Matt Turner <mattst88@gmail.com>
7 years agoglsl: Restrict functions to not return arrays or SOAs in GLSL 1.00.
Eric Anholt [Mon, 1 May 2017 23:00:47 +0000 (16:00 -0700)]
glsl: Restrict functions to not return arrays or SOAs in GLSL 1.00.

From the spec,

    Arrays are allowed as arguments, but not as the return type. [...] The
    return type can also be a structure if the structure does not contain
    an array.

Fixes DEQP shaders.functions.invalid.return_array_in_struct_fragment.

v2: Spec cite wording change

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Tested-by: Matt Turner <mattst88@gmail.com>
7 years agonir: fix (hopefully) windows build
Rob Clark [Mon, 8 May 2017 17:34:53 +0000 (13:34 -0400)]
nir: fix (hopefully) windows build

Fixes: 53aa109b ("nir: add pass to lower atomic counters to SSBO")
Signed-off-by: Rob Clark <robdclark@gmail.com>
7 years agoradeonsi: rename si_eliminate_const_vs_outputs -> si_optimize_vs_outputs
Marek Olšák [Mon, 8 May 2017 14:41:09 +0000 (16:41 +0200)]
radeonsi: rename si_eliminate_const_vs_outputs -> si_optimize_vs_outputs

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
7 years agoac: fix broken elimination of duplicated VS exports
Marek Olšák [Mon, 8 May 2017 14:37:26 +0000 (16:37 +0200)]
ac: fix broken elimination of duplicated VS exports

The renumbering code didn't take into account that multiple VS exports
can have the same PARAM index. This also significantly simplifies
the renumbering. Thankfully, we have piglits for this:

    spec@arb_gpu_shader5@arb_gpu_shader5-interpolateatcentroid-packing
    spec@glsl-1.50@execution@interface-blocks-complex-vs-fs

Reported by Michel Dänzer.

Fixes: b08715499e61 ("ac: eliminate duplicated VS exports")
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
7 years agoegl: Fix -Wint-to-pointer-cast
Chad Versace [Fri, 5 May 2017 19:27:03 +0000 (12:27 -0700)]
egl: Fix -Wint-to-pointer-cast

main/egldisplay.c: In function '_eglParseX11DisplayAttribList':
main/egldisplay.c:491:38: warning: cast to pointer from integer of different size [-Wint-to-pointer-cast]
          display->Options.Platform = (void *)value;

The fix: cast to uinptr_t before void*.
                                      ^
Fixes: ddb99127 egl/x11: Honor the EGL_PLATFORM_X11_SCREEN_EXT attribute
Cc: Adam Jackson <ajax@redhat.com>
Reviewed-by: Eric Engestrom <eric@engestrom.ch>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
7 years agost/mesa: remove unused st parameter in init_velement_lowered
Marek Olšák [Sun, 30 Apr 2017 12:49:45 +0000 (14:49 +0200)]
st/mesa: remove unused st parameter in init_velement_lowered

Tested-by: Edmondo Tommasina <edmondo.tommasina@gmail.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
7 years agost/mesa: use PIPE_MAX_ATTRIBS as the max number of vertex buffers
Marek Olšák [Sun, 30 Apr 2017 12:28:34 +0000 (14:28 +0200)]
st/mesa: use PIPE_MAX_ATTRIBS as the max number of vertex buffers

Tested-by: Edmondo Tommasina <edmondo.tommasina@gmail.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
7 years agost/mesa: simplify code due to unification to st_common_program
Marek Olšák [Sun, 30 Apr 2017 14:31:07 +0000 (16:31 +0200)]
st/mesa: simplify code due to unification to st_common_program

v2: use the st_common_program() helper

Tested-by: Edmondo Tommasina <edmondo.tommasina@gmail.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>