mesa.git
8 years agovk: Advertise number of timestamp bits
Kristian Høgsberg Kristensen [Sat, 9 Jan 2016 08:51:14 +0000 (00:51 -0800)]
vk: Advertise number of timestamp bits

We have 36 bits.

8 years agovk: Expose correct timestampPeriod for SKL
Kristian Høgsberg Kristensen [Sat, 9 Jan 2016 08:50:04 +0000 (00:50 -0800)]
vk: Expose correct timestampPeriod for SKL

Skylake uses 83.333ms per tick.

8 years agovk: Mark VkEvent and VkSemaphore as done
Kristian Høgsberg Kristensen [Sat, 9 Jan 2016 08:48:41 +0000 (00:48 -0800)]
vk: Mark VkEvent and VkSemaphore as done

8 years agovk: Assert on use of uninitialized surface state
Kristian Høgsberg Kristensen [Sat, 9 Jan 2016 07:43:20 +0000 (23:43 -0800)]
vk: Assert on use of uninitialized surface state

This exposes a case where we want to anv_CmdCopyBufferToImage() on an
image that wasn't created with VK_IMAGE_USAGE_COLOR_ATTACHMENT_BIT and
end up using uninitialized color_rt_surface_state from the meta image
view.

8 years agovk: Only begin subpass if we're continuing a render pass
Kristian Høgsberg Kristensen [Sat, 9 Jan 2016 06:50:18 +0000 (22:50 -0800)]
vk: Only begin subpass if we're continuing a render pass

If VK_COMMAND_BUFFER_USAGE_RENDER_PASS_CONTINUE_BIT is not set in
pBeginInfo->flags, we don't have a render pass or framebuffer. Change
the condition that guard looking up render pass and framebuffer to test
for VK_COMMAND_BUFFER_USAGE_RENDER_PASS_CONTINUE_BIT instead of
VK_COMMAND_BUFFER_LEVEL_SECONDARY.

Fixes all remaining crashes in dEQP-VK.api.command_buffers.*.

8 years agovk: Remove unsupported warnings for Skylake and Broxton
Kristian Høgsberg Kristensen [Sat, 9 Jan 2016 06:24:58 +0000 (22:24 -0800)]
vk: Remove unsupported warnings for Skylake and Broxton

These are working as well as Broadwell and Cherryiew. The recent merge
from mesa master brings in Kabylake device info and that should be all
we need to enable that.

8 years agoMerge ../mesa into vulkan
Kristian Høgsberg Kristensen [Sat, 9 Jan 2016 06:16:43 +0000 (22:16 -0800)]
Merge ../mesa into vulkan

8 years agoanv/reloc_list: Make valgrind explicitly check relocation data
Jason Ekstrand [Sat, 9 Jan 2016 00:44:54 +0000 (16:44 -0800)]
anv/reloc_list: Make valgrind explicitly check relocation data

8 years agomesa: merge bind_atomic_buffers_{base|range}
Nicolai Hähnle [Wed, 6 Jan 2016 22:30:18 +0000 (17:30 -0500)]
mesa: merge bind_atomic_buffers_{base|range}

Reduced code duplication should make the code more maintainable.

Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
8 years agomesa: merge bind_shader_storage_buffers_{base|range}
Nicolai Hähnle [Wed, 6 Jan 2016 22:26:14 +0000 (17:26 -0500)]
mesa: merge bind_shader_storage_buffers_{base|range}

Reduced code duplication should make the code more maintainable.

Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
8 years agomesa: merge bind_uniform_buffers_{base|range}
Nicolai Hähnle [Wed, 6 Jan 2016 22:20:57 +0000 (17:20 -0500)]
mesa: merge bind_uniform_buffers_{base|range}

Reduced code duplication should make the code more maintainable.

Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
8 years agomesa: merge bind_xfb_buffers_{base|range}
Nicolai Hähnle [Wed, 6 Jan 2016 20:47:01 +0000 (15:47 -0500)]
mesa: merge bind_xfb_buffers_{base|range}

Reduced code duplication should make the code more maintainable.

Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
8 years agoglsl: Don't add nir files to libglsl_la_SOURCES
Kristian Høgsberg Kristensen [Fri, 8 Jan 2016 23:23:56 +0000 (15:23 -0800)]
glsl: Don't add nir files to libglsl_la_SOURCES

SCons doesn't understand nir yet and doesn't want to compile the glsl to
nir pass. Move the files to their own variable so we can add it only for
automake.

Tested-by: Brian Paul <brianp@vmware.com>
8 years agonir/spirv: Add matrix determinants and inverses
Jason Ekstrand [Sat, 9 Jan 2016 00:02:06 +0000 (16:02 -0800)]
nir/spirv: Add matrix determinants and inverses

8 years agonv50,nvc0: use a face sysval to avoid the useless back-and-forth conversion
Ilia Mirkin [Fri, 8 Jan 2016 22:32:56 +0000 (17:32 -0500)]
nv50,nvc0: use a face sysval to avoid the useless back-and-forth conversion

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
8 years agoglsl: Move _mesa_shader_stage_to_string/abbrev to shader_enums.c
Kristian Høgsberg Kristensen [Fri, 8 Jan 2016 20:35:48 +0000 (12:35 -0800)]
glsl: Move _mesa_shader_stage_to_string/abbrev to shader_enums.c

These are used by code that doesn't necessarily link to libglsl.la. Move
them to shader_enums.[ch] where we keep similar helpers.

Reviewed-by: Matt Turner <mattst88@gmail.com>
8 years agoi965: Move GLSL lowering passes out of libi965_compiler.la
Kristian Høgsberg Kristensen [Fri, 8 Jan 2016 20:35:38 +0000 (12:35 -0800)]
i965: Move GLSL lowering passes out of libi965_compiler.la

The scope of libi965_compiler.la is to be able to take nir shaders and
generate i965 EU code.  As such, we don't want the GLSL IR lowering
passes in the library. With this change, libi965_compiler.la no longer
needs to link to libglsl.la.

Reviewed-by: Matt Turner <mattst88@gmail.com>
8 years agoglsl: Move glsl_to_nir files to LIBGLSL_FILES
Kristian Høgsberg Kristensen [Fri, 8 Jan 2016 20:35:18 +0000 (12:35 -0800)]
glsl: Move glsl_to_nir files to LIBGLSL_FILES

libglsl_la_SOURCES includes both NIR_FILES and LIBGLSL_FILES, so for
libglsl.la consumers, this is a no-op. libnir.la however no longer uses
any GLSL IR infrastructure and can be used without also linking to
libglsl.la.

Acked-by: Matt Turner <mattst88@gmail.com>
8 years agomesa: Use separate indices for UBO & SSBO during binding
Jordan Justen [Sat, 24 Oct 2015 00:08:33 +0000 (17:08 -0700)]
mesa: Use separate indices for UBO & SSBO during binding

Previously we were treating the binding index for Uniform Buffer
Objects and Shader Storage Buffer Objects as being part of the
combined BufferInterfaceBlocks array.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=93322
Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Juha-Pekka Heikkila <juhapekka.heikkila@gmail.com>
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
8 years agomesa: Map program UBOs and SSBOs to Interface Blocks
Jordan Justen [Sat, 24 Oct 2015 00:07:42 +0000 (17:07 -0700)]
mesa: Map program UBOs and SSBOs to Interface Blocks

v2:
 * Fill UboInterfaceBlockIndex and SsboInterfaceBlockIndex in
   split_ubos_and_ssbos (Iago)

Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Juha-Pekka Heikkila <juhapekka.heikkila@gmail.com>
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
8 years agoanv: Increate dynamic pool block size from 2k to 16k
Jordan Justen [Fri, 8 Jan 2016 20:15:29 +0000 (12:15 -0800)]
anv: Increate dynamic pool block size from 2k to 16k

This is needed because compute push constant data is replicated per
invocation. For gen7, this can be up to 64. With a push constant data
max of 128 bytes, this is 8k of data. We need additional space for
local-id payloads, so we are going with 16k for now.

Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
8 years agonir/spirv: Fix a small bug in row-major matrix loading
Jason Ekstrand [Fri, 8 Jan 2016 20:27:25 +0000 (12:27 -0800)]
nir/spirv: Fix a small bug in row-major matrix loading

8 years agomesa: docs: Add link to planet.freedesktop.org
Sarah Sharp [Thu, 29 Oct 2015 22:56:18 +0000 (15:56 -0700)]
mesa: docs: Add link to planet.freedesktop.org

The freedesktop.org blog feeds aren't mentioned on either mesa3d.org or
any of the graphics project wikis (including the DRI wiki) on
freedeskop.org.  Fix that by linking to it from the sidebar.

Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
8 years agofreedreno: add ir3_compiler to gitignore
Ilia Mirkin [Fri, 8 Jan 2016 20:09:26 +0000 (15:09 -0500)]
freedreno: add ir3_compiler to gitignore

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
8 years agogallium: add a RESQ opcode to query info about a resource
Ilia Mirkin [Mon, 14 Dec 2015 03:11:25 +0000 (22:11 -0500)]
gallium: add a RESQ opcode to query info about a resource

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agogallium: add PIPE_CAP_SHADER_BUFFER_OFFSET_ALIGNMENT
Ilia Mirkin [Sun, 3 Jan 2016 02:56:45 +0000 (21:56 -0500)]
gallium: add PIPE_CAP_SHADER_BUFFER_OFFSET_ALIGNMENT

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agogallium: add PIPE_SHADER_CAP_MAX_SHADER_BUFFERS
Ilia Mirkin [Sun, 27 Sep 2015 00:27:42 +0000 (20:27 -0400)]
gallium: add PIPE_SHADER_CAP_MAX_SHADER_BUFFERS

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agotgsi: update atomic op docs
Ilia Mirkin [Sun, 27 Sep 2015 05:23:38 +0000 (01:23 -0400)]
tgsi: update atomic op docs

Specify that the operation only applies to the x component, not
per-component as previously specified. This is unnecessary for GL and
creates additional complications for images which need to support these
operations as well.

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agotgsi: add a is_store property
Ilia Mirkin [Sat, 26 Sep 2015 21:35:41 +0000 (17:35 -0400)]
tgsi: add a is_store property

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agotgsi: provide a way to encode memory qualifiers for SSBO
Ilia Mirkin [Sat, 7 Nov 2015 07:25:20 +0000 (02:25 -0500)]
tgsi: provide a way to encode memory qualifiers for SSBO

Each load/store on most hardware can specify what caching to do. Since
SSBO allows individual variables to also have separate caching modes,
allow loads/stores to have the qualifiers instead of attempting to
encode them in declarations.

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agoureg: add buffer support to ureg
Ilia Mirkin [Sat, 19 Sep 2015 22:19:13 +0000 (18:19 -0400)]
ureg: add buffer support to ureg

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agotgsi: add ureg support for image decls
Ilia Mirkin [Sat, 20 Sep 2014 06:54:16 +0000 (02:54 -0400)]
tgsi: add ureg support for image decls

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agoglsl: Ensure 64bits shift is used.
Jose Fonseca [Fri, 8 Jan 2016 14:03:38 +0000 (14:03 +0000)]
glsl: Ensure 64bits shift is used.

I believe that `1u << x`, where x >= 32 yields undefined results
according to the C standard.

Particularly MSVC says `warning C4334: '<<' : result of 32-bit shift
implicitly converted to 64 bits (was 64-bit shift intended?)`.

Reviewed-by: Brian Paul <brianp@vmware.com>
8 years agomesa/main: Avoid `void function returning a value` warning.
Jose Fonseca [Fri, 8 Jan 2016 13:59:16 +0000 (13:59 +0000)]
mesa/main: Avoid `void function returning a value` warning.

Trivial.

Reviewed-by: Brian Paul <brianp@vmware.com>
8 years agoconfigure.ac: add --enable-profile
Oded Gabbay [Thu, 7 Jan 2016 15:20:47 +0000 (17:20 +0200)]
configure.ac: add --enable-profile

For profiling mesa's code, especially llvmpipe, PROFILE should be
defined. Currently, this define can only be generated if mesa is
built using scons.
This patch makes it possible to generate this define also when building
mesa through automake tools.

v2:

- Change --enable-llvmpipe-profile to --enable-profile
- Add -fno-omit-frame-pointer to CFLAGS and CXXFLAGS when enabling profile

Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
8 years agonir/spirv: Use create_ssa_value for block_load_store
Jason Ekstrand [Fri, 8 Jan 2016 19:50:32 +0000 (11:50 -0800)]
nir/spirv: Use create_ssa_value for block_load_store

8 years agonir/spirv: Add real support for outer products
Jason Ekstrand [Fri, 8 Jan 2016 19:38:59 +0000 (11:38 -0800)]
nir/spirv: Add real support for outer products

8 years agonir/spirv: Add support for add, subtract, and negate on matrices
Jason Ekstrand [Fri, 8 Jan 2016 19:18:47 +0000 (11:18 -0800)]
nir/spirv: Add support for add, subtract, and negate on matrices

8 years agonir/spirv: Split ALU operations out into their own file
Jason Ekstrand [Fri, 8 Jan 2016 19:02:17 +0000 (11:02 -0800)]
nir/spirv: Split ALU operations out into their own file

8 years agonine: allow fragment shader POSITION and FACE to be system values
Marek Olšák [Fri, 8 Jan 2016 01:11:16 +0000 (02:11 +0100)]
nine: allow fragment shader POSITION and FACE to be system values

Reported-by: Axel Davy <axel.davy@ens.fr>
8 years agovl: allow fragment shader POSITION to be a system value
Marek Olšák [Thu, 7 Jan 2016 22:14:55 +0000 (23:14 +0100)]
vl: allow fragment shader POSITION to be a system value

Reviewed-by: Edward O'Callaghan <eocallaghan@alterapraxis.com
Reviewed-by: Brian Paul <brianp@vmware.com>
8 years agoutil/pstipple: allow fragment shader POSITION to be a system value
Marek Olšák [Thu, 7 Jan 2016 18:48:56 +0000 (19:48 +0100)]
util/pstipple: allow fragment shader POSITION to be a system value

Reviewed-by: Edward O'Callaghan <eocallaghan@alterapraxis.com
Reviewed-by: Brian Paul <brianp@vmware.com>
8 years agost/mesa: add support for POSITION and FACE system values
Marek Olšák [Sat, 2 Jan 2016 21:45:10 +0000 (22:45 +0100)]
st/mesa: add support for POSITION and FACE system values

Reviewed-by: Edward O'Callaghan <eocallaghan@alterapraxis.com
Reviewed-by: Brian Paul <brianp@vmware.com>
8 years agotgsi/scan: update for POSITION and FACE sytem values
Marek Olšák [Fri, 8 Jan 2016 00:45:34 +0000 (01:45 +0100)]
tgsi/scan: update for POSITION and FACE sytem values

Reviewed-by: Edward O'Callaghan <eocallaghan@alterapraxis.com
Reviewed-by: Brian Paul <brianp@vmware.com>
8 years agogallium: add caps for POSITION and FACE system values
Marek Olšák [Sat, 2 Jan 2016 19:45:00 +0000 (20:45 +0100)]
gallium: add caps for POSITION and FACE system values

v2: document the integer behavior

Reviewed-by: Edward O'Callaghan <eocallaghan@alterapraxis.com
Reviewed-by: Brian Paul <brianp@vmware.com>
8 years agoprogram: add a helper for rewriting FP position input to sysval
Marek Olšák [Sat, 2 Jan 2016 22:08:27 +0000 (23:08 +0100)]
program: add a helper for rewriting FP position input to sysval

Reviewed-by: Edward O'Callaghan <eocallaghan@alterapraxis.com
Reviewed-by: Brian Paul <brianp@vmware.com>
8 years agoglsl: optionally declare gl_FragCoord & gl_FrontFacing as system values
Marek Olšák [Sat, 2 Jan 2016 19:16:16 +0000 (20:16 +0100)]
glsl: optionally declare gl_FragCoord & gl_FrontFacing as system values

Reviewed-by: Edward O'Callaghan <eocallaghan@alterapraxis.com
Reviewed-by: Brian Paul <brianp@vmware.com>
8 years agotgsi/ureg: handle redundant declarations in ureg_DECL_system_value
Marek Olšák [Thu, 7 Jan 2016 22:37:53 +0000 (23:37 +0100)]
tgsi/ureg: handle redundant declarations in ureg_DECL_system_value

Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Edward O'Callaghan <eocallaghan@alterapraxis.com>
8 years agotgsi/ureg: remove index parameter from ureg_DECL_system_value
Marek Olšák [Thu, 7 Jan 2016 22:25:48 +0000 (23:25 +0100)]
tgsi/ureg: remove index parameter from ureg_DECL_system_value

It can be trivially derived from the number of already declared system
values. This allows ureg users not to worry about which index to choose.

Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Edward O'Callaghan <eocallaghan@alterapraxis.com>
8 years agost/mesa: remove dead code from mesa_to_tgsi
Marek Olšák [Sat, 2 Jan 2016 18:58:26 +0000 (19:58 +0100)]
st/mesa: remove dead code from mesa_to_tgsi

These aren't part of ARB_fragment_program.

Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Edward O'Callaghan <eocallaghan@alterapraxis.com>
8 years agoradeon, si: Use TGSI chan name defines in lp_build_emit_fetch() calls
Edward O'Callaghan [Thu, 7 Jan 2016 16:44:46 +0000 (03:44 +1100)]
radeon, si: Use TGSI chan name defines in lp_build_emit_fetch() calls

Signed-off-by: Edward O'Callaghan <eocallaghan@alterapraxis.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
8 years agogallium/aux: Use TGSI chan name defines inplace of literals
Edward O'Callaghan [Thu, 7 Jan 2016 16:44:45 +0000 (03:44 +1100)]
gallium/aux: Use TGSI chan name defines inplace of literals

Signed-off-by: Edward O'Callaghan <eocallaghan@alterapraxis.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
8 years agomesa: check that internalformat of CopyTexImage*D is not 1, 2, 3, 4
Nicolai Hähnle [Thu, 7 Jan 2016 20:27:52 +0000 (15:27 -0500)]
mesa: check that internalformat of CopyTexImage*D is not 1, 2, 3, 4

The piglit copyteximage check has recently been augmented to test this, but
apparently it hasn't been fixed in Mesa so far.

This language also already appears in the OpenGL 2.1 spec (Ian).

Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
8 years agonir/spirv: Add support for SSBO atomics
Jason Ekstrand [Fri, 8 Jan 2016 04:44:04 +0000 (20:44 -0800)]
nir/spirv: Add support for SSBO atomics

8 years agonir/spirv: Rework UBOs and SSBOs
Jason Ekstrand [Fri, 8 Jan 2016 00:55:56 +0000 (16:55 -0800)]
nir/spirv: Rework UBOs and SSBOs

This completely reworks all block load/store operations.  In particular, it
should get row-major matrices working.

8 years agoanv/gen9: Fix cube surface state
Chad Versace [Fri, 8 Jan 2016 01:05:22 +0000 (17:05 -0800)]
anv/gen9: Fix cube surface state

For gen9 SURFTYPE_CUBE, the RENDER_SURFACE_STATE's Depth,
MinimumArrayElement, and RenderTargetViewExtent is in units of full
cubes and so must be divided by 6.

Fixes 'dEQP-VK.pipeline.image.view_type.cube_array.cube_array.*'.

Now all of 'dEQP-VK.pipeline.image.*' passes.

8 years agoanv/gen8: Refactor genX_image_view_init()
Chad Versace [Fri, 8 Jan 2016 01:02:49 +0000 (17:02 -0800)]
anv/gen8: Refactor genX_image_view_init()

Drop the temporary variables for RENDER_SURFACE_STATE's Depth and
RenderTargetViewExtent. Instead, assign them in-place.

This simplifies the next commit, which fixes gen9 cube surfaces.

8 years agovk: Make sure we emit binding table pointers after push constants
Kristian Høgsberg Kristensen [Fri, 8 Jan 2016 00:25:49 +0000 (16:25 -0800)]
vk: Make sure we emit binding table pointers after push constants

SKL needs this to make sure we flush the push constants. It gets a
little tricky, since we also need to emit binding tables before push
constants, since that may affect the push constants (dynamic buffer
offsets and storage image parameters).  This patch splits emitting
binding tables from emitting the pointers so that we can emit push
constants after binding tables but before emitting binding table
pointers.

8 years agovk: Implement VK_QUERY_RESULT_WITH_AVAILABILITY_BIT
Kristian Høgsberg Kristensen [Thu, 7 Jan 2016 05:57:24 +0000 (21:57 -0800)]
vk: Implement VK_QUERY_RESULT_WITH_AVAILABILITY_BIT

8 years agovk: Add missing DepthStallEnable to OQ pipe control
Kristian Høgsberg Kristensen [Thu, 7 Jan 2016 00:42:14 +0000 (16:42 -0800)]
vk: Add missing DepthStallEnable to OQ pipe control

8 years agovk: Issue PIPELINE_SELECT before setting up render pass
Kristian Høgsberg Kristensen [Thu, 7 Jan 2016 00:41:22 +0000 (16:41 -0800)]
vk: Issue PIPELINE_SELECT before setting up render pass

We need to make sure we're selected the 3D pipeline before we start
setting up depth and stencil buffers.

8 years agoanv/gen7: Setup state to enable barrier() function
Jordan Justen [Fri, 8 Jan 2016 01:10:02 +0000 (17:10 -0800)]
anv/gen7: Setup state to enable barrier() function

Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
8 years agoanv/gen8: Setup state to enable barrier() function
Jordan Justen [Fri, 8 Jan 2016 00:25:35 +0000 (16:25 -0800)]
anv/gen8: Setup state to enable barrier() function

Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
8 years agoi965/compiler: Enable more lowering in NIR
Jason Ekstrand [Wed, 6 Jan 2016 23:30:39 +0000 (15:30 -0800)]
i965/compiler: Enable more lowering in NIR

We don't need these for GLSL or ARB, but we need them for SPIR-V

Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
8 years agonir/algebraic: Add more lowering
Jason Ekstrand [Wed, 6 Jan 2016 23:30:38 +0000 (15:30 -0800)]
nir/algebraic: Add more lowering

This commit adds lowering options for the following opcodes:

 - nir_op_fmod
 - nir_op_bitfield_insert
 - nir_op_uadd_carry
 - nir_op_usub_borrow

Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
8 years agonir/opcodes: Fix up uadd_carry and usub_borrow
Jason Ekstrand [Wed, 6 Jan 2016 23:30:37 +0000 (15:30 -0800)]
nir/opcodes: Fix up uadd_carry and usub_borrow

Both were defined as returning bool but the gpu_shader5 functions are
defined to return int.  Also, we had the parameters for usub borrwo
backwards in the folding expression.

Reviewed-by: Matt Turner <mattst88@gmail.com>
8 years agonvc0: add ARB_indirect_parameters support
Ilia Mirkin [Sat, 2 Jan 2016 16:38:42 +0000 (11:38 -0500)]
nvc0: add ARB_indirect_parameters support

I chose to make separate macros for this due to the additional
complexity and extra scratch usage.

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
8 years agost/mesa: expose ARB_indirect_parameters when the backend driver allows
Ilia Mirkin [Thu, 31 Dec 2015 21:17:19 +0000 (16:17 -0500)]
st/mesa: expose ARB_indirect_parameters when the backend driver allows

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agomesa: add support for ARB_indirect_parameters draw functions
Ilia Mirkin [Thu, 31 Dec 2015 21:11:56 +0000 (16:11 -0500)]
mesa: add support for ARB_indirect_parameters draw functions

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agomesa: add parameter buffer, used for ARB_indirect_parameters
Ilia Mirkin [Thu, 31 Dec 2015 20:47:17 +0000 (15:47 -0500)]
mesa: add parameter buffer, used for ARB_indirect_parameters

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agoglapi: add ARB_indirect_parameters definitions
Ilia Mirkin [Thu, 31 Dec 2015 20:19:51 +0000 (15:19 -0500)]
glapi: add ARB_indirect_parameters definitions

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agonvc0: add support for real ARB_multi_draw_indirect
Ilia Mirkin [Sat, 2 Jan 2016 05:45:56 +0000 (00:45 -0500)]
nvc0: add support for real ARB_multi_draw_indirect

The draw groups are now split up into groups of 32 if there's a
non-packed stride, or in groups of 400-500 if the draw data is packed.

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
8 years agonvc0: adjust indirect draw macros to handle multiple draws at once
Ilia Mirkin [Sat, 2 Jan 2016 05:06:22 +0000 (00:06 -0500)]
nvc0: adjust indirect draw macros to handle multiple draws at once

These are still invoked one at a time, but the underlying macro can
handle multiple draws.

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
8 years agost/mesa: add support for new mesa indirect draw interface
Ilia Mirkin [Thu, 31 Dec 2015 19:11:07 +0000 (14:11 -0500)]
st/mesa: add support for new mesa indirect draw interface

This shifts all indirect draws to go through the new function. If the
driver doesn't have support for multi draws, we break those up and
perform N draws. Otherwise, we pass everything through for just a single
draw call.

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agogallium: add caps to expose support for multi indirect draws
Ilia Mirkin [Thu, 31 Dec 2015 18:30:13 +0000 (13:30 -0500)]
gallium: add caps to expose support for multi indirect draws

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agogallium: add sufficient draw interface to allow new indirect features
Ilia Mirkin [Thu, 31 Dec 2015 18:07:49 +0000 (13:07 -0500)]
gallium: add sufficient draw interface to allow new indirect features

This makes it possible to support indirect multidraws as well as having
the number of such draws to come from a separate GPU resource.

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agovbo: create a new draw function interface for indirect draws
Ilia Mirkin [Wed, 30 Dec 2015 23:10:56 +0000 (18:10 -0500)]
vbo: create a new draw function interface for indirect draws

All indirect draws are passed to the new draw function. By default
there's a fallback implementation which pipes it right back to
draw_prims, but eventually both the fallback and draw_prim's support for
indirect drawing should be removed.

This should allow a backend to properly support ARB_multi_draw_indirect
and ARB_indirect_parameters.

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Acked-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
8 years agollvmpipe: do 64bit plane calculations in the sse path
Roland Scheidegger [Sat, 2 Jan 2016 03:59:09 +0000 (04:59 +0100)]
llvmpipe: do 64bit plane calculations in the sse path

The sse path was pretty much disabled for practical purposes because the
largest allowed fb size was 128x128. So, adapt it for 64bit plane calculations.
This is actually not that difficult, though a problem is that we can't do
a signed 32x32->64bit mul, only unsigned, so need to fix that up. Overall,
the code still looks reasonable, though it's not like changes there in
setup really make much of a difference in the end...

Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
8 years agollvmpipe: don't store eo as 64bit int
Roland Scheidegger [Sat, 2 Jan 2016 03:58:37 +0000 (04:58 +0100)]
llvmpipe: don't store eo as 64bit int

eo, just like dcdx and dcdy, cannot overflow 32bit.
Store it as unsigned though just in case (it cannot be negative, but
in theory twice as big as dcdx or dcdy so this gives it one more bit).
This doesn't really change anything, albeit it might help minimally on
32bit archs.

Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
8 years agollvmpipe: use aligned data for the assembly program in setup
Roland Scheidegger [Thu, 31 Dec 2015 02:20:38 +0000 (03:20 +0100)]
llvmpipe: use aligned data for the assembly program in setup

Back in the day (before 24678700edaf5bb9da9be93a1367f1a24cfaa471) the values
were not actually in a struct but even then I can't see why we didn't simply
align the values. Especially since it's trivial to do so.
(Not that it actually matters since the code is pretty much unused for now.)

Reviewed-by: Oded Gabbay <oded.gabbay@gmail.com>
8 years agodraw: initialize prim header flags when clipping lines
Roland Scheidegger [Thu, 7 Jan 2016 18:38:15 +0000 (19:38 +0100)]
draw: initialize prim header flags when clipping lines

Otherwise, clipped lines would have undefined stippling reset bit if line
stippling is enabled.
(Untested, and I just assume copying over the bits from the original line
is actually the right thing to do.)

Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
8 years agodraw: fix line stippling with unfilled prims
Roland Scheidegger [Wed, 6 Jan 2016 22:49:30 +0000 (23:49 +0100)]
draw: fix line stippling with unfilled prims

The unfilled stage was not filling in the prim header, and the line stage
then decided to reset the stipple counter or not based on the uninitialized
data. This causes some failures in conform linestipple test (albeit quite
randomly happening depending on environment).
So fill in the prim header in the unfilled stage - I am not entirely sure
if anybody really needs determinant after that stage, but there's at least
later stages (wide line for instance) which copy over the determinant as well.

Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
8 years agoglsl: replace null check with assert
Timothy Arceri [Tue, 14 Jul 2015 13:30:27 +0000 (23:30 +1000)]
glsl: replace null check with assert

This was added in 54f583a20 since then error handling has improved.

The test this was added to fix now fails earlier since 01822706ec

Reviewed-by: Matt Turner <mattst88@gmail.com>
8 years agoi965: use _mesa_delete_buffer_object
Nicolai Hähnle [Wed, 6 Jan 2016 02:51:27 +0000 (21:51 -0500)]
i965: use _mesa_delete_buffer_object

This is more future-proof, plugs the memory leak of Label and properly
destroys the buffer mutex.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Cc: "11.0 11.1" <mesa-stable@lists.freedesktop.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
8 years agoi915: use _mesa_delete_buffer_object
Nicolai Hähnle [Wed, 6 Jan 2016 02:51:13 +0000 (21:51 -0500)]
i915: use _mesa_delete_buffer_object

This is more future-proof, plugs the memory leak of Label and properly
destroys the buffer mutex.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Cc: "11.0 11.1" <mesa-stable@lists.freedesktop.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
8 years agoradeon: use _mesa_delete_buffer_object
Nicolai Hähnle [Wed, 6 Jan 2016 02:49:37 +0000 (21:49 -0500)]
radeon: use _mesa_delete_buffer_object

This is more future-proof, plugs the memory leak of Label and properly
destroys the buffer mutex.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Cc: "11.0 11.1" <mesa-stable@lists.freedesktop.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
8 years agost/mesa: use _mesa_delete_buffer_object
Nicolai Hähnle [Wed, 6 Jan 2016 02:49:11 +0000 (21:49 -0500)]
st/mesa: use _mesa_delete_buffer_object

This is more future-proof than the current code.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Cc: "11.0 11.1" <mesa-stable@lists.freedesktop.org>
8 years agomesa/bufferobj: make _mesa_delete_buffer_object externally accessible
Nicolai Hähnle [Wed, 6 Jan 2016 02:47:04 +0000 (21:47 -0500)]
mesa/bufferobj: make _mesa_delete_buffer_object externally accessible

gl_buffer_object has grown more complicated and requires cleanup. Using this
function from drivers will be more future-proof.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Cc: "11.0 11.1" <mesa-stable@lists.freedesktop.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
8 years agoanv/meta: Fix hardcoded format size in anv_CmdCopy*
Chad Versace [Thu, 7 Jan 2016 21:46:20 +0000 (13:46 -0800)]
anv/meta: Fix hardcoded format size in anv_CmdCopy*

When looping through VkBufferImageCopy regions, for each region we
incremented the offset into the VkBuffer assuming the format size was 4.

Fixes CTS tests dEQP-VK.pipeline.image.view_type.cube_array.3d.* on
Skylake.

8 years agollvmpipe: use sse2 conv code for altivec
Oded Gabbay [Thu, 7 Jan 2016 17:50:12 +0000 (19:50 +0200)]
llvmpipe: use sse2 conv code for altivec

In lp_build_conv() and lp_build_conv_auto(), there is a special case of
conversion when sse2 is present. That code path is suitable without any
changes to altivec, because all the functions that are called in that
code path already support altivec.

This patch increase the FPS in POWER arch across the board
between 10%-25%

I checked ipers, glxgears, glxspheres64, openarena, xonotic and glmark2.

Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
8 years agoisl: Add missing break statement in array pitch calculation
Chad Versace [Thu, 7 Jan 2016 19:07:44 +0000 (11:07 -0800)]
isl: Add missing break statement in array pitch calculation

Fixes regression in ed98c374bd3f1952fbab3031afaf5ff4d178ef41.

8 years agoisl/gen9: Fix array pitch of 3d surfaces
Chad Versace [Thu, 7 Jan 2016 19:00:29 +0000 (11:00 -0800)]
isl/gen9: Fix array pitch of 3d surfaces

For tiled 3D surfaces, the array pitch must aligned to the tile height.

From the Skylake BSpec >> RENDER_SURFACE_STATE >> Surface QPitch:

   Tile Mode != Linear: This field must be set to an integer multiple of
   the tile height

Fixes CTS tests 'dEQP-VK.pipeline.image.view_type.3d.format.r8g8b8a8_unorm.*'.
Fixes Crucible tests 'func.miptree.r8g8b8a8-unorm.aspect-color.view-3d.*'.

8 years agoisl: Refactor func isl_calc_array_pitch_sa_rows
Chad Versace [Thu, 7 Jan 2016 18:58:29 +0000 (10:58 -0800)]
isl: Refactor func isl_calc_array_pitch_sa_rows

Update the function to calculate the array pitch is *element rows*, and
it rename it accordingly to isl_calc_array_pitch_el_rows.

8 years agoisl: Assert that alignments are not 0 for isl_align
Jordan Justen [Wed, 6 Jan 2016 23:42:18 +0000 (15:42 -0800)]
isl: Assert that alignments are not 0 for isl_align

Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
8 years agoanv: Assert that alignments are not 0 for align_*
Jordan Justen [Wed, 6 Jan 2016 23:40:01 +0000 (15:40 -0800)]
anv: Assert that alignments are not 0 for align_*

Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
8 years agoisl: Fix image alignment calculation
Jordan Justen [Wed, 6 Jan 2016 23:43:11 +0000 (15:43 -0800)]
isl: Fix image alignment calculation

The previous code was resulting in an alignment of 0.

Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
8 years agoradeonsi: adjust the parameters of si_shader_dump
Marek Olšák [Wed, 6 Jan 2016 01:30:13 +0000 (02:30 +0100)]
radeonsi: adjust the parameters of si_shader_dump

The function will be extended to dump all binaries shaders will consist of,
so si_shader* makes sense here.

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
8 years agoradeonsi: move si_shader_dump call out of si_compile_llvm
Marek Olšák [Sun, 3 Jan 2016 16:18:04 +0000 (17:18 +0100)]
radeonsi: move si_shader_dump call out of si_compile_llvm

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
8 years agoradeonsi: inline si_shader_binary_read
Marek Olšák [Sun, 3 Jan 2016 16:05:05 +0000 (17:05 +0100)]
radeonsi: inline si_shader_binary_read

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
8 years agoradeonsi: move si_shader_dump call out of si_shader_binary_read
Marek Olšák [Sun, 3 Jan 2016 16:03:24 +0000 (17:03 +0100)]
radeonsi: move si_shader_dump call out of si_shader_binary_read

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>