anv: Advertise VK_EXT_shader_demote_to_helper_invocation

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>

spirv: Implement SPV_EXT_demote_to_helper_invocation

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>

spirv: Update the headers from latest Khronos master

This corresponds to 29c11140baaf9f7fdaa39a583672c556bf1795a1 in
https://github.com/KhronosGroup/SPIRV-Headers.

Acked-by: Jason Ekstrand <jason@jlekstrand.net>

intel/fs: Implement "demote to helper invocation"

The "demote" intrinsic works like "discard" but don't change the
control flow, allowing derivative operations to work.  This is the
semantics of D3D discard.

The "is_helper_invocation" intrinsic will return true for helper
invocations -- both the ones that started as helpers and the ones that
where demoted.  This is needed to avoid changing the behavior of
gl_HelperInvocation which is an input (so not expected to change
during shader execution).

v2: Emit the discard jump and comment why it is safe.  (Jason)
    Rework the is_helper_invocation() that was stomping f0.1.  (Jason)

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>

nir: Add demote and is_helper_invocation intrinsics

From SPV_EXT_demote_to_helper_invocation.  Demote will be implemented
as a variant of discard, so mark uses_discard if it is used.

v2: Add CAN_ELIMINATE flag to the new intrinsic.  (Jason)

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>

radv: do not emit VGT_FLUSH on GFX10

We don't need it.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

ac/nir: Remove now-unused interp_deref handling

Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>

radeonsi/nir: Use NIR barycentric intrinsics

This is simpler than radv, since the driver_location is already assigned
for us.

Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>

radeonsi/nir: Delete unreachable code

We always get gl_FragCoord as a system value, not a varying, so this is
never hit. We already set PIXEL_CENTER_INTEGER elsewhere.

Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>

compiler: Add color system value

This is nice to have with radeonsi, where color varyings are handled
specially to avoid recompiles.

Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>

radv: Use NIR barycentric intrinsics

We have to add a few lowering to deal with things that used to be dealt
with inline when creating inputs. We also move the code that fills out
the radv_shader_variant_info struct for linking purposes to
radv_shader.c, as it's no longer tied to the NIR->LLVM lowering.

Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

ac/nir: Implement barycentric intrinsics

Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>

intel/nir: Extract add_const_offset_to_base

Pretty much every driver using nir_lower_io_to_temporaries followed by
nir_lower_io is going to want this. In particular, radv and radeonsi in
the next commits.

Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

nir/lower_io_to_temporaries: Handle interpolation intrinsics

These weren't properly supported. This does pretty much the same thing
that the radv code did.

Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

nir: Avoid coalescing vars created by lower_io_to_temporaries

Right now nir_copy_prop_vars is effectively undoing
nir_lower_io_to_temporaries for inputs by propagating the original
variable through the copy created in lower_io_to_temporaries. A
theoretical variable coalescing pass would have the same issue with
output variables, although that doesn't exist yet. To fix this, add a
new bit to nir_variable, and disable copy propagation when it's set.

This doesn't seem to affect any drivers now, probably since since no one
uses lower_io_to_temporaries for inputs as well as copy_prop_vars, but
it will fix radv once we flip on lower_io_to_temporaries for fs inputs.

Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

nir: Return correct size in nir_assign_io_var_locations()

It was double-counting cases where multiple variables were assigned to
the same slot, and not handling the case where the last variable is a
compact variable.

Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

nir: Handle compact variables when assigning i/o locations

These are used in Vulkan for clip/cull distances, instead of the GLSL
lowering when the clip/cull arrays are shared.

Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

nir: Move st_nir_assign_var_locations() to common code

It isn't really doing anything Gallium-specific, and it's needed for
handling component packing, overlapping, etc.

Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>

radv: Make FragCoord a sysval

load_fragcoord is already handled in common code for radeonsi, so we
don't need to do anything to handle it. However, there were some passes
creating NIR with the varying, so we switch them over to the sysval. In
the case of nir_lower_input_attachments which is used by both radv and
anv, we add handling for both until intel switches to using a sysval.

Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

spirv: Add an option for making FragCoord a sysval

On AMD, FragCoord should be a sysval because it is handled separately
from all the other inputs. We were already doing this in radeonsi, but
we weren't doing it with radv. It'll be much more annoying to handle
VARYING_SLOT_POS in fragment shaders when we let NIR lower FS inputs for
us, so here we add an option so that radv can get it as a system value.

Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv: Lower input attachments in NIR.

v2 (Connor)
- Fix warning in release mode using MAYBE_UNUSED

Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv: Implement nir_intrinsic_load_layer_id().

Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>

anv,nir: Move lower_input_attachments pass from ANV to NIR.

Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: don't emit PFP packets on ME.

This was done for all previous GPUs.

This fixes Talos Principle launch hangs.

Fixes: 7e43022e8c8 (radv/gfx10: add gfx10_cs_emit_cache_flush)
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

ac: select the GFX ring when halting waves with UMR on GFX10

GFX10 has two rings, so UMR want to know which one to halt.
Select the first one by default.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: Move NGG output handling outside of giant if-statement.

In merged shaders we put a big if around each shader, so both stages
can have a different number of threads. However, the NGG output code
still needs to run if the first shader is not executed.

This can happen when there are more gs threads than vs/es threads, or
when there are 0 es/vs threads (why? no clue).

Fixes: ee21bd7440c "radv/gfx10: implement NGG support (VS only)"
Reviewed-by: Dave Airlie <airlied@redhat.com>

radv: Actually use VK formats for the format table.

No ETC2 or ASTC on navi so nothing to add.

Fixes: 3dc5ec5d167 "radv/gfx10: generate gfx10_format_table.h"
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Dave Airlie <airlied@redhat.com>

anv: fix VkExternalBufferProperties for host allocation

It was reported as unsupported previously.  It should be importable
and is compatible with itself.

Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
Fixes: 69cc6272fbc199 ("anv: Implement VK_EXT_external_memory_host")
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>

anv: fix VkExternalBufferProperties for unsupported handles

compatibleHandleTypes must include the queried handle type.

Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
Cc: <mesa-stable@lists.freedesktop.org>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>

radv: Handle cmask being disallowed by addrlib.

alignment=0 does weird things with align64.

CC: <mesa-stable@lists.freedesktop.org>
Reviewed-by: Dave Airlie <airlied@redhat.com>

radv/gfx10: enable support for NAVI10, NAVI12 and NAVI14

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: Use GS rectlist when needed.

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>

radv/gfx10: implement NGG support (VS only)

This needs to be cleaned up a bit, and it probably contains
missing stuff and/or bugs.

This doesn't fix the "half of the triangles" issue.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv: Combine vs and tes output keys parts.

That way the same deref is valid for both shader stages.

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>

radv/gfx10: Use new uconfig reg index packet for GFX10+.

Otherwise the hardware/firmware seems to not set the registers.

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>

radv/gfx10: Set MEM_ORDERED flags on shaders.

Scattered because depending on stage they are at offset 24/25/27/30.

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>

radv/gfx10: emit GE_CNTL instead of IA_MULTI_VGT_PARAM for legacy mode

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: double the number of tessellation offchip buffers per SE

Each gfx10 shader engine corresponds to two gfx9 shader engines, so scale
the number of offchip buffers accordingly.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: require LLVM 9+

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: disable geometry and tessellation shaders

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: disable binning

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: disable CLEAR_STATE

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: disable VK_EXT_transform_feedback

It requires a bunch of work, so disable for now.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: set user data base registers

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: add gfx10_cs_emit_cache_flush

The cache flush logic on GFX10 is quite different and it's
implemented with a new function.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: set the DCC constant encoding flag

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: do not declare streamout SGPRS

Streamout is completely different on GFX10.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: do not set stream output shader config

Transform feedback is really different on GFX10.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: emit VGT_VERTEX_REUSE_BLOCK_CNTL during gfx initialization

The value doesn't need to be updated for tess.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: update shader-related fields in si_emit_graphics()

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: implement si_emit_compute()

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: mask DCC tile swizzle by alignment

DCC alignment can be less than the alignment of the main surface. In that
case, the DCC tile swizzle needs to be masked accordingly. Should have no
impact on pre-gfx10.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: initialize GE_{MAX,MIN}_VTX_INDX/INDX_OFFSET

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: implement radv_flush_vertex_descriptors() change

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: implement fill_geom_tess_rings()

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: implement radv_CmdBindDescriptorSets()

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: implement write_buffer_descriptor()

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: use the correct register for image descriptor dumping

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: implement radv_pipeline_generate_hw_hs()

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: implement radv_fill_shader_variant()

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: implement radv_pipeline_generate_geometry_shader()

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: implement radv_init_sampler()

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: fix PS exports for SPI_SHADER_32_AR

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: implement radv_get_device_name()

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: set RADV_FORCE_FAMILY

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: fix a possible hang with exp pos0 with done=0 and exec=0

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: set PA_SC_TILE_STEERING_OVERRIDE

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: set cache control registers

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: set llvm_has_working_vgpr_indexing

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: update DB_DFSM_CONTROL register

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: update DB_Z_INFO register

GFX10 uses the same register as GFX8.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: implement radv_emit_global_shader_pointers()

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: implement radv_emit_tess_factor_ring()

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: implement radv_emit_fb_ds_state()

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: implement radv_initialise_ds_surface()

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: implement radv_emit_fb_color_state()

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: implement radv_initialise_color_surface()

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: implement radv_init_dcc_control_reg()

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: implement radv_make_buffer_descriptor()

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: implement si_set_mutable_tex_desc_fields()

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: add gfx10_make_texture_descriptor

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: generate gfx10_format_table.h

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: increase maximum number of layers to 8192

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: increase maximum number of levels to 14

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv/gfx10: set MAX_ALLOC_COUNT

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

ac/nir: unpacked GS invocation ID on GFX10+

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

ac: add missing formats to ac_get_tbuffer_format() for GFX10

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

vulkan/overlay: fix command buffer stats

Begin/Reset of command buffer both reset the content of the command
buffer. Don't forget to wipe them on Begin.

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Fixes: 4438188f492e1f ("vulkan/overlay: record stats in command buffers and accumulate on exec/submit")
Acked-by: Eric Engestrom <eric.engestrom@intel.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

anv: manually add KHR_display to the list of platforms

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Fixes: 38305e6c94ea31 ("anv: replace hard-coded platform list with vk.xml parse")
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=111078
Reviewed-by: Eric Engestrom <eric.engestrom@intel.com>

docs/features: add shader buffer and atomic support for llvmpipe

llvmpipe: enable ARB_shader_storage_buffer_object

Reviewed-by: Roland Scheidegger <sroland@vmware.com>

llvmpipe: add support for shader buffer binding.

This add support for setting shader buffers and passing them
to draw or binding them to the fragment shader jit.

Reviewed-by: Roland Scheidegger <sroland@vmware.com>

draw: add shader buffer interfaces.

This adds the interface to add mapped shader buffers,
and sets up the jit linkage for them.

Reviewed-by: Roland Scheidegger <sroland@vmware.com>

gallivm: add buffer operations to the tgsi->llvm conversion.

This adds load, store and atomic operations. These operations
have to respect the exec_mask, and can't operate in lanes where
the execute is off. This is needed to avoid side effects seen
outside the shaders.

There is also bounds checking on the ssbo accesses vs the size
ptr.

Reviewed-by: Roland Scheidegger <sroland@vmware.com>

gallivm: move mask_vec function up higher so it can be reused.

Reviewed-by: Roland Scheidegger <sroland@vmware.com>

tgsi: denote which load/store/atomic channels are unsigned

llvmpipe will need this info.

Reviewed-by: Roland Scheidegger <sroland@vmware.com>

llvmpipe: add support for ssbo to the fragment shader jit.

This just adds the ssbo ptrs to the jit fragment shader api.

Reviewed-by: Roland Scheidegger <sroland@vmware.com>

draw: add support for ssbo ptrs to jit tables.

This adds ssbo/num_ssbo ptrs to the vs/gs jit tables.

Reviewed-by: Roland Scheidegger <sroland@vmware.com>

gallivm: add some basic SSBO limits. (v2)

v2: update ssbo size

Reviewed-by: Roland Scheidegger <sroland@vmware.com>

util: add util_copy_shader_buffer.

This just adds an inline to copy a pipe_shader_buffer.

Reviewed-by: Roland Scheidegger <sroland@vmware.com>