radv: add tessellation ring allocation support. (v2)

This patch adds support for the offchip rings for storing
tessellation factors and attribute data.

It includes the register setup for the TF ring

v2: always do tess ring size calcs (Bas)

Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Signed-off-by: Dave Airlie <airlied@redhat.com>

radv: add support for some device specific tess information.

Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Signed-off-by: Dave Airlie <airlied@redhat.com>

radv/ac: add tess changes to shader keys/info

This adds the tess pieces for shader keys and shader info,
it adds the necessary bits to the vertex key/info as well.

Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Signed-off-by: Dave Airlie <airlied@redhat.com>

radv: add tess shader stage user data support.

This just adds support for tess to the shader stage conversion
and emits the per-stage descriptors/constants for tess stages.

Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Signed-off-by: Dave Airlie <airlied@redhat.com>

radv: use defines for ring descriptor offsets.

Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Signed-off-by: Dave Airlie <airlied@redhat.com>

radv: add helper function to denote if tess is enabled on a pipeline.

Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Signed-off-by: Dave Airlie <airlied@redhat.com>

radv: handle clip dist in es outputs.

Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Signed-off-by: Dave Airlie <airlied@redhat.com>

radv: drop unneeded start

Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Signed-off-by: Dave Airlie <airlied@redhat.com>

radv: fixup geometry clip emission since using the geom pass

Fixes: 2b35b60d: radv: move to using nir clip/cull merge pass.
Signed-off-by: Dave Airlie <airlied@redhat.com>

radeonsi/gfx9: allow CMASK fast clear with RB+

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

radeonsi/gfx9: don't compare src_va w/ dst_va for CP_DMA_CLEAR

src_va contains the clear value in this case.

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

radeonsi/gfx9: fix 1D array fetches with derivs, bias, or Z compare value

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

radeonsi/gfx9: fix and enable single-sample CMASK fast clear

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

radeonsi/gfx9: fix and enable MSAA compression

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

radeonsi/gfx9: disable CE

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

radeonsi/gfx9: fix linear mipmap CPU access

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

radeonsi: add tests verifying that VM faults don't hang

GFX9 hangs instead of writing VM faults to dmesg.

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

radeonsi: unify HS max_offchip_buffers workarounds

Vulkan doesn't set more than 508.

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

radeonsi: adjust checking for SC bug workarounds

no change in behavior, just making sure that no later chips will use
the workarounds

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

glsl: use -O1 optimization for builtin_functions.cpp with MinGW

Some versions of MinGW-w64 such as 5.3.1 and 6.2.0 produce bad code
with -O2 or -O3 causing a random driver crash when running programs
that use GLSL.  Most Mesa demos in the glsl/ directory trigger the
bug, but not the fragcoord.c test.

Use a #pragma to force -O1 for this file for later MinGW versions.
Luckily, this is basically one-time setup code.  I suspect the bug
is related to the sheer size of this file.

This should let us move to newer versions of MinGW-w64 for Mesa.

Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>

tnl: remove unused var to silence warning

Trivial.

st/wgl: Replace variable name hdc with hDrawDC

Reviewed-by: Brian Paul <brianp@vmware.com>

st/wgl: add support for WGL_ARB_make_current_read

This adds the wglMakeContextCurrentARB() and wglGetCurrentReadDCARB()
functions.

Signed-off-by: Brian Paul <brianp@vmware.com>

stw/wgl: add null context check in wglBindTexImageARB()

To avoid dereferencing a null pointer in case wglMakeCurrent() wasn't
called.  Found while debugging SWKOTOR game.

Reviewed-by: Neha Bhende <bhenden@vmware.com>
Reviewed-by: Charmaine Lee <charmainel@vmware.com>

radeonsi: decompress DCC in set_sampler_view instead of create_sampler_view (v2)

v2: don't add a new decompress helper function

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

radeonsi: decompress DCC in set_framebuffer_state instead of create_surface (v2)

for threaded gallium, which can't use pipe_context in create_surface

v2: don't add a new decompress helper function

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

st/glsl_to_tgsi: fix 64-bit integer bit shifts

Fix a bug that was caused by a type mismatch in the shift count between
GLSL and TGSI. I briefly considered adjusting the TGSI semantics, but
since both LLVM and AMD GCN require both arguments to be of the same type,
it makes more sense to keep TGSI as-is -- it reflects the underlying
implementation better.

I'm also sending out piglit tests that expose this error.

v2: use the right number of components for the temporary register

Reviewed-by: Marek Olšák <marek.olsak@amd.com>

tgsi: fix printing of 64-bit integer immediates

Reviewed-by: Marek Olšák <marek.olsak@amd.com>

intel: genxml: fix out of tree builds

v2: use Emil's recommendation
    change rule to closer to genxml/genX_bits.h

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>

gbm/dri: Check dri extension version before flush after unmap

The commit mentioned below required the __DRI2FlushExtension to have
version 4 or above, for GBM functionality. That broke GBM with some
classic dri drivers. Relax that requirement so that we only flush
after unmap if we have version 4 or above. Drivers that require the flush
for correct functionality should implement the desired version.

Fixes: ba8df228 ("gbm/dri: Flush after unmap")
Cc: <mesa-stable@lists.freedesktop.org>
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Tested-by: Dylan Baker <dylan@pnwbakers.com>

radeonsi: implement ARB_shader_group_vote

Reviewed-by: Marek Olšák <marek.olsak@amd.com>

radeonsi: enable ARB_shader_clock

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>

radeonsi: emit TGSI_OPCODE_CLOCK

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>

st/mesa: implement ARB_shader_clock

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>

tgsi: add CLOCK opcode

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>

gallium: add PIPE_CAP_TGSI CLOCK

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>

glsl: fix clockARB builtin function

The underlying intrinsic is defined to always have a uvec2 return type.

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>

anv: change BLOCK_POOL_MEMFD_SIZE to 1GB

This allows us to run 32bit Vulkan apps on Android, ftruncate
call would fail on 2GB (max size being 2GB - 1).

Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>

android: add libmesa_genxml as dep to libmesa_isl

This is to fix following compile error with libmesa_isl:
   mesa/src/intel/isl/isl.c:28:10: fatal error: 'genxml/genX_bits.h' file not found

Fixes: f0eaf38 ("genxml: New generated header genX_bits.h (v6)")
Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: Emli Velikov <emil.velikov@collabora.com>

mesa: remove MESA_GLSL=opt

This is unused.

Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Reviewed-by: Emil Velikov <emli.velikov@collabora.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>

mesa: remove MESA_GLSL=no_opts env option

This is confusing because is only applys to GL_ARB_vertex/fragment_program,
and because of that its also not very useful.

If someone requires this for debugging they can just make an ad-hoc
code change.

Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>

mesa: move FLUSH_VERTICES() call to meta

There is no need for this to be in the common code.

Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>

mesa/vbo: remove redundant _mesa_is_bufferobj() calls

This is already called inside the vbo_exec_vtx_{unmap,map}()
functions.

Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>

mesa/glthread: add async support to ARB_gpu_shader_int64 uniform functions

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

mesa/glthread: add async support to ARB_gpu_shader_fp64 uniform functions

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

aubinator: enable snb/ilk through --gen

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>

intel: genxml: compress all gen files into one

Combining all the files into a single string didn't make any
difference in the size of the aubinator binary.

With this change we now also embed gen4/4.5/5 descriptions, which
increases the aubinator size by ~16Kb.

v2 (Lionel): rebase makefiles

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>

radv: Use the guard band.

Signed-off-by: Bas Nieuwenhuizen <basni@google.com>
Reviewed-by: Dave Airlie <airlied@redhat.com>

radv: Prepare for not using the guard band for lines & points.

Vulkan Clipping is defined in terms of vertices, the scissor based
clipping happens on pixels. There is a difference with points and
lines, as a vertex can be outside the viewport while some pixels are in.
On Vulkan thoise pixels shouldn't be drawn, while they would be with
the guardband.

Signed-off-by: Bas Nieuwenhuizen <basni@google.com>
Reviewed-by: Dave Airlie <airlied@redhat.com>

radv: Drop the default viewport when 0 viewports are given.

Signed-off-by: Bas Nieuwenhuizen <basni@google.com>
Reviewed-by: Dave Airlie <airlied@redhat.com>

radv: Set proper viewport & scissor for meta draws.

Signed-off-by: Bas Nieuwenhuizen <basni@google.com>
Reviewed-by: Dave Airlie <airlied@redhat.com>

mesa: Fix trailing whitespace in polygon.c

Signed-off-by: Lyude <lyude@redhat.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>

mesa: Fix gross indenting in _mesa_PolygonMode()

Signed-off-by: Lyude <lyude@redhat.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>

r300: Fix indenting in r300_get_param()

Signed-off-by: Lyude <lyude@redhat.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>

vc4: Fix indenting in vc4_screen_get_param()

Signed-off-by: Lyude <lyude@redhat.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>

intel: Add INTEL_CFLAGS to aubinator CFLAGS.

It still needs intel_aub.h.  Fixes the build.

nir: Add support for 8 and 16-bit types

Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Reviewed-by: Eduardo Lima Mitev <elima@igalia.com>

nir/constant_expressions: Don't switch on bit size when not needed

For opcodes such as the nir_op_pack_64_2x32 for which all sources and
destinations have explicit sizes, the bit_size parameter to the evaluate
function is pointless and *should* do nothing.  Previously, we were
always switching on the bit_size and asserting if it isn't one of the
sizes in the list.  This generates way more code than needed and is a
bit cruel because it doesn't let us have a bit_size of zero on an ALU op
which shouldn't need a bit_size.

Reviewed-by: Eduardo Lima Mitev <elima@igalia.com>

nir/constant_expressions: Pull the guts out into a helper block

Reviewed-by: Eduardo Lima Mitev <elima@igalia.com>

i965: Stop using legacy dri_bufmgr_* and intel_* names.

Eric renamed these from dri_bufmgr_* and intel_bufmgr_* to drm_intel_*
in libdrm commit 4b9826408f65976a1a13387beda748b65e03ec52, circa 2008,
but we've been using the legacy names this whole time.

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

intel: automake: move INTEL_CFLAGS as applicable

Only common/decoder.[ch] requires it [for intel_aub.h].

v2: The code was moved to from intel/tools to intel/common,
update accordingly.

Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

intel: android: remove libdrm_intel requirement

The only part which requires libdrm_intel tools/aubinator is not built
on Android.

Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

Partially revert "amd/addrlib: silence warnings" to fix builds with DEBUG

This partially reverts commit 8a74140a21fe6b0d2e8a60b065b890f797f2db51.

ddebug: implement clear_texture

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

radeonsi: fix an unused-variable warning in a release build

vdpau: fix a maybe-uninitialized warning

softpipe: fix a maybe-uninitialized warning

/home/marek/dev/mesa-main/src/gallium/drivers/softpipe/sp_compute.c:178:
 warning: 'grid_size' may be used uninitialized in this function
 [-Wmaybe-uninitialized]

gallivm: fix a maybe-uninitialized warning

/home/marek/dev/mesa-main/src/gallium/auxiliary/gallivm/lp_bld_sample_soa.c:3598:
 warning: 'level' may be used uninitialized in this function [-Wmaybe-uninitialized]
       out1 = lp_build_cmp(&leveli_bld, PIPE_FUNC_GREATER, level, last_level);
            ^

gallium/radeon: s/dcc_disable/disable_dcc/

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Tested-by: Edmondo Tommasina <edmondo.tommasina@gmail.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>

radeonsi: handle incompatible DCC formats in resource_copy_region

Required because of later commits.

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

radeonsi: remove a workaround for inexact *8_SNORM blits

All tests pass on Fiji now. This prevents DCC disablement due to
incompatible DCC formats due to the fallback.

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

gallium/radeon: add and use a new helper vi_dcc_enabled

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Tested-by: Edmondo Tommasina <edmondo.tommasina@gmail.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>

gallium/radeon: formalize that r600_query_hw_add_result doesn't need a context

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Tested-by: Edmondo Tommasina <edmondo.tommasina@gmail.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>

radeonsi: don't make a copy of pipe_index_buffer in draw_vbo

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Tested-by: Edmondo Tommasina <edmondo.tommasina@gmail.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>

gallium/util: use const in u_index_modify helpers

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Tested-by: Edmondo Tommasina <edmondo.tommasina@gmail.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>

winsys/amdgpu: remove AMDGPU_INFO_NUM_EVICTIONS

This is now exposed with libdrm_amdgpu 2.4.76.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>

radeonsi: add Vega10 PCI IDs

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

radeon/uvd: set correct vega10 db pitch alignment

Create new function to get correct alignment based on Asics, and change
the corresponding decode message buffer and dpb buffer size calculations

Signed-off-by: Boyuan Zhang <boyuan.zhang@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>

radeon/vce: add vce support for firmware 53.19.4

v2: squashed with other similar commits

Signed-off-by: Leo Liu <leo.liu@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>

radeon/vce: adapt gfx9 surface to vce

Signed-off-by: Leo Liu <leo.liu@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>

winsys/surface: add height pitch for gfx9

Signed-off-by: Leo Liu <leo.liu@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>

radeon/uvd: clear message buffer when reuse

As required by firmware

Signed-off-by: Leo Liu <leo.liu@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>

radeon/uvd: adapt gfx9 surface to uvd

Signed-off-by: Leo Liu <leo.liu@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>

radeon/uvd: add uvd soc15 register

Signed-off-by: Leo Liu <leo.liu@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>

radeonsi/gfx9: disable features that don't work

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

radeonsi/gfx9: only allow GL 3.1

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

radeonsi/gfx9: add linear address computations for texture transfers

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

radeonsi/gfx9: don't generate LS and ES states

these shaders don't exist on GFX9

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

radeonsi/gfx9: SPI_SHADER_USER_DATA changes

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

winsys/amdgpu: set/get BO tiling flags for GFX9

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

radeonsi/gfx9: handle pitch and offset overrides for texture_from_handle

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

radeonsi/gfx9: set/validate GFX9 BO metadata

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

radeonsi/gfx9: add radeon_surf.gfx9.surf_offset

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

radeonsi/gfx9: don't write mipmap level offsets to BO metadata

GFX9 doesn't have (usable) mipmap offsets.

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

radeonsi/gfx9: flush CB & DB caches with an EOP TS event

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

radeonsi/gfx9: use ACQUIRE_MEM

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

radeonsi/gfx9: only use CE RAM for most-used descriptors

because the CE RAM size decreased to 4 KB.

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

radeonsi/gfx9: emit FLUSH_DFSM where required

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

radeonsi/gfx9: emit BREAK_BATCH in emit_framebuffer_state

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

radeonsi/gfx9: fix MIP0_WIDTH & MIP0_HEIGHT for compressed texture blits

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