mesa.git
8 years agoradeonsi: Store inputs to memory when not using a TCS.
Bas Nieuwenhuizen [Tue, 3 May 2016 19:31:00 +0000 (21:31 +0200)]
radeonsi: Store inputs to memory when not using a TCS.

We need to copy the VS outputs to memory. I decided to do this
using a shader key, as the value depends on other shaders.

I also switch the fixed function TCS over to monolithic, as
otherwisze many of the user SGPR's need to be passed to the
epilog, which increases register pressure, or complexity to
avoid that. The main body of the fixed function TCS is not
that interesting to precompile anyway, since we do it on
demand and it is very small.

v2: Use u_bit_scan64.

Signed-off-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agoradeonsi: Add offchip buffer address calculation.
Bas Nieuwenhuizen [Mon, 9 May 2016 22:49:39 +0000 (00:49 +0200)]
radeonsi: Add offchip buffer address calculation.

Instead of creating a memory area per patch and per vertex, we put
the same attribute of every vertex & patch together. Most loads
and stores access the same attribute across all lanes, only for
different patches and vertices.

For the TCS this results in tightly packed data for 4-component
stores.

For the TES this is not the case as within a patch the loads
often also access the same vertex. However if there are < 4
vertices/patch, this still results in a reduction of the number
of cache lines. In the LDS situation we only do better than worst
case if the data per patch < 64 bytes, which due to the
tessellation factors is pretty much never.

We do not use hardware swizzling for this. It would slightly reduce
the number of executed VALU instructions, but I had issues with
increased wait times that I haven't been able to solve yet.
Furthermore, the tbuffer_store intrinsic does not support both
VGPR offset and an index, so we have a problem storing
indirectly indexed outputs. This can be solved by temporarily
storing arrays in LDS and then copying them, but I don't think
that is worth the effort. The difference in VALU cycles
hardware swizzling gives is about 0.2% of total busy cycles.
That is without handling the array case.

I chose for attributes instead of components as they are often
accessed together, and the software swizzling takes VALU cycles
for calculating offsets.

v2: - Rename functions to get_tcs_tes_buffer_address.
    - multiply by 16 as late as possible.
    - Use  tgsi_full_src_register_from_dst.
    - Remove some bad comments.

Signed-off-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agoradeonsi: Add user SGPR for the layout of the offchip buffer.
Bas Nieuwenhuizen [Mon, 9 May 2016 22:48:55 +0000 (00:48 +0200)]
radeonsi: Add user SGPR for the layout of the offchip buffer.

Signed-off-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agoradeonsi: Use correct parameter index for LS_OUT_LAYOUT.
Bas Nieuwenhuizen [Sun, 1 May 2016 18:35:40 +0000 (20:35 +0200)]
radeonsi: Use correct parameter index for LS_OUT_LAYOUT.

This happens to be in the right position, but that changes
when TCS/TES get new parameters.

Signed-off-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agoradeonsi: Add buffer load functions.
Bas Nieuwenhuizen [Mon, 2 May 2016 12:39:56 +0000 (14:39 +0200)]
radeonsi: Add buffer load functions.

v2: - Use llvm.admgcn.buffer.load instrinsics for new LLVM.
    - Code style fixes.

v3: - Code style fix.

Signed-off-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agoradeonsi: Define build_tbuffer_store_dwords earlier to support new users.
Bas Nieuwenhuizen [Mon, 2 May 2016 12:20:19 +0000 (14:20 +0200)]
radeonsi: Define build_tbuffer_store_dwords earlier to support new users.

Signed-off-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agoradeonsi: Add offchip tessellation parameters.
Bas Nieuwenhuizen [Mon, 2 May 2016 11:20:43 +0000 (13:20 +0200)]
radeonsi: Add offchip tessellation parameters.

Signed-off-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agoradeonsi: Add buffer for offchip storage between TCS and TES.
Bas Nieuwenhuizen [Mon, 2 May 2016 07:54:11 +0000 (09:54 +0200)]
radeonsi: Add buffer for offchip storage between TCS and TES.

The buffer is quite large, but should only be allocated if the
application uses tessellation. Most non-games don't.

v2: - Use the correct register for SI.
    - Add define for block size.

Signed-off-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agotgsi: fix coverity out-of-bounds warning
Rob Clark [Thu, 26 May 2016 15:11:32 +0000 (11:11 -0400)]
tgsi: fix coverity out-of-bounds warning

CID 1271532 (#1 of 1): Out-of-bounds read (OVERRUN)34. overrun-local:
Overrunning array of 2 16-byte elements at element index 2 (byte offset
32) by dereferencing pointer &inst.Dst[i].

Signed-off-by: Rob Clark <robclark@freedesktop.org>
Reviewed-by: Brian Paul <brianp@vmware.com>
8 years agotgsi: fix out of bounds access
Rob Clark [Thu, 26 May 2016 14:22:33 +0000 (10:22 -0400)]
tgsi: fix out of bounds access

Not sure why coverity calls this an out-of-bounds read vs out-of-bounds
write.

CID 1358920 (#1 of 1): Out-of-bounds read (OVERRUN)9. overrun-local:
Overrunning array r of 3 16-byte elements at element index 3 (byte
offset 48) using index chan (which evaluates to 3).

Signed-off-by: Rob Clark <robclark@freedesktop.org>
Reviewed-by: Brian Paul <brianp@vmware.com>
8 years agoi965: Don't use fast copy blit in case of logical operations other than GL_COPY
Anuj Phogat [Wed, 25 May 2016 18:33:51 +0000 (11:33 -0700)]
i965: Don't use fast copy blit in case of logical operations other than GL_COPY

XY_FAST_COPY_BLT command doesn't have a field for raster operation. So, fall
back to using XY_SRC_COPY_BLT to handle those cases.

Fixes piglit test gl-1.1-xor-copypixels when fast copy blit is enabled
for all tiling formats.

Signed-off-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
8 years agoi965/gen9: Remove the halign/valign field setup code in fast copy blit
Anuj Phogat [Sat, 12 Dec 2015 03:14:24 +0000 (19:14 -0800)]
i965/gen9: Remove the halign/valign field setup code in fast copy blit

Experimentation with different values of src/dst horizontal/vertical
alignment showed that these fileds are not used on gen9 hardware.

A recent update in graphics specs has removed these fields from
XY_FAST_COPY_BLT command.

Cc: Ben Widawsky <ben@bwidawsk.net>
Cc: Chad Versace <chad.versace@intel.com>
Signed-off-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Ben Widawsky <ben@bwidawsk.net>
8 years agonvc0: allow to monitor MP perf counters with compute shaders
Samuel Pitoiset [Wed, 25 May 2016 21:36:48 +0000 (23:36 +0200)]
nvc0: allow to monitor MP perf counters with compute shaders

To read out MP perf counters we use a compute shader and need to upload
input data like a 64-bits addr used to store the values and a sequence
ID for synchronization. Currently, this input data is uploaded as user
uniforms which means that it's sticked to c0[], but if a compute shader
from a real application is used, monitoring those performance counters
will just overwrite some data and miserably crash.

Instead, sticking the 64-bits addr and the sequence into the driver
constant buffer seems like much better and will allow to monitor
counters with GL 4.3 apps.

Tested on GF119 and GK110, but should not hurt anything on GK104.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
8 years agomesa: Move robustness code to main/robustness.c
Kristian Høgsberg Kristensen [Wed, 25 May 2016 22:29:41 +0000 (15:29 -0700)]
mesa: Move robustness code to main/robustness.c

Signed-off-by: Kristian Høgsberg Kristensen <krh@bitplanet.net>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Brian Paul <brianp@vmware.com>
8 years agodocs: Mark GL_KHR_robustness done for GLES3.2 as well
Kristian Høgsberg Kristensen [Wed, 25 May 2016 22:22:52 +0000 (15:22 -0700)]
docs: Mark GL_KHR_robustness done for GLES3.2 as well

Signed-off-by: Kristian Høgsberg Kristensen <krh@bitplanet.net>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Brian Paul <brianp@vmware.com>
8 years agoegl: Additional attribute validation for eglCreatePbufferSurface
Plamena Manolova [Wed, 25 May 2016 16:29:55 +0000 (17:29 +0100)]
egl: Additional attribute validation for eglCreatePbufferSurface

eglCreatePbufferSurface should generate an EGL_BAD_MATCH error if:
1: The EGL_TEXTURE_FORMAT attribute is EGL_NO_TEXTURE and EGL_TEXTURE_TARGET
is something other than EGL_NO_TEXTURE
2: EGL_TEXTURE_FORMAT is something other than EGL_NO_TEXTURE and
EGL_TEXTURE_TARGET is EGL_NO_TEXTURE.

This fixes the dEQP-EGL.functional.negative_api.create_pbuffer_surface test.

Signed-off-by: Plamena Manolova <plamena.manolova@intel.com>
Reviewed-by: Ben Widawsky <ben@bwidawsk.net>
8 years agogallium/radeon: add the kernel version into the renderer string
Marek Olšák [Tue, 24 May 2016 23:00:53 +0000 (01:00 +0200)]
gallium/radeon: add the kernel version into the renderer string

Example:
Gallium 0.4 on AMD TONGA (DRM 3.2.0 / 4.5.0, LLVM 3.9.0)

My kernel version is pretty long already (4.5.0-amd-01025-g32791c1)
and adding "kernel" into the string would make too it long for glxinfo
to display.

Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
8 years agowinsys/amdgpu: add back multithreaded command submission
Marek Olšák [Tue, 8 Mar 2016 00:19:31 +0000 (01:19 +0100)]
winsys/amdgpu: add back multithreaded command submission

Ported from the initial amdgpu winsys from the private AMD branch.

The thread creates the buffer list, submits IBs, and cleans up
the submission context, which can also destroy buffers.

3-5% reduction in CPU overhead is expected for apps submitting a lot
of IBs per frame. This is most visible with DMA IBs.

v2: use a semaphore instead of a busy loop in amdgpu_ws_queue_cs
    add another amdgpu_cs_sync_flush call into amdgpu_bo_map

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
8 years agogallium/tgsi: use _mesa_roundevenf in micro_rnd
Lars Hamre [Thu, 19 May 2016 21:34:00 +0000 (15:34 -0600)]
gallium/tgsi: use _mesa_roundevenf in micro_rnd

Fixes the following piglit tests (for softpipe):

/spec/glsl-1.30/execution/built-in-functions/...
fs-roundeven-float
fs-roundeven-vec2
fs-roundeven-vec3
fs-roundeven-vec4
vs-roundeven-float
vs-roundeven-vec2
vs-roundeven-vec3
vs-roundeven-vec4

/spec/glsl-1.50/execution/built-in-functions/...
gs-roundeven-float
gs-roundeven-vec2
gs-roundeven-vec3
gs-roundeven-vec4

Signed-off-by: Lars Hamre <chemecse@gmail.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
8 years ago.mailmap: use Jakob Bornecrantz's personal email
Emil Velikov [Thu, 26 May 2016 12:57:32 +0000 (13:57 +0100)]
.mailmap: use Jakob Bornecrantz's personal email

The VMware one is bouncing.

Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
8 years agonvc0: add note about where the viewport mask would go
Ilia Mirkin [Thu, 26 May 2016 04:02:57 +0000 (00:02 -0400)]
nvc0: add note about where the viewport mask would go

Not piping this all the way through yet, but no better place to note
this down. This will can be used with NV_viewport_array2.

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
8 years agonvc0: enable 32 textures on kepler+
Ilia Mirkin [Sat, 21 May 2016 23:09:32 +0000 (19:09 -0400)]
nvc0: enable 32 textures on kepler+

For fermi, this likely will require use of linked tsc mode. However on
bindless architectures, we can have as many as we want. As it stands,
the AUX_TEX_INFO has 32 teture handles reserved.

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
8 years agoglsl: add unit tests data vertex/expected outcome for uninitialized warning
Alejandro Piñeiro [Wed, 20 Apr 2016 08:02:45 +0000 (10:02 +0200)]
glsl: add unit tests data vertex/expected outcome for uninitialized warning

v2: fix 025 test. Add three more tests (Ian Romanick)

Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
8 years agoglsl: add warning-test
Alejandro Piñeiro [Tue, 19 Apr 2016 19:03:07 +0000 (21:03 +0200)]
glsl: add warning-test

It executes compiler-glsl on all the available shaders, and it checks
that the outcome is the expected.

Bash code based on the already existing optimization-test

v2: rebasing: use --version option

Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
8 years agoglsl: add just-log option for the standalone compiler.
Alejandro Piñeiro [Tue, 19 Apr 2016 18:26:32 +0000 (20:26 +0200)]
glsl: add just-log option for the standalone compiler.

Add an option in order to ask to just print the InfoLog, without any
header or separator. Useful if we want to use the standalone compiler
to track only the warning/error messages.

v2: all printfs goes on its own line (Ian Romanick)
v3: rebasing: move just_log to standalone.h/cpp

Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
8 years agoglsl: do not raise uninitialized warning with out function parameters
Alejandro Piñeiro [Tue, 19 Apr 2016 09:17:27 +0000 (11:17 +0200)]
glsl: do not raise uninitialized warning with out function parameters

It silence by default warnings with function parameters, as the
parameters need to be processed in order to have the actual and the
formal parameter, and the function signature. Then it raises the
warning if needed at verify_parameter_modes where other in/out/inout modes
checks are done.

v2: fix comment style, multi-line condition style, simplify check,
    remove extra blank (Ian Romanick)
v3: inout function parameters can raise the warning too (Ian
    Romanick)

Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
8 years agoglsl: add a empty set_is_lhs on ast_node
Alejandro Piñeiro [Tue, 19 Apr 2016 09:15:54 +0000 (11:15 +0200)]
glsl: add a empty set_is_lhs on ast_node

Just to allow to call set_is_lhs on any ast_node without a casting. Useful
when processing a ast_node list that we know it contain ast_expression.

v2: comment out new_value to avoid unused parameter warning (Ian Romanick)

Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
8 years agoglsl: handle implicit sized arrays in ssbo
Dave Airlie [Wed, 25 May 2016 03:31:41 +0000 (13:31 +1000)]
glsl: handle implicit sized arrays in ssbo

The current code disallows unsized arrays except at the end of
an SSBO but it is a bit overzealous in doing so.

struct a {
int b[];
int f[4];
};

is valid as long as b is implicitly sized within the shader,
i.e. it is accessed only by integer indices.

I've submitted some piglit tests to test for this.

This also has no regressions on piglit on my Haswell.
This fixes:
GL45-CTS.shader_storage_buffer_object.basic-syntax
GL45-CTS.shader_storage_buffer_object.basic-syntaxSSO

This patch moves a chunk of the linker code down, so
that we don't link the uniform blocks until after we've
merged all the variables. The logic went something like:

Removing the checks for last ssbo member unsized from
the compiler and into the linker, meant doing the check
in the link_uniform_blocks code. However to do that the
array sizing had to happen first, so we knew that the
only unsized arrays were in the last block. But array
sizing required the variable to be merged, otherwise
you'd get two different array sizes in different
version of two variables, and one would get lost
when merged. So the solution was to move array sizing
up, after variable merging, but before uniform block
visiting.

Reviewed-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
8 years agoglsl: fix error message on uniform block mismatch
Dave Airlie [Wed, 25 May 2016 21:42:16 +0000 (07:42 +1000)]
glsl: fix error message on uniform block mismatch

This looks like a cut-paste from above.

Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
8 years agoglsl/ast: assign explicit_xfb_buffer from correct place
Dave Airlie [Wed, 25 May 2016 23:23:54 +0000 (09:23 +1000)]
glsl/ast: assign explicit_xfb_buffer from correct place

This fixes:
GL44-CTS.tessellation_shader.tessellation_control_to_tessellation_evaluation.data_pass_through

As the OUT_TC interface structures weren't matching because
one of them had explicit_xfb_buffer set when it shouldn't.

Reviewed-by: Timothy Arceri <timothy.arceri@collabora.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
8 years agoswr: [rasterizer] Correctly select optimized primitive assembly.
Bruce Cherniak [Tue, 24 May 2016 20:00:17 +0000 (15:00 -0500)]
swr: [rasterizer] Correctly select optimized primitive assembly.

Indexed primitives were always using cut-aware primitive assembly,
whether primitive_restart was enabled or not.  Correctly pass down
primitive_restart and select optimized PA when possible.

Reviewed-by: Tim Rowley <timothy.o.rowley@intel.com>
8 years agodocs: Mention i965/gen8+ supports GL 4.2 in release notes.
Kenneth Graunke [Wed, 25 May 2016 21:22:56 +0000 (14:22 -0700)]
docs: Mention i965/gen8+ supports GL 4.2 in release notes.

8 years agodocs: Update GL_OES_copy_image status.
Kenneth Graunke [Wed, 25 May 2016 21:22:30 +0000 (14:22 -0700)]
docs: Update GL_OES_copy_image status.

8 years agoi965: Enable OES_copy_image (and EXT) on Gen8+ and Baytrail.
Kenneth Graunke [Fri, 20 May 2016 04:44:59 +0000 (21:44 -0700)]
i965: Enable OES_copy_image (and EXT) on Gen8+ and Baytrail.

For now, only enable it on platforms that actually support ETC2.

At this point, Broadwell is only failing 5 (out of 8358) dEQP tests:
dEQP-GLES31.functional.copy_image.non_compressed.viewclass_32_bits.
   srgb8_alpha8_r11f_g11f_b10f.renderbuffer_to_texture3d
   srgb8_alpha8_rgb10_a2ui.renderbuffer_to_cubemap
   srgb8_alpha8_rgb10_a2ui.renderbuffer_to_renderbuffer
   srgb8_alpha8_rgb10_a2.renderbuffer_to_texture2d
   srgb8_alpha8_rgb9_e5.renderbuffer_to_texture3d

These fail with all methods (meta, blorp, blitter, memcpy).

All are blacklisted from the Android mustpass list, which makes me
wonder whether there's an issue with the tests.  The formats in
question work with other targets, and the targets in question work
with other formats...

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Chris Forbes <chrisforbes@google.com>
8 years agoi965: Implement a BLORP path for CopyImage and prefer it over Meta.
Kenneth Graunke [Fri, 20 May 2016 04:13:29 +0000 (21:13 -0700)]
i965: Implement a BLORP path for CopyImage and prefer it over Meta.

We're dropping Meta in favor of BLORP everywhere we can.

This also fixes bugs when copying cubemaps to 2D, which is currently
broken in the meta pass.  BLORP just works.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=94198
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Chris Forbes <chrisforbes@google.com>
8 years agoi965: Make the CopyImage BLT path bail for stencil images.
Kenneth Graunke [Fri, 20 May 2016 04:10:14 +0000 (21:10 -0700)]
i965: Make the CopyImage BLT path bail for stencil images.

The BLT can't handle S8 because it's W-tiled (at least without
additional funny business, and I'm not sure we care).  Disallow
it so it falls back to the CPU path, which works.

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Chris Forbes <chrisforbes@google.com>
8 years agoi965: Also copy stencil miptree data.
Kenneth Graunke [Fri, 20 May 2016 03:50:06 +0000 (20:50 -0700)]
i965: Also copy stencil miptree data.

The Meta path handles this, but the CPU/BLT fallbacks did not.

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Chris Forbes <chrisforbes@google.com>
8 years agoi965: Make a helper function for CopyImage of a miptree.
Kenneth Graunke [Fri, 20 May 2016 03:46:22 +0000 (20:46 -0700)]
i965: Make a helper function for CopyImage of a miptree.

Currently, it only contains the BLT/CPU fallbacks, so the name is a bit
too generic.  But eventually this will use BLORP as well, at which point
the name will make more sense.

The next patch will introduce a second call.

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Chris Forbes <chrisforbes@google.com>
8 years agoi965: Combine src/dest tex vs. rb checks in intel_copy_image_sub_data.
Kenneth Graunke [Fri, 20 May 2016 03:29:04 +0000 (20:29 -0700)]
i965: Combine src/dest tex vs. rb checks in intel_copy_image_sub_data.

This simplifies things a little - now we only have one (tex or rb?)
if-ladder for src, and a second for dst, rather than four.

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Chris Forbes <chrisforbes@google.com>
8 years agoi965: Account for MinLayer in CopyImageSubData's blitter/CPU paths.
Kenneth Graunke [Fri, 20 May 2016 02:20:12 +0000 (19:20 -0700)]
i965: Account for MinLayer in CopyImageSubData's blitter/CPU paths.

Fixes Piglit's arb_copy_image-texview test with the Meta path disabled
(so we hit the blitter/CPU fallback paths).

v2: Add MinLayer even for cube maps (suggested by Ilia).

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Chris Forbes <chrisforbes@google.com>
8 years agofreedreno/ir3: cmdline compiler for glsl
Rob Clark [Sat, 14 May 2016 17:38:13 +0000 (13:38 -0400)]
freedreno/ir3: cmdline compiler for glsl

Use glsl/libstandalone.la to add support for taking glsl src files (in
addition to .tgsi) as input.  Then glsl->nir and feed the result into
the ir3 backend as normal.

Signed-off-by: Rob Clark <robclark@freedesktop.org>
8 years agoglsl: split out libstandalone
Rob Clark [Sat, 14 May 2016 15:59:26 +0000 (11:59 -0400)]
glsl: split out libstandalone

Split standalone glsl_compiler into a libstandalone.la and a thin
main.cpp.  This way drivers can re-use the glsl standalone frontend in
their own standalone compilers.

Signed-off-by: Rob Clark <robclark@freedesktop.org>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
8 years agoandroid: drop build of standalone glsl_compiler
Rob Clark [Wed, 25 May 2016 13:59:02 +0000 (09:59 -0400)]
android: drop build of standalone glsl_compiler

It's only a tool for debugging the glsl compiler, and should not be
installed.

Signed-off-by: Rob Clark <robclark@freedesktop.org>
Tested-by: Rob Herring <robh@kernel.org>
Acked-by: Emil Velikov <emil.l.velikov@gmail.com>
8 years agoi965: Mark fallthrough in switch statement.
Matt Turner [Tue, 24 May 2016 19:23:00 +0000 (12:23 -0700)]
i965: Mark fallthrough in switch statement.

Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Eric Engestrom <eric.engestrom@imgtec.com>
8 years agoi965: Assert that a depth_mt exists when using HiZ.
Matt Turner [Tue, 24 May 2016 20:06:36 +0000 (13:06 -0700)]
i965: Assert that a depth_mt exists when using HiZ.

Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Eric Engestrom <eric.engestrom@imgtec.com>
8 years agonir: Strengthen assertion that 'out' is nonnull.
Matt Turner [Tue, 24 May 2016 19:24:56 +0000 (12:24 -0700)]
nir: Strengthen assertion that 'out' is nonnull.

Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Eric Engestrom <eric.engestrom@imgtec.com>
8 years agospirv: Mark default cases unreachable().
Matt Turner [Tue, 24 May 2016 19:29:30 +0000 (12:29 -0700)]
spirv: Mark default cases unreachable().

Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Eric Engestrom <eric.engestrom@imgtec.com>
8 years agoisl: Mark default cases unreachable.
Matt Turner [Tue, 24 May 2016 20:13:04 +0000 (13:13 -0700)]
isl: Mark default cases unreachable.

Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Eric Engestrom <eric.engestrom@imgtec.com>
8 years agoisl: Remove useless qualifier from return type.
Matt Turner [Tue, 24 May 2016 19:29:51 +0000 (12:29 -0700)]
isl: Remove useless qualifier from return type.

Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Eric Engestrom <eric.engestrom@imgtec.com>
8 years agonvc0: add descriptions for hardware perf counters/metrics
Samuel Pitoiset [Wed, 25 May 2016 09:22:44 +0000 (11:22 +0200)]
nvc0: add descriptions for hardware perf counters/metrics

The GALLIUM_HUD does not yet expose a description for each events, but
this might be useful for developers who want to have a long description
of hw perf counters directly in the source code.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
8 years agomesa: 80-column wrapping for _context_lost_GetSynciv()
Brian Paul [Wed, 25 May 2016 17:49:31 +0000 (11:49 -0600)]
mesa: 80-column wrapping for _context_lost_GetSynciv()

Reviewed-by: Kristian Høgsberg Kristensen <krh@bitplanet.net>
8 years agomesa: add GLAPIENTRY to new _context_lost_X functions
Brian Paul [Wed, 25 May 2016 17:45:13 +0000 (11:45 -0600)]
mesa: add GLAPIENTRY to new _context_lost_X functions

To fix MSVC build.  Any function which goes into the dispatch table
needs to have the GLAPIENTRY (__stdcall) tag.

Reviewed-by: Kristian Høgsberg Kristensen <krh@bitplanet.net>
8 years agoscons: support 2.5.0
Giuseppe Bilotta [Wed, 25 May 2016 13:32:08 +0000 (07:32 -0600)]
scons: support 2.5.0

The get_implicit_deps changed in SCons 2.5, expecting a callable rather
than a path as third argument. Detect the SCons versions and set the
argument appropriately to support both 2.5 and earlier versions.

This closes #95211.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=95211
Signed-off-by: Giuseppe Bilotta <giuseppe.bilotta@gmail.com>
Cc: mesa-stable@lists.freedesktop.org
Acked-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
8 years agoscons: whitespace cleanup
Giuseppe Bilotta [Wed, 25 May 2016 13:30:06 +0000 (07:30 -0600)]
scons: whitespace cleanup

This text transformation was done automatically via the following shell
command:

$ find -name SCons\* -exec sed -i s/\\s\\+$// '{}' \;

Signed-off-by: Giuseppe Bilotta <giuseppe.bilotta@gmail.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
8 years agoi965/fs: take into account doubles when emitting system values
Alejandro Piñeiro [Tue, 24 May 2016 13:00:30 +0000 (15:00 +0200)]
i965/fs: take into account doubles when emitting system values

Fixes the following cts test:
GL42-CTS.vertex_attrib_64bit.limits_test

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
8 years agoi965: Fix shadowing of 'height' parameter
Kristian Høgsberg Kristensen [Wed, 25 May 2016 16:30:26 +0000 (09:30 -0700)]
i965: Fix shadowing of 'height' parameter

The nested declaration of 'height' shadows a parameter and uses
uninitialized memory. Fix by renaming to 'plane_height' which also makes
the code clearer.

This would typically break the bo size computation, but we don't use
that except when mmaping, and we don't mmap YUV buffers much.

Signed-off-by: Kristian Høgsberg Kristensen <krh@bitplanet.net>
Reported-by: Mathias Fröhlich <Mathias.Froehlich@gmx.net>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
8 years agomesa: Add .gitignore entries for make check binaries
Kristian Høgsberg Kristensen [Wed, 25 May 2016 04:07:10 +0000 (21:07 -0700)]
mesa: Add .gitignore entries for make check binaries

Signed-off-by: Kristian Høgsberg Kristensen <krh@bitplanet.net>
Acked-by: Matt Turner <mattst88@gmail.com>
8 years agoi965: Enable GL_KHR_robustness
Kristian Høgsberg Kristensen [Tue, 24 May 2016 05:49:51 +0000 (22:49 -0700)]
i965: Enable GL_KHR_robustness

GL_KHR_robustness adds the GL_CONTEXT_LOST error and five new entry
points that we already implement.  This patch adds a new dispatch table
that returns GL_CONTEXT_LOST from all entry points and implements the
GL_LOSE_CONTEXT_ON_RESET strategy by setting that table when we learn
that we've lost the context.

With the GL_CONTEXT_LOST reporting in place and dispatch for the new
entry points we can turn on GL_KHR_robustness.

Signed-off-by: Kristian Høgsberg Kristensen <krh@bitplanet.net>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Acked-by: Ilia Mirkin <imirkin@alum.mit.edu>
8 years ago.mailmap: Use Chia-I Wu personal e-mail.
Emil Velikov [Wed, 25 May 2016 16:38:06 +0000 (17:38 +0100)]
.mailmap: Use Chia-I Wu personal e-mail.

The LunarG one is bouncing.

Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
8 years ago.mailmap: Use my (Emil Velikov) personal e-mail.
Emil Velikov [Wed, 25 May 2016 16:28:22 +0000 (17:28 +0100)]
.mailmap: Use my (Emil Velikov) personal e-mail.

Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
8 years agodocs: add missing GL_OES/EXT_gpu_shader5 enablement note
Ilia Mirkin [Wed, 25 May 2016 00:03:22 +0000 (20:03 -0400)]
docs: add missing GL_OES/EXT_gpu_shader5 enablement note

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
8 years agoglsl: add GL_EXT_clip_cull_distance define, add helpers
Ilia Mirkin [Tue, 24 May 2016 23:57:47 +0000 (19:57 -0400)]
glsl: add GL_EXT_clip_cull_distance define, add helpers

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Tobias Klausmann <tobias.johannes.klausmann@mni.thm.de>
8 years agotgsi: print TGSI_PROPERTY_NEXT_SHADER value as string, not an integer
Brian Paul [Tue, 24 May 2016 23:44:30 +0000 (17:44 -0600)]
tgsi: print TGSI_PROPERTY_NEXT_SHADER value as string, not an integer

Print "GEOM" instead of "2", for example.

v2: also update the text parsing code, per Ilia.

Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agotgsi: s/6/PIPE_SHADER_TYPES/ for tgsi_processor_type_names array size
Brian Paul [Tue, 24 May 2016 23:44:08 +0000 (17:44 -0600)]
tgsi: s/6/PIPE_SHADER_TYPES/ for tgsi_processor_type_names array size

Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agonir/spirv: Handle location decorations on structure members
Jason Ekstrand [Tue, 24 May 2016 21:17:16 +0000 (14:17 -0700)]
nir/spirv: Handle location decorations on structure members

8 years agonir/spirv: Add explicit handling for all decorations
Jason Ekstrand [Tue, 24 May 2016 20:59:10 +0000 (13:59 -0700)]
nir/spirv: Add explicit handling for all decorations

From time to time we have had cases where glslang has added a decoration we
don't handle and it has caused problems.  This audit ensures that, for
every decoration, we either handle it or hit an unreachable() with an
accurate description of why we don't have to.

8 years agoi965/draw: Use the correct buffer index for interleaved VBO sizes
Jason Ekstrand [Tue, 24 May 2016 23:57:38 +0000 (16:57 -0700)]
i965/draw: Use the correct buffer index for interleaved VBO sizes

The buffer_range_* arrays are indexed by buffer index not element index.

Reviewed-by: Kristian Høgsberg <krh@bitplanet.net>
8 years agoi965/gen7: Fix gl_HelperInvocation
Jordan Justen [Tue, 24 May 2016 00:34:51 +0000 (17:34 -0700)]
i965/gen7: Fix gl_HelperInvocation

It appears that UV immediates aren't working on Ivy Bridge. In this
case, a signed version will work, and this fixes the piglit
tests/spec/glsl-4.50/execution/helper-invocation.shader_test test.

Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
8 years agomesa_glinterop: make GL interop version field bidirectional
Emil Velikov [Thu, 21 Apr 2016 16:29:16 +0000 (17:29 +0100)]
mesa_glinterop: make GL interop version field bidirectional

This allows clear and easy communication between the two.

Caller: Requesting information (struct vN)
Callee: I know how to deal with older version (vN-1) only. Here is your
data and the version I support.
Caller: Older version ? Sure I'll cap all access to the fields provided
by the older version (vN-1)

Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Tested-by: Tom Stellard <thomas.stellard@amd.com>
8 years agomesa_glinterop: drop mesa_glinterop_device_info::interop_version
Emil Velikov [Thu, 21 Apr 2016 16:16:49 +0000 (17:16 +0100)]
mesa_glinterop: drop mesa_glinterop_device_info::interop_version

One cannot use a single version to control both export_in and export_out
versions. Using this forces us to always extend/bump both structs at the
same time.

An alternative scheme is coming with next patch.

Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Tested-by: Tom Stellard <thomas.stellard@amd.com>
8 years agost/dri: add note about GL interop version checks
Emil Velikov [Tue, 3 May 2016 10:13:12 +0000 (11:13 +0100)]
st/dri: add note about GL interop version checks

... and make them more explicit.

Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Tested-by: Tom Stellard <thomas.stellard@amd.com>
8 years agomesa_glinterop: rename MESA_GLINTEROP_INVALID_{VALUE,VERSION}
Emil Velikov [Tue, 3 May 2016 10:10:54 +0000 (11:10 +0100)]
mesa_glinterop: rename MESA_GLINTEROP_INVALID_{VALUE,VERSION}

Be more explicit what it actually does.

Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Tested-by: Tom Stellard <thomas.stellard@amd.com>
8 years agomesa_glinterop: s/struct_version/version/
Emil Velikov [Thu, 21 Apr 2016 15:18:39 +0000 (16:18 +0100)]
mesa_glinterop: s/struct_version/version/

OCD polish for consistency with other mesa interfaces.

Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Tested-by: Tom Stellard <thomas.stellard@amd.com>
8 years agomesa_glinterop: fix GL interop *_VERSION comments
Emil Velikov [Thu, 21 Apr 2016 15:16:07 +0000 (16:16 +0100)]
mesa_glinterop: fix GL interop *_VERSION comments

Using the macro to set the version is wrong and ill-advised. Please don't
do it.

Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Tested-by: Tom Stellard <thomas.stellard@amd.com>
8 years agomesa_glinterop: remove inclusion of EGL header
Emil Velikov [Tue, 3 May 2016 11:25:53 +0000 (12:25 +0100)]
mesa_glinterop: remove inclusion of EGL header

Analogous to previous commit, but for EGL.

Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Tested-by: Tom Stellard <thomas.stellard@amd.com>
8 years agomesa_glinterop: remove inclusion of GLX header
Emil Velikov [Tue, 3 May 2016 11:25:34 +0000 (12:25 +0100)]
mesa_glinterop: remove inclusion of GLX header

Since we only need partial information about the GLX symbols we can
forward declare them and drop the include. Obviously each user of the
said API will needs more than what's provides, so they'll include the
GLX header.

If they don't, the compiler will give us a nice warning ;-)

Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Tested-by: Tom Stellard <thomas.stellard@amd.com>
8 years agomesa_glinterop: remove unneeded GLAPI/GLAPIENTRY/APIENTRYP symbols
Emil Velikov [Tue, 3 May 2016 11:14:26 +0000 (12:14 +0100)]
mesa_glinterop: remove unneeded GLAPI/GLAPIENTRY/APIENTRYP symbols

These come from windows.h, gl.h, glcorearb.h and/or glext.h.

The interop interface is aimed at non-Windows platforms while the macros
are used/derived due to Windows specifics. Thus we can safely remove
them.

Strictly speaking there should be GLXAPIENTRY/EGLAPIENTRY and alike
macros, although a) there is no GLX ones and b) this brings us even
further from decoupling the file from the GLX/EGL header dependency.

Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Tested-by: Tom Stellard <thomas.stellard@amd.com>
8 years agomesa_glinterop: replace GL types with their native counterpart.
Emil Velikov [Tue, 3 May 2016 11:13:43 +0000 (12:13 +0100)]
mesa_glinterop: replace GL types with their native counterpart.

Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Tested-by: Tom Stellard <thomas.stellard@amd.com>
8 years agomesa_glinterop: use generic variable types for the GL interop
Emil Velikov [Thu, 21 Apr 2016 15:36:01 +0000 (16:36 +0100)]
mesa_glinterop: use generic variable types for the GL interop

Thus we can preserve the ABI, while avoiding the inclusion of some/all
of the following:

  EGL/egl.h
  GL/gl.h
  GL/glcorearb.h
  GLES/gl.h
  GLES2/gl2.h
  GLES3/gl3.h
  GLES3/gl31.h

This will allow us to build/use it alongside any combination of APIs.

Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Tested-by: Tom Stellard <thomas.stellard@amd.com>
8 years agomesa_glinterop: use consistent naming scheme for GL interop
Emil Velikov [Thu, 21 Apr 2016 15:20:45 +0000 (16:20 +0100)]
mesa_glinterop: use consistent naming scheme for GL interop

Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Tested-by: Tom Stellard <thomas.stellard@amd.com>
8 years agoRevert "mesa: Build EGL without X11 headers after interop patchset"
Emil Velikov [Tue, 24 May 2016 13:21:31 +0000 (14:21 +0100)]
Revert "mesa: Build EGL without X11 headers after interop patchset"

This reverts commit 4e2c9a04354b6b133845b8b93c0c5d34261a91d0.

The solution was incomplete and fragile. An alternative one is coming
shortly.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Tested-by: Tom Stellard <thomas.stellard@amd.com>
8 years agodocs: Note that GL_OES_geometry_shader and GL_OES_tessellation_shader are started
Ian Romanick [Tue, 24 May 2016 19:43:18 +0000 (12:43 -0700)]
docs: Note that GL_OES_geometry_shader and GL_OES_tessellation_shader are started

The GL_OES_geometry_shader work is on the oes_shader_io_blocks branch
of idr's fd.o repository.

The GL_OES_tessellation_shader work is on the tess-gles branch
of kwg's fd.o repository.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
8 years agoc11/threads: resolve link issues with -O0
Emil Velikov [Tue, 24 May 2016 15:23:09 +0000 (16:23 +0100)]
c11/threads: resolve link issues with -O0

Add weak symbol notation for the pthread_mutexattr* symbols, thus making
the linker happy. When building with -O1 or greater the optimiser will
kick in and remove the said functions as they are dead/unreachable code.

Ideally we'll enable the optimisations locally, yet that does not seem
to work atm.

v2: Add the AX_GCC_FUNC_ATTRIBUTE([weak]) hunk in configure.

Cc: Alejandro Piñeiro <apinheiro@igalia.com>
Cc: Ben Widawsky <ben@bwidawsk.net>
Cc: Ilia Mirkin <imirkin@alum.mit.edu>
Cc: Rob Herring <robh@kernel.org>
Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Tested-by: Rob Clark <robdclark@gmail.com>
Tested-by: Mark Janes <mark.a.janes@intel.com>
8 years agoswr: [rasterizer] remove containers.hpp
Tim Rowley [Fri, 20 May 2016 00:08:53 +0000 (18:08 -0600)]
swr: [rasterizer] remove containers.hpp

Reviewed-by: Bruce Cherniak <bruce.cherniak@intel.com>
8 years agoswr: [rasterizer core] remove utility dead code
Tim Rowley [Thu, 19 May 2016 22:23:07 +0000 (16:23 -0600)]
swr: [rasterizer core] remove utility dead code

Reviewed-by: Bruce Cherniak <bruce.cherniak@intel.com>
8 years agoswr: [rasterizer core] buckets fixes
Tim Rowley [Tue, 17 May 2016 23:26:27 +0000 (17:26 -0600)]
swr: [rasterizer core] buckets fixes

1. Don't clear bucket descriptions to fix issues with sim level
   buckets getting out of sync.

2. Close out threadviz file descriptors in ClearThreads().

3. Skip buckets for jitter based buckets when multithreaded. We need
   thread local storage through llvm jit functions to be fixed before
   we can enable this.

4. Fix buckets StopCapture to correctly detect capture complete.

Reviewed-by: Bruce Cherniak <bruce.cherniak@intel.com>
8 years agoswr: [rasterizer core] move centroid setup out of CalcCentroidBarycentrics
Tim Rowley [Tue, 17 May 2016 22:32:08 +0000 (16:32 -0600)]
swr: [rasterizer core] move centroid setup out of CalcCentroidBarycentrics

Reviewed-by: Bruce Cherniak <bruce.cherniak@intel.com>
8 years agoswr: [rasterizer jitter] implement InstanceID/VertexID in fetch jit
Tim Rowley [Fri, 20 May 2016 16:15:43 +0000 (11:15 -0500)]
swr: [rasterizer jitter] implement InstanceID/VertexID in fetch jit

Reviewed-by: Bruce Cherniak <bruce.cherniak@intel.com>
8 years agomesa: Silence unused parameter warnings
Ian Romanick [Thu, 19 May 2016 01:26:34 +0000 (18:26 -0700)]
mesa: Silence unused parameter warnings

Neither shProg nor name was used.  Remove them both.

main/shader_query.cpp:779:53: warning: unused parameter ‘shProg’ [-Wunused-parameter]
 program_resource_location(struct gl_shader_program *shProg,
                                                     ^
main/shader_query.cpp:780:72: warning: unused parameter ‘name’ [-Wunused-parameter]
                           struct gl_program_resource *res, const char *name,
                                                                        ^

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Timothy Arceri <timothy.arceri@collabora.com>
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
8 years agoglsl/linker: Silence unused parameter warning
Ian Romanick [Thu, 19 May 2016 19:06:55 +0000 (12:06 -0700)]
glsl/linker: Silence unused parameter warning

The parameter is required for the interface.

glsl/link_uniforms.cpp:689:61: warning: unused parameter ‘record_type’ [-Wunused-parameter]
                             bool row_major, const glsl_type *record_type,
                                                             ^

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Timothy Arceri <timothy.arceri@collabora.com>
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
8 years agodri: Add YVU formats
Kristian Høgsberg Kristensen [Wed, 27 Apr 2016 22:00:54 +0000 (15:00 -0700)]
dri: Add YVU formats

Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
8 years agoi965: Allow creating planar YUV __DRIimages
Kristian Høgsberg Kristensen [Mon, 2 May 2016 04:25:35 +0000 (21:25 -0700)]
i965: Allow creating planar YUV __DRIimages

Lift the resctriction we had before and allow creation of images with
multiple planes. We still require all the planes to be within the same
bo.

Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Chad Versace <chad.versace@intel.com>
8 years agoi965: Invoke lowering pass for YUV textures
Kristian Høgsberg Kristensen [Mon, 2 May 2016 04:22:54 +0000 (21:22 -0700)]
i965: Invoke lowering pass for YUV textures

Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
8 years agoi965: Support textures with multiple planes
Kristian Høgsberg Kristensen [Mon, 2 May 2016 04:20:02 +0000 (21:20 -0700)]
i965: Support textures with multiple planes

Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
8 years agoi965: Create multiple miptrees for planar YUV images
Kristian Høgsberg Kristensen [Mon, 2 May 2016 04:24:00 +0000 (21:24 -0700)]
i965: Create multiple miptrees for planar YUV images

Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
8 years agoi965: Refactor intel_set_texture_image_bo() to create_mt_for_dri_image()
Kristian Høgsberg Kristensen [Tue, 10 May 2016 05:53:47 +0000 (22:53 -0700)]
i965: Refactor intel_set_texture_image_bo() to create_mt_for_dri_image()

This function now only creates the mt and we then call
intel_set_texture_image_mt() in intel_image_target_texture_2d() to set
it for the texture image.

Reviewed-by: Chad Versace <chad.versace@intel.com>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
8 years agoi965: Use intel_set_texture_image_mt() in intelSetTexBuffer2()
Kristian Høgsberg Kristensen [Tue, 10 May 2016 00:02:45 +0000 (17:02 -0700)]
i965: Use intel_set_texture_image_mt() in intelSetTexBuffer2()

Create the mt for the drawable bo directly and call our new
intel_miptree_create_for_bo() helper instead.

Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Chad Versace <chad.versace@intel.com>
8 years agoi965: Add new intel_set_texture_image_mt() helper
Kristian Høgsberg Kristensen [Mon, 9 May 2016 23:53:06 +0000 (16:53 -0700)]
i965: Add new intel_set_texture_image_mt() helper

This factors out the work of setting up a miptree as the backing for a
texture image into a new helper.

Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Chad Versace <chad.versace@intel.com>
8 years agonir: Add a lowering pass for YUV textures
Kristian Høgsberg Kristensen [Mon, 2 May 2016 04:13:37 +0000 (21:13 -0700)]
nir: Add a lowering pass for YUV textures

This lowers sampling from YUV textures to 1) one or more texture
instructions to sample each plane and 2) color space conversion to RGB.

Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
8 years agonir: Handle NULL in nir_copy_deref()
Kristian Høgsberg Kristensen [Tue, 10 May 2016 22:08:38 +0000 (15:08 -0700)]
nir: Handle NULL in nir_copy_deref()

Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>