platform/upstream/mesa.git
5 years agoradeonsi: add radeonsi_sync_compile option
Nicolai Hähnle [Mon, 25 Mar 2019 14:44:45 +0000 (15:44 +0100)]
radeonsi: add radeonsi_sync_compile option

Force the driver thread to sync immediately with a compiler thread (but
compilation still happens in a separate thread).

This can be useful to simplify debugging compiler issues.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
5 years agoradeonsi: add radeonsi_aux_debug option for aux context debug dumps
Nicolai Hähnle [Thu, 14 Mar 2019 08:51:43 +0000 (09:51 +0100)]
radeonsi: add radeonsi_aux_debug option for aux context debug dumps

Enabling this option will create ddebug-style dumps for the aux context,
except that instead of intercepting the pipe_context layer
we just dump the IB contents on flush.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
5 years agoddebug: expose some helper functions as non-inline
Nicolai Hähnle [Thu, 14 Mar 2019 08:48:47 +0000 (09:48 +0100)]
ddebug: expose some helper functions as non-inline

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
5 years agoddebug: dump driver state into a separate file
Nicolai Hähnle [Tue, 26 Feb 2019 15:34:34 +0000 (16:34 +0100)]
ddebug: dump driver state into a separate file

Due to asynchronous execution, it's not clear which of the draws the state
may refer to.

This also works around an issue encountered with radeonsi where dumping
the driver state itself caused a hang.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
5 years agoddebug: log calls to pipe->flush
Nicolai Hähnle [Tue, 26 Feb 2019 15:22:53 +0000 (16:22 +0100)]
ddebug: log calls to pipe->flush

This can be useful when internal draws lead to a hang.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
5 years agoddebug: set thread name
Nicolai Hähnle [Tue, 26 Feb 2019 12:07:30 +0000 (13:07 +0100)]
ddebug: set thread name

For better debuggability.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
5 years agoutil/u_log: flush auto loggers before starting a new page
Nicolai Hähnle [Tue, 26 Feb 2019 15:22:02 +0000 (16:22 +0100)]
util/u_log: flush auto loggers before starting a new page

Without this, command stream dumps of radeonsi may misleadingly end up
in a later page.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
5 years agoradeonsi: add si_debug_options for convenient adding/removing of options
Nicolai Hähnle [Fri, 15 Mar 2019 13:56:36 +0000 (14:56 +0100)]
radeonsi: add si_debug_options for convenient adding/removing of options

Move the definition of radeonsi_clear_db_cache_before_clear there,
as well as radeonsi_enable_nir.

This removes the AMD_DEBUG=nir option.

We currently still have two places for options: the driconf machinery
and AMD_DEBUG/R600_DEBUG. If we are to have a single place for options,
then the driconf machinery should be preferred since it's more flexible.

The only downside of the driconf machinery was that adding new options
was quite inconvenient. With this change, a simple boolean option can
be added with a single line of code, same as for AMD_DEBUG.

One technical limitation of this particular implementation is that while
almost all driconf features are available, the translation machinery doesn't
pick up the description strings for options added in si_debvug_options. In
practice, translations haven't been provided anyway, and this is intended
for developer options, so I'm not too worried. It could always be added
later if anybody really cares.

v2:
- use bool instead of uint8_t for options
- si_debug_options.inc -> si_debug_options.h

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
5 years agogitlab-ci: Use meson buildtype debug instead of default debugoptimized
Michel Dänzer [Wed, 17 Apr 2019 10:53:23 +0000 (12:53 +0200)]
gitlab-ci: Use meson buildtype debug instead of default debugoptimized

This can save a lot of time for some of the meson CI jobs.

Acked-by: Eric Engestrom <eric.engestrom@intel.com>
5 years agoRevert "intel/compiler: split is_partial_write() into two variants"
Juan A. Suarez Romero [Wed, 24 Apr 2019 10:38:28 +0000 (12:38 +0200)]
Revert "intel/compiler: split is_partial_write() into two variants"

This reverts commit 40b3abb4d16af4cef0307e1b4904c2ec0924299e.

It is not clear that this commit was entirely correct, and unfortunately
it was pushed by error.

CC: Jason Ekstrand <jason@jlekstrand.net>
Acked-by: Jason Ekstrand <jason@jlekstrand.net>
5 years agonir: fix nir_remove_unused_varyings()
Timothy Arceri [Thu, 25 Apr 2019 01:17:42 +0000 (11:17 +1000)]
nir: fix nir_remove_unused_varyings()

We were only setting the used mask for the first component of a
varying. Since the linking opts split vectors into scalars this
has mostly worked ok.

However this causes an issue where for example if we split a
struct on one side of the interface but not the other, then we
can possibly end up removing the first components on the side
that was split and then incorrectly remove the whole struct
on the other side of the varying.

With this change we simply mark all 4 components for each slot
used by a struct. We could possibly make this more fine gained
but that would require a more complex change.

This fixes a bug in Strange Brigade on RADV when tessellation
is enabled, all credit goes to Samuel Pitoiset for tracking down
the cause of the bug.

Fixes: f1eb5e639997 ("nir: add component level support to remove_unused_io_vars()")

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
5 years agoi965: fix icelake performance query enabling
Lionel Landwerlin [Wed, 24 Apr 2019 21:27:34 +0000 (05:27 +0800)]
i965: fix icelake performance query enabling

This was a rebase issue which lost of change to a file moved from i965
to src/intel/perf.

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Fixes: 134e750e16bfc5 ("i965: extract performance query metrics")
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
5 years agoradeonsi: add BOs after need_cs_space
Marek Olšák [Wed, 24 Apr 2019 21:33:53 +0000 (17:33 -0400)]
radeonsi: add BOs after need_cs_space

need_cs_space may clear the buffer list.

Fixes: 951d60f8cdc88 "radeonsi: delay adding BOs at the beginning of IBs until the first draw"

Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
5 years agoglsl: handle interactions between EXT_gpu_shader4 and texture extensions
Marek Olšák [Wed, 24 Apr 2019 17:02:43 +0000 (13:02 -0400)]
glsl: handle interactions between EXT_gpu_shader4 and texture extensions

also, EXT_texture_buffer_object has to be enabled separately.

Reviewed-by: Eric Anholt <eric@anholt.net>
5 years agost/mesa: expose EXT_gpu_shader4 if GLSL 1.40 is supported
Marek Olšák [Tue, 7 Aug 2018 22:32:31 +0000 (18:32 -0400)]
st/mesa: expose EXT_gpu_shader4 if GLSL 1.40 is supported

Tested-by: Dieter Nützel <Dieter@nuetzel-hh.de>
Reviewed-by: Eric Anholt <eric@anholt.net>
5 years agomesa: only allow EXT_gpu_shader4 in the compatibility profile
Marek Olšák [Tue, 21 Aug 2018 03:42:22 +0000 (23:42 -0400)]
mesa: only allow EXT_gpu_shader4 in the compatibility profile

Tested-by: Dieter Nützel <Dieter@nuetzel-hh.de>
Reviewed-by: Eric Anholt <eric@anholt.net>
5 years agomesa: expose EXT_texture_buffer_object
Marek Olšák [Wed, 8 Aug 2018 04:06:51 +0000 (00:06 -0400)]
mesa: expose EXT_texture_buffer_object

This is needed for exposing the samplerBuffer functions under
EXT_gpu_shader4.

v2: - expose it in the compat profile only
    - make it an alias of EXT_gpu_shader4

Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com> (v1)
Tested-by: Dieter Nützel <Dieter@nuetzel-hh.de>
Reviewed-by: Eric Anholt <eric@anholt.net>
5 years agoglsl: allow "varying out" for fragment shader outputs with EXT_gpu_shader4
Marek Olšák [Fri, 5 Apr 2019 21:01:55 +0000 (17:01 -0400)]
glsl: allow "varying out" for fragment shader outputs with EXT_gpu_shader4

Reviewed-by: Eric Anholt <eric@anholt.net>
5 years agoglsl: add texture builtin functions for EXT_gpu_shader4
Marek Olšák [Wed, 8 Aug 2018 01:31:09 +0000 (21:31 -0400)]
glsl: add texture builtin functions for EXT_gpu_shader4

v2: some fixes to texture functions thanks to piglit tests

Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com> (v1)
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com> (v1)
Tested-by: Dieter Nützel <Dieter@nuetzel-hh.de> (v1)
Reviewed-by: Eric Anholt <eric@anholt.net>
5 years agoglsl: add arithmetic builtin functions for EXT_gpu_shader4
Marek Olšák [Wed, 8 Aug 2018 01:31:09 +0000 (21:31 -0400)]
glsl: add arithmetic builtin functions for EXT_gpu_shader4

Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Tested-by: Dieter Nützel <Dieter@nuetzel-hh.de>
Reviewed-by: Eric Anholt <eric@anholt.net>
5 years agoglsl: add builtin variables for EXT_gpu_shader4
Marek Olšák [Tue, 7 Aug 2018 22:30:19 +0000 (18:30 -0400)]
glsl: add builtin variables for EXT_gpu_shader4

Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Tested-by: Dieter Nützel <Dieter@nuetzel-hh.de>
Reviewed-by: Eric Anholt <eric@anholt.net>
5 years agoglsl: apply some 1.30 and other rules to EXT_gpu_shader4 as well
Marek Olšák [Tue, 7 Aug 2018 23:56:44 +0000 (19:56 -0400)]
glsl: apply some 1.30 and other rules to EXT_gpu_shader4 as well

Reviewed-by: Eric Anholt <eric@anholt.net>
5 years agoglsl: enable types for EXT_gpu_shader4
Chris Forbes [Thu, 18 Jul 2013 10:41:21 +0000 (22:41 +1200)]
glsl: enable types for EXT_gpu_shader4

Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
Tested-by: Dieter Nützel <Dieter@nuetzel-hh.de>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
5 years agoglsl: add `unsigned int` type for EXT_GPU_shader4
Marek Olšák [Tue, 7 Aug 2018 21:18:40 +0000 (17:18 -0400)]
glsl: add `unsigned int` type for EXT_GPU_shader4

Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Tested-by: Dieter Nützel <Dieter@nuetzel-hh.de>
Reviewed-by: Eric Anholt <eric@anholt.net>
5 years agoglsl: enable noperspective|flat|centroid for EXT_gpu_shader4
Chris Forbes [Thu, 18 Jul 2013 10:43:26 +0000 (22:43 +1200)]
glsl: enable noperspective|flat|centroid for EXT_gpu_shader4

Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Tested-by: Dieter Nützel <Dieter@nuetzel-hh.de>
Reviewed-by: Eric Anholt <eric@anholt.net>
5 years agoglsl: add scaffolding for EXT_gpu_shader4
Chris Forbes [Thu, 18 Jul 2013 09:44:58 +0000 (21:44 +1200)]
glsl: add scaffolding for EXT_gpu_shader4

Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Tested-by: Dieter Nützel <Dieter@nuetzel-hh.de>
Reviewed-by: Eric Anholt <eric@anholt.net>
5 years agomesa: enable glGet for EXT_gpu_shader4
Marek Olšák [Tue, 7 Aug 2018 21:49:42 +0000 (17:49 -0400)]
mesa: enable glGet for EXT_gpu_shader4

Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Tested-by: Dieter Nützel <Dieter@nuetzel-hh.de>
Reviewed-by: Eric Anholt <eric@anholt.net>
5 years agov3d: Disable SSBOs and atomic counters on vertex shaders.
Eric Anholt [Mon, 22 Apr 2019 18:24:55 +0000 (11:24 -0700)]
v3d: Disable SSBOs and atomic counters on vertex shaders.

The CTS fails on
dEQP-GLES31.functional.shaders.opaque_type_indexing.atomic_counter.*vertex
when they are enabled, due to the VS being run for both bin and render.  I
think this behavior is expected to be valid, but I can't find text in
atomic counters or SSBO specs saying so (the closed I found was in
shader_image_load_store).  Just disable it for now, since the closed
source driver doesn't expose vertex atomic counters/SSBOs either.

5 years agost/mesa: Don't set atomic counter size != 0 if MAX_SHADER_BUFFERS == 0.
Eric Anholt [Wed, 1 Aug 2018 23:07:45 +0000 (16:07 -0700)]
st/mesa: Don't set atomic counter size != 0 if MAX_SHADER_BUFFERS == 0.

This is just asking for tests to get confused about the HW supporting
atomics in this shader stage or not, such as
dEQP-GLES31.functional.shaders.opaque_type_indexing.atomic_counter.const_expression_vertex.

v2: Rebase on the other atomic cleanups that have happened since posting.
v3: Commit message tweak by Marek.

Signed-off-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
5 years agoiris: Advertise EXT_texture_sRGB_R8 support
Kenneth Graunke [Wed, 24 Apr 2019 20:24:18 +0000 (13:24 -0700)]
iris: Advertise EXT_texture_sRGB_R8 support

Using the luminance format, like both brw and anv do.

5 years agoiris: Enable GL_AMD_depth_clamp_separate
Kenneth Graunke [Wed, 24 Apr 2019 20:04:53 +0000 (13:04 -0700)]
iris: Enable GL_AMD_depth_clamp_separate

We support this, we just forgot to turn it on.

5 years agoutil: fix a compile failure in u_compute.c on windows
Marek Olšák [Wed, 24 Apr 2019 22:29:26 +0000 (18:29 -0400)]
util: fix a compile failure in u_compute.c on windows

5 years agoiris: enable preemption support for gen10
Mike Blumenkrantz [Fri, 19 Apr 2019 13:04:59 +0000 (09:04 -0400)]
iris: enable preemption support for gen10

this automatically enables preemption on gen10 where it is disabled by
default but still available

Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Rafael Antognolli <rafael.antognolli@intel.com>
5 years agoiris: add preemption support on gen9
Mike Blumenkrantz [Thu, 11 Apr 2019 14:07:15 +0000 (10:07 -0400)]
iris: add preemption support on gen9

this is basically just porting the following two commits to gallium:
d8b50e152a0d5df0971c05b8db132fa688794001
5c454661c66fa2624cf4bba1071175070724869a

resolves kwg/mesa#49

Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Rafael Antognolli <rafael.antognolli@intel.com>
5 years agoiris: Split iris_flush_and_dirty_for_history into two helpers.
Kenneth Graunke [Wed, 24 Apr 2019 19:11:39 +0000 (12:11 -0700)]
iris: Split iris_flush_and_dirty_for_history into two helpers.

We create two new helpers, iris_flush_bits_for_history, and
iris_dirty_for_history, then use them in the existing function.

The first accumulates flush bits based on res->bind_history, but doesn't
actually perform a flush.  This allows us to accumulate flush bits by
looping over multiple resources, but ultimately emit a single flush for
all of them.

The latter flags dirty bits without flushing, which again allows us to
handle multiple resources, but also is more convenient when writing from
the CPU where we don't need a flush (as in commit 4d12236072).

5 years agointel/compiler: fix uninit non-static variable. (v2)
Dave Airlie [Thu, 11 Apr 2019 10:31:59 +0000 (20:31 +1000)]
intel/compiler: fix uninit non-static variable. (v2)

Pointed out by coverity.

v2: init nir_locals also.

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
5 years agovirgl/drm: insert correct handles into the table. (v3)
Dave Airlie [Tue, 9 Apr 2019 05:00:48 +0000 (15:00 +1000)]
virgl/drm: insert correct handles into the table. (v3)

This inserts a handle for the flink name and a handle the correct
gem handle for the bo.

v2: fix handles/names confusion (Lepton Wu)
v3: set flink name correctly (Lepton Wu)

Reviewed-by: Chia-I Wu <olvaffe@gmail.com>
5 years agovirgl/drm: handle flink name better.
Dave Airlie [Tue, 9 Apr 2019 04:54:27 +0000 (14:54 +1000)]
virgl/drm: handle flink name better.

This realigns this code with code from radeon.

Reviewed-by: Chia-I Wu <olvaffe@gmail.com>
5 years agovirgl/drm: cleanup buffer from handle creation (v2)
Dave Airlie [Tue, 9 Apr 2019 04:49:01 +0000 (14:49 +1000)]
virgl/drm: cleanup buffer from handle creation (v2)

This cleans up and realigns this code with what is in radeon

v2: fix names->handles (Lepton Wu)

Reviewed-by: Chia-I Wu <olvaffe@gmail.com>
5 years agoiris: Actually put Mesa in GL_RENDERER string
Kenneth Graunke [Wed, 24 Apr 2019 19:53:30 +0000 (12:53 -0700)]
iris: Actually put Mesa in GL_RENDERER string

I constructed the right thing and then returned the other one.

5 years agova: use a compute shader for the blit
Jiang, Sonny [Tue, 2 Apr 2019 17:44:12 +0000 (17:44 +0000)]
va: use a compute shader for the blit

Signed-off-by: Sonny Jiang <sonny.jiang@amd.com>
Signed-off-by: Marek Olšák <marek.olsak@amd.com>
5 years agogallium: add PIPE_CAP_PREFER_COMPUTE_BLIT_FOR_MULTIMEDIA
Marek Olšák [Thu, 18 Apr 2019 21:01:52 +0000 (17:01 -0400)]
gallium: add PIPE_CAP_PREFER_COMPUTE_BLIT_FOR_MULTIMEDIA

5 years agodocs: update calendar, and news item and link release notes for 19.0.3
Dylan Baker [Wed, 24 Apr 2019 17:52:20 +0000 (10:52 -0700)]
docs: update calendar, and news item and link release notes for 19.0.3

5 years agodocs: Add SHA256 sums for mesa 19.0.3
Dylan Baker [Wed, 24 Apr 2019 17:44:27 +0000 (10:44 -0700)]
docs: Add SHA256 sums for mesa 19.0.3

5 years agodocs: add relnotes for 19.0.3
Dylan Baker [Wed, 24 Apr 2019 17:39:04 +0000 (10:39 -0700)]
docs: add relnotes for 19.0.3

5 years agogallium: set PIPE_CAP_MAX_FRAMES_IN_FLIGHT to 2 for all drivers
Marek Olšák [Wed, 24 Apr 2019 01:23:22 +0000 (21:23 -0400)]
gallium: set PIPE_CAP_MAX_FRAMES_IN_FLIGHT to 2 for all drivers

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
5 years agointel/isl: Resize clear color buffer to full cacheline
Rafael Antognolli [Tue, 16 Apr 2019 13:31:06 +0000 (16:31 +0300)]
intel/isl: Resize clear color buffer to full cacheline

Fixes MCS fast clear gpu hangs with Vulkan CTS on ICL in CI.

v2 (Nanley): In the title s/Align/Resize/

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Nanley Chery <nanley.g.chery@intel.com>
Tested-by: Topi Pohjolainen <topi.pohjolainen@intel.com>
Signed-off-by: Rafael Antognolli <rafael.antognolli@intel.com>
5 years agoanv/descriptor_set: Properly align descriptor buffer to a page
Jason Ekstrand [Wed, 24 Apr 2019 02:40:31 +0000 (21:40 -0500)]
anv/descriptor_set: Properly align descriptor buffer to a page

Instead of aligning and then taking inline uniforms into account, we
need to take inline uniforms into account and then align to a page.
Otherwise, we may not be aligned to a page and allocation may fail.

Fixes: 43f40dc7cb2 "anv: Implement VK_EXT_inline_uniform_block"
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
5 years agoanv/descriptor_set: Only vma_heap_finish if we have a descriptor buffer
Jason Ekstrand [Wed, 24 Apr 2019 02:35:30 +0000 (21:35 -0500)]
anv/descriptor_set: Only vma_heap_finish if we have a descriptor buffer

Fixes: 7bb34ecff98 "anv: release memory allocated by bo_heap when..."
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
5 years agoanv/descriptor_set: Destroy sets before pool finalization
Jason Ekstrand [Wed, 24 Apr 2019 02:46:32 +0000 (21:46 -0500)]
anv/descriptor_set: Destroy sets before pool finalization

Fixes: 105002bd2d "anv: destroy descriptor sets when pool gets..."
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
5 years agoanv/descriptor_set: Unlink sets from the pool in set_destroy
Jason Ekstrand [Wed, 24 Apr 2019 03:13:55 +0000 (22:13 -0500)]
anv/descriptor_set: Unlink sets from the pool in set_destroy

anv_descriptor_pool_free_set is called on the clean-up path of
anv_descriptor_set_create and the set may not have been added to the
pool's list of sets yet.  While we're here, we move adding it to that
list into set_create for symmetry.

Fixes: 105002bd2d "anv: destroy descriptor sets when pool gets..."
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
5 years agoandroid/iris: fix driinfo header filename
Tapani Pälli [Tue, 23 Apr 2019 09:52:55 +0000 (12:52 +0300)]
android/iris: fix driinfo header filename

Fixes iris driver Android build.

Fixes: faa52e328e3 "iris: Add mechanism for iris-specific driconf options"
Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
5 years agointel/fs: Fix D to W conversion in opt_combine_constants
Ian Romanick [Fri, 19 Apr 2019 18:32:59 +0000 (11:32 -0700)]
intel/fs: Fix D to W conversion in opt_combine_constants

Found by GCC warning:

src/intel/compiler/brw_fs_combine_constants.cpp: In function ‘bool needs_negate(const fs_reg*, const imm*)’:
src/intel/compiler/brw_fs_combine_constants.cpp:306:34: warning: comparison of unsigned expression < 0 is always false [-Wtype-limits]
       return ((reg->d & 0xffffu) < 0) != (imm->w < 0);
               ~~~~~~~~~~~~~~~~~~~^~~

The result of the bit-and is a 32-bit value with the top bits all zero.
This will never be < 0.  Instead of masking off the bits, just cast to
int16_t and let the compiler handle the actual conversion.

Fixes: e64be391dd0 ("intel/compiler: generalize the combine constants pass")
Cc: Iago Toral Quiroga <itoral@igalia.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
5 years agopanfrost/midgard: Remove assembler
Alyssa Rosenzweig [Wed, 24 Apr 2019 02:43:23 +0000 (02:43 +0000)]
panfrost/midgard: Remove assembler

This code is outdated and unused; now that the compiler is mature,
there's no point keeping it around in-tree (or at all).

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
5 years agopanfrost: Adds Bifrost shader disassembler utility
Ryan Houdek [Mon, 22 Apr 2019 03:41:09 +0000 (20:41 -0700)]
panfrost: Adds Bifrost shader disassembler utility

This code is stable and can live upstream independently while the rest
of the Bifrost stack comes up.

v2: Added a verbose flag to hide away some of the more verbose features
that nobody really needs

[The Bifrost disassembler is written by Connor Abbott, Lyude Paul, and
Ryan Houdek.]

Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
5 years agopanfrost/midgard: Add "op commutes?" property
Alyssa Rosenzweig [Wed, 24 Apr 2019 02:18:28 +0000 (02:18 +0000)]
panfrost/midgard: Add "op commutes?" property

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
5 years agopanfrost/midgard: Refactor opcode tables
Alyssa Rosenzweig [Wed, 24 Apr 2019 01:15:15 +0000 (01:15 +0000)]
panfrost/midgard: Refactor opcode tables

We create an all-encompassing opcode table for handling name and
properties, removing a number of ad hoc opcode tables which became
brittle and quickly out of date. While we're at it, we fix some
incorrect opcodes relating to ball/bany, and move a small function out
to midgard_compile.c. Together these changes should allow compilation
without warnings, along with helping the codebase health considerably.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
5 years agopanfrost/midgard: Optimize MIR in progress loop
Alyssa Rosenzweig [Mon, 22 Apr 2019 04:58:53 +0000 (04:58 +0000)]
panfrost/midgard: Optimize MIR in progress loop

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
5 years agopanfrost/midgard: Implement copy propagation
Alyssa Rosenzweig [Mon, 22 Apr 2019 04:56:25 +0000 (04:56 +0000)]
panfrost/midgard: Implement copy propagation

Most copy prop should occur at the NIR level, but we generate a fair
number of moves implicitly ourselves, etc... long story short, it's a
net win to also do simple copy prop + DCE on the MIR. As a bonus, this
fixes the weird imov precision bug once and for good, I think.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
5 years agopanfrost/midgard: Set integer mods
Alyssa Rosenzweig [Mon, 22 Apr 2019 03:25:42 +0000 (03:25 +0000)]
panfrost/midgard: Set integer mods

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
5 years agopanfrost/midgard: Document sign-extension/zero-extension bits (vector)
Alyssa Rosenzweig [Mon, 22 Apr 2019 03:08:25 +0000 (03:08 +0000)]
panfrost/midgard: Document sign-extension/zero-extension bits (vector)

For floating point ops, these bits determine the "negate?" and "abs?"
modifiers. For integer ops, it turns out they control how sign/zero
extension work, useful for mixing types.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
5 years agopanfrost/midgard: Update integer op list
Alyssa Rosenzweig [Mon, 22 Apr 2019 02:56:53 +0000 (02:56 +0000)]
panfrost/midgard: Update integer op list

In the future, we might want to switch to a table-based approach, but
for now, at least have it current.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
5 years agopanfrost/midgard: Remove unused mir_next_block
Alyssa Rosenzweig [Sun, 21 Apr 2019 19:13:27 +0000 (19:13 +0000)]
panfrost/midgard: Remove unused mir_next_block

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
5 years agopanfrost/midgard: Fix off-by-one in successor analysis
Alyssa Rosenzweig [Sun, 21 Apr 2019 19:12:10 +0000 (19:12 +0000)]
panfrost/midgard: Fix off-by-one in successor analysis

This reduces register pressure substantially since we get smaller
liveness ranges.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
5 years agopanfrost/midgard: Track loop depth
Alyssa Rosenzweig [Sun, 21 Apr 2019 16:22:44 +0000 (16:22 +0000)]
panfrost/midgard: Track loop depth

This fixes nested loops.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
5 years agopanfrost/midgard: Dead code eliminate MIR
Alyssa Rosenzweig [Sun, 21 Apr 2019 16:11:11 +0000 (16:11 +0000)]
panfrost/midgard: Dead code eliminate MIR

We reshuffle the existing "dead move elimination" pass into a generic
dead code elimination layer, fixing bugs incurred with looping in the
process.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
5 years agopanfrost: Use actual imov instruction
Alyssa Rosenzweig [Sun, 21 Apr 2019 05:11:37 +0000 (05:11 +0000)]
panfrost: Use actual imov instruction

The bug this worked around is no longer applicable, it seems -- remove
the hack that breaks more than it fixes.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
5 years agopanfrost: Disable indirect outputs for now
Alyssa Rosenzweig [Sun, 21 Apr 2019 05:11:02 +0000 (05:11 +0000)]
panfrost: Disable indirect outputs for now

The hardware needs this lowered anyway; for now, might as well use
mesa's default lowering for pure conformance reasons.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
5 years agopanfrost/midgard: imul can only run on *mul
Alyssa Rosenzweig [Sun, 21 Apr 2019 04:09:10 +0000 (04:09 +0000)]
panfrost/midgard: imul can only run on *mul

This restriction makes sense logically. Not sure why it wasn't obeyed
before. In conjunction with previous commit's disclaimer, fixes
dEQP-GLES2.functional.shaders.loop.for_dynamic_iterations.*

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
5 years agopanfrost/midgard: Don't try to inline constants on branches
Alyssa Rosenzweig [Sun, 21 Apr 2019 03:59:05 +0000 (03:59 +0000)]
panfrost/midgard: Don't try to inline constants on branches

Along with a corresponding fix to the move elimination pass (not
included here yet -- I just have it disabled for now), this will fix
dEQP-GLES2.functional.shaders.loops.for_uniform_iterations.*

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
5 years agopanfrost: Respect backwards branches in RA
Alyssa Rosenzweig [Sun, 21 Apr 2019 03:29:47 +0000 (03:29 +0000)]
panfrost: Respect backwards branches in RA

Fixes a bunch of issues with looping. Honestly, I'm not sure why loops
worked at all before.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
5 years agopanfrost/midgard: Remove useless MIR dump
Alyssa Rosenzweig [Sun, 21 Apr 2019 01:43:08 +0000 (01:43 +0000)]
panfrost/midgard: Remove useless MIR dump

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
5 years agopanfrost/midgard: Respect component of bcsel condition
Alyssa Rosenzweig [Sun, 21 Apr 2019 00:09:13 +0000 (00:09 +0000)]
panfrost/midgard: Respect component of bcsel condition

Fixes a bunch of non-vec4 indexing.varying_array tests.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
5 years agopanfrost/midgard: Implement indirect loads of varyings/UBOs
Alyssa Rosenzweig [Sat, 20 Apr 2019 23:52:42 +0000 (23:52 +0000)]
panfrost/midgard: Implement indirect loads of varyings/UBOs

This adds preliminary support for indirect loads of varying arrays and
uniform arrays, bringing a few new tests in shader.indexing.* to
passing, although there remains a number of cases still missing.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
5 years agopanfrost/midgard: Pipe through varying arrays
Alyssa Rosenzweig [Sat, 20 Apr 2019 23:39:29 +0000 (23:39 +0000)]
panfrost/midgard: Pipe through varying arrays

Varying arrays sometimes are lowered to a series of directly accessed
varyings (which we handled okay), but when indirectly accessed, they
appear as a single array; we need to handle this as well.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
5 years agopanfrost/mdg/disasm: Print raw varying_parameters
Alyssa Rosenzweig [Sat, 20 Apr 2019 23:37:14 +0000 (23:37 +0000)]
panfrost/mdg/disasm: Print raw varying_parameters

The semantics of this field are not well understood; it is better to
print it unconditionally along with the other unknown state, rather than
silently eat the value. Without this change, some critical state was
being lost in some shaders (notably, the offset for load/store
scratchpad intructions found in shaders that spill registers.)

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
5 years agoiris: Prefer staging blits when destination supports CCS_E.
Kenneth Graunke [Wed, 24 Apr 2019 00:40:09 +0000 (17:40 -0700)]
iris: Prefer staging blits when destination supports CCS_E.

Otherwise our textures don't get color compression.  Thanks to
Eero Tamminen for noticing this was missing!

Improves performance of GLB27_FillTestC24Z16 on my Apollolake
laptop with single channel RAM by 2.3x.

Reported-by: Eero Tamminen <eero.t.tamminen@intel.com>
5 years agogallium: replace drm_driver_descriptor::configuration with driconf_xml
Marek Olšák [Tue, 23 Apr 2019 00:00:10 +0000 (20:00 -0400)]
gallium: replace drm_driver_descriptor::configuration with driconf_xml

PIPE_CAPs are better.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
5 years agogallium: replace DRM_CONF_SHARE_FD with PIPE_CAP_DMABUF
Marek Olšák [Mon, 22 Apr 2019 21:50:00 +0000 (17:50 -0400)]
gallium: replace DRM_CONF_SHARE_FD with PIPE_CAP_DMABUF

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
5 years agogallium: replace DRM_CONF_THROTTLE with PIPE_CAP_MAX_FRAMES_IN_FLIGHT
Marek Olšák [Mon, 22 Apr 2019 21:35:27 +0000 (17:35 -0400)]
gallium: replace DRM_CONF_THROTTLE with PIPE_CAP_MAX_FRAMES_IN_FLIGHT

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
5 years agost/dri: simplify throttling code
Marek Olšák [Mon, 22 Apr 2019 21:05:18 +0000 (17:05 -0400)]
st/dri: simplify throttling code

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
5 years agogallium: document conservative rasterization flags
Marek Olšák [Fri, 19 Apr 2019 23:11:34 +0000 (19:11 -0400)]
gallium: document conservative rasterization flags

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
5 years agointel/compiler: Lower ffma on Gen4 and Gen5
Ian Romanick [Fri, 19 Apr 2019 00:48:15 +0000 (17:48 -0700)]
intel/compiler: Lower ffma on Gen4 and Gen5

flrp32 is also a 3-source instruction, but there is another pending
series that handles that for Gen4 and Gen5.

v2: Rebase on "intel/compiler: Don't have sepearate, per-Gen
nir_options"

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Matt Turner <mattst88@gmail.com>
5 years agointel/compiler: Don't have sepearate, per-Gen nir_options
Ian Romanick [Fri, 19 Apr 2019 20:11:34 +0000 (13:11 -0700)]
intel/compiler: Don't have sepearate, per-Gen nir_options

Instead, just have separate scalar vs. vector nir_options and do
per-Gen "fix ups".

Suggested-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Matt Turner <mattst88@gmail.com>
5 years agoglsl: Silence may unused parameter warnings in glsl/ir.h
Ian Romanick [Thu, 18 Apr 2019 19:33:39 +0000 (12:33 -0700)]
glsl: Silence may unused parameter warnings in glsl/ir.h

Every file that included glsl/ir.h had a warning like:

src/compiler/glsl/ir.h: In member function ‘virtual bool ir_rvalue::is_lvalue(const _mesa_glsl_parse_state*) const’:
src/compiler/glsl/ir.h:236:64: warning: unused parameter ‘state’ [-Wunused-parameter]
    virtual bool is_lvalue(const struct _mesa_glsl_parse_state *state = NULL) const
                                                                ^
Cc: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Fixes: fa4ebf6b8d9 ("glsl: add _mesa_glsl_parse_state object to is_lvalue()")
Reviewed-by: Sagar Ghuge <sagar.ghuge@intel.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
5 years agost/mesa/radeonsi: fix race between destruction of types and shader compilation
Timothy Arceri [Tue, 23 Apr 2019 02:54:38 +0000 (12:54 +1000)]
st/mesa/radeonsi: fix race between destruction of types and shader compilation

Commit 624789e3708c moved the destruction of types out of atexit() and
made use of a ref count instead. This is useful for avoiding a crash
where drivers such as radeonsi are still compiling in a thread when the app
exits and has not called MakeCurrent to change from the current context.

While the above scenario is technically an app bug we shouldn't crash.
However that change caused another race condition between the shader
compilation tread in radeonsi and context teardown functions.

This patch makes two changes to fix this new problem:

First we explicitly call _mesa_destroy_shader_compiler_types() when destroying
the st context rather than calling it indirectly via _mesa_free_context_data().
We do this as we must call it after st_destroy_context_priv() so that we don't
destory the glsl types before the compilation threads finish.

Next wait for the shader threads to finish in si_destroy_context() this
also means we need to call context destroy before destroying the queues
in si_destroy_screen().

Fixes: 624789e3708c ("compiler/glsl: handle case where we have multiple users for types")

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
5 years agoradv: Add adaptive_sync driconfig option and enable it by default.
Bas Nieuwenhuizen [Tue, 16 Apr 2019 23:51:10 +0000 (01:51 +0200)]
radv: Add adaptive_sync driconfig option and enable it by default.

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
5 years agovulkan/wsi: Add X11 adaptive sync support based on dri options.
Bas Nieuwenhuizen [Tue, 16 Apr 2019 23:30:49 +0000 (01:30 +0200)]
vulkan/wsi: Add X11 adaptive sync support based on dri options.

The dri options are optional. When the dri options are not provided
the WSI will not  use adaptive sync.

FWIW I think for xf86-video-amdgpu this still requires an X11 config
option, so only people who opt in can get possible regressions from this.

So then the remaining question is: why do this in the WSI?

It has been suggested in another MR that the application sets this.
However, I disagree with that as I don't think we'll ever get a
reasonable set of applications setting it.

The next questions is whether this can be a layer. It definitely
can be as implemented now. However, I think this generally fits
well with the function of the WSI. Furthemore, for e.g. the DISPLAY
WSI this is much harder to do in a layer.

Of course, most of the WSI could almost be a layer, but I think
this still fits best in the WSI.

Acked-by: Jason Ekstrand <jason@jlekstrand.net>
5 years agoradv: Add support for driconf.
Bas Nieuwenhuizen [Sun, 14 Apr 2019 22:32:27 +0000 (00:32 +0200)]
radv: Add support for driconf.

This includes 0 options.

The cache parsing is located at a position where we can easily add
config filtering by VkApplicationInfo.

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
5 years agoiris: add support for INTEL_conservative_rasterization
Mike Blumenkrantz [Thu, 18 Apr 2019 17:21:56 +0000 (13:21 -0400)]
iris: add support for INTEL_conservative_rasterization

this hooks up the iris gallium driver to existing mesa bits which handle
the implementation

resolves kwg/mesa#8

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
5 years agost/mesa: indicate intel extension support for inner_coverage based on cap
Mike Blumenkrantz [Thu, 18 Apr 2019 17:18:43 +0000 (13:18 -0400)]
st/mesa: indicate intel extension support for inner_coverage based on cap

if the driver (iris) indicates support for the inner_coverage pipe cap, this
will set the necessary states in the driver flags and rasterizer structs

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
5 years agogallium: add pipe cap for inner_coverage conservative raster mode
Mike Blumenkrantz [Thu, 18 Apr 2019 17:18:03 +0000 (13:18 -0400)]
gallium: add pipe cap for inner_coverage conservative raster mode

this can be used by drivers which support the extension to indicate support

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
5 years agoiris: Fix DrawTransformFeedback math when there's a buffer offset
Kenneth Graunke [Fri, 22 Mar 2019 07:42:56 +0000 (00:42 -0700)]
iris: Fix DrawTransformFeedback math when there's a buffer offset

We need to subtract the starting offset from the final offset before
dividing by the stride.  See src/intel/vulkan/genX_cmd_buffer.c:3142.

Not known to fix anything.

5 years agoiris: Make some offset math helpers take a const isl_surf pointer
Kenneth Graunke [Thu, 14 Mar 2019 08:19:59 +0000 (01:19 -0700)]
iris: Make some offset math helpers take a const isl_surf pointer

5 years agospirv: Handle SpvOpDecorateId
Caio Marcelo de Oliveira Filho [Mon, 22 Apr 2019 23:09:56 +0000 (16:09 -0700)]
spirv: Handle SpvOpDecorateId

This operation decorate with an Id instead of a Literal or String.

It is used by HlslCounterBufferGOOGLE (provided by
SPV_GOOGLE_hlsl_functionality1).  Even if we don't do anything with
that decoration, we must be able to parse SPIR-V that uses it.

Fixes: 891886da2f9 "spirv: Add no-op support for VK_GOOGLE_hlsl_functionality1"
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
5 years agospirv: Rename vtn_decoration literals to operands
Caio Marcelo de Oliveira Filho [Mon, 22 Apr 2019 23:17:58 +0000 (16:17 -0700)]
spirv: Rename vtn_decoration literals to operands

Decorations (and ExecutionModes) can have not only literals, but also
Ids associated with them.  So rename the field to the more general
name "Operand" used by the spec.

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
5 years agoanv: fix argument name for vkCmdEndQuery
Lionel Landwerlin [Mon, 22 Apr 2019 21:09:11 +0000 (22:09 +0100)]
anv: fix argument name for vkCmdEndQuery

Doesn't fix anything but it's not the right function prototype.

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Fixes: 673f33c77dd765 ("anv: Implement CmdBegin/EndQueryIndexed")
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Sagar Ghuge <sagar.ghuge@intel.com>
5 years agovirgl: skip empty cmdbufs
Chia-I Wu [Thu, 18 Apr 2019 22:34:46 +0000 (15:34 -0700)]
virgl: skip empty cmdbufs

Several empty cmdbufs are submitted by app/xserver per frame, from
glamor_block_handler for example.  Let's skip them.

Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
Reviewed-by: Gurchetan Singh <gurchetansingh@chromium.org>
5 years agogallium: Remove the malloc pipebuffer manager.
Eric Anholt [Fri, 19 Apr 2019 22:19:58 +0000 (15:19 -0700)]
gallium: Remove the malloc pipebuffer manager.

This has been unused since r600 stopped using it in 2010.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Reviewed-by: Kristian Høgsberg <hoegsberg@gmail.com>
5 years agogallium: Remove the "alt" pipebuffer manager interface.
Eric Anholt [Fri, 19 Apr 2019 22:10:30 +0000 (15:10 -0700)]
gallium: Remove the "alt" pipebuffer manager interface.

This one would allocate from two underlying pools, but has never been
used.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Reviewed-by: Kristian Høgsberg <hoegsberg@gmail.com>