swr/rast: Small editorial changes

Reviewed-by: Bruce Cherniak <bruce.cherniak@intel.com>

swr/rast: Use new processor detection mechanism

Use specific avx512 selection mechanism based on avx512er bit instead of
getHostCPUName().  LLVM 6.0.0 has a bug that reports wrong string for KNL
(fixed in 6.0.1).

Reviewed-by: Bruce Cherniak <bruce.cherniak@intel.com>

swr/rast: Output rasterizer dir to console since it's process specific

Reviewed-by: Bruce Cherniak <bruce.cherniak@intel.com>

swr/rast: Add TranslateGfxAddress for shader

Also add GFX_MEM_CLIENT_SHADER

Reviewed-by: Bruce Cherniak <bruce.cherniak@intel.com>

swr/rast: jit PRINT improvements.

Sign-extend integer types to 32bit when specifying "%d" and add new %u
which zero-extends to 32bit. Improves  printing of sub 32bit integer types
(i1 specifically).

Reviewed-by: Bruce Cherniak <bruce.cherniak@intel.com>

swr/rast: Fix regressions.

Bump jit cache revision number to force recompile.

Reviewed-by: Bruce Cherniak <bruce.cherniak@intel.com>

swr/rast: Cleanup old cruft.

Reviewed-by: Bruce Cherniak <bruce.cherniak@intel.com>

swr/rast: Package events.proto with core output

However only if the file exists in DEBUG_OUTPUT_DIR. The expectation is
that AR rasterizerLauncher will start placing it there when launching
a workload (which is in a subsequent checkin)

Reviewed-by: Bruce Cherniak <bruce.cherniak@intel.com>

swr/rast: Fix init in EventHandlerWorkerStats

Make sure we initialize variables.

Reviewed-by: Bruce Cherniak <bruce.cherniak@intel.com>

swr/rast: Fix return type of VCVTPS2PH.

expecting <8xi16> return.

Reviewed-by: Bruce Cherniak <bruce.cherniak@intel.com>

swr/rast: WIP Translation handling

Reviewed-by: Bruce Cherniak <bruce.cherniak@intel.com>

swr/rast: Use different handing for stream masks

Reviewed-by: Bruce Cherniak <bruce.cherniak@intel.com>

swr/rast: Silence warnings

Reviewed-by: Bruce Cherniak <bruce.cherniak@intel.com>

swr/rast: Add support for TexelMask evaluation

Reviewed-by: Bruce Cherniak <bruce.cherniak@intel.com>

swr/rast: Internal core change

Reviewed-by: Bruce Cherniak <bruce.cherniak@intel.com>

swr/rast: Fix x86 lowering 64-bit float handling

- 64-bit cvt-to-float needs to be explicitly handled
- gathers need the right parameter types to work with doubles

Fixes draw-vertices piglit tests

Reviewed-by: Bruce Cherniak <bruce.cherniak@intel.com>

swr/rast: Add some SIMD_T utility functors

VecEqual and VecHash

Reviewed-by: Bruce Cherniak <bruce.cherniak@intel.com>

swr/rast: Fix wrong type allocation

ALLOCA pointer elements, not pointers.

Reviewed-by: Bruce Cherniak <bruce.cherniak@intel.com>

swr: touch generated files to update timestamp

previous change in generators necessitates this change

Reviewed-by: Bruce Cherniak <bruce.cherniak@intel.com>

swr/rast: Fix byte offset for non-indexed draws

for the case when USE_SIMD16_SHADERS == FALSE

Reviewed-by: Bruce Cherniak <bruce.cherniak@intel.com>

util/u_queue: fix a deadlock in util_queue_finish

Cc: 18.0 18.1 <mesa-stable@lists.freedesktop.org>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>

meson: fix race condition revealed by using 0.44

Previously there was a special target that blocked for the generation of
anv_entrypoints.h, with meson 0.44 we don't need this, we can use a new
language feature instead. The problem is that previously that blocking
target would hide a race condition for the generation of another header,
anv_extensions.h. Now the build sometimes fails when anv_extensions.h is
not generated in time.

v2: - clarify the race condition in the commit message (Emil)

CC: Mark Janes <mark.a.janes@intel.com>
Fixes: 92550d9b16d2b295bdac087f31b1fd6d0f808e02
       ("meson: remove workaround for custom target creating .h and .c files")
Signed-off-by: Dylan Baker <dylan.c.baker@intel.com>
Reviewed-by: Scott D Phillips <scott.d.phillips@intel.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>

bin: force git show to use default pretty setting

I have pretty default to short, which breaks this script.

v2: - Fix both places that don't define a --pretty (Emil)

cc: Juan A. Suarez <jasuarez@igalia.com>
Signed-off-by: Dylan Baker <dylan.c.baker@intel.com>
Reviewed-by: Andres Gomez <agomez@igalia.com> (v1)
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>

mesa: add TBO support for GL_EXT_texture_norm16

Earlier plumbing missed interaction with texture buffer objects.

Fixes: 7f467d4f73 "mesa: GL_EXT_texture_norm16 extension plumbing"
Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

ac: fix texture query LOD for 1D textures on GFX9

1D textures are allocated as 2D which means we only need
one coordinate for texture query LOD.

Fixes: 625dcbbc456 ("amd/common: pass address components individually to
ac_build_image_intrinsic")
Cc: 18.1 <mesa-stable@lists.freedesktop.org>
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>

etnaviv: remove not needed includes

Signed-off-by: Christian Gmeiner <christian.gmeiner@gmail.com>
Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>

etnaviv: remove redundant include

Signed-off-by: Christian Gmeiner <christian.gmeiner@gmail.com>
Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>

glsl: replace some asserts with unreachable when processing the ast

Reviewed-by: Matt Turner <mattst88@gmail.com>

mesa: drop the buffer mode param from the DrawBuffer driver function

No drivers used it.

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

anv/icl: Enable Vulkan on Ice Lake

This patch enables the Vulkan driver on Ice Lake h/w
with added warning about preliminary support.

Signed-off-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Nanley Chery <nanley.g.chery@intel.com>

anv: enable VK_EXT_shader_viewport_index_layer

Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>

anv/allocator: Don't shrink either end of the block pool

Previously, we only tried to ensure that we didn't shrink either end
below what was already handed out.  However, due to the way we handle
relocations with block pools, we can't shrink the back end at all.  It's
probably best to not shrink in either direction.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=105374
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=106147
Tested-by: Eero Tamminen <eero.t.tamminen@intel.com>
Reviewed-by: Scott D Phillips <scott.d.phillips@intel.com>
Cc: mesa-stable@lists.freedesktop.org

broadcom/vc5: Add support for centroid varyings.

It would be nice to share the flags packet emit logic with flat shade
flags, but I couldn't come up with a good way while still using our pack
macros.  We need to refactor this to shader record setup at compile time,
anyway.

Fixes ext_framebuffer_multisample-interpolation * centroid-*

broadcom/vc5: Add an assert about GFXH-1559.

Our TF outputs always start at 6 or 7 currently, so we don't hit the
broken 8 case.  Let's make sure that doesn't change somehow.

broadcom/vc5: Add validation that we don't violate GFXH-1633 requirements.

We don't use ldunifa yet, but we will eventually for UBOs.

broadcom/vc5: Add validation that we don't violate GFXH-1625 requirements.

We don't use TMUWT yet, but we will once we do SSBOs.

broadcom/vc5: Implement GFXH-1742 workaround (emit 2 dummy stores on 4.x).

This should fix help with intermittent GPU hangs in tests switching
formats while rendering small frames.  Unfortunately, it didn't help with
the tests I'm having troubles with.

broadcom/vc5: Add QPU validation for register writes after thrend.

The next shader gets to start writing the register file during these
slots, so make sure we don't stomp over them.

The only case of hitting this that I could imagine would be dead writes.

st: Choose a 2101010 format for GL_RGB/GL_RGBA with a 2_10_10_10 type.

GLES's GL_EXT_texture_type_2_10_10_10_REV allows uploading this type to an
unsized internalformat, and it should be non-color-renderable.
fbobject.c's implementation of the check for color-renderable is checks
that the texture has a 2101010 mesa format, so make sure that we have
chosen a 2101010 format so that check can do what it meant to.

Fixes KHR-GLES3.packed_pixels.pbo_rectangle.rgb on vc5.

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

st/mesa: fix missing setting of _ElementSize in new_draw_rasterpos_stage

With this patch, _ElementSize is initialized along with the rest
of the vertex array attributes in new_draw_rasterpos_stage().
This fixes a crash in st_pipe_vertex_format() when running
topogun-1.06-orc-84k-resize trace file with VMware svga driver.

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

st/va: Fix typos

s/attibute/attribute/
s/suface/surface/

v2: rebased(Leo)

Reviewed-by: Leo Liu <leo.liu@amd.com>

st/va: Fix potential buffer overread

VASurfaceAttribExternalBuffers.pitches is indexed by
plane. Current implementation only supports single plane layout.

Reviewed-by: Kristian H. Kristensen <hoegsberg@chromium.org>
Reviewed-by: Leo Liu <leo.liu@amd.com>

radeon/vcn: fix mpeg4 msg buffer settings

Previous bit-fields assignments are incorrect and will result certain mpeg4
decode failed due to wrong flag values. This patch fixes these assignments.

Signed-off-by: Boyuan Zhang <boyuan.zhang@amd.com>
Reviewed-by: Leo Liu <leo.liu@amd.com>

radeon: Drop broken front_buffer_reading/drawing optimization

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Acked-by: Timothy Arceri <tarceri@itsqueeze.com>

radeon: Use _mesa_is_front_buffer_drawing

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Acked-by: Timothy Arceri <tarceri@itsqueeze.com>

radv: set ac_surf_info::num_channels correctly

num_channels has been introduced since "ac/surface: don't set
the display flag for obviously unsupported cases".

Based on RadeonSI.

Fixes: e29facff315 ("ac/surface: don't set the display flag for obviously unsupported cases (v2)")
Cc: 18.1 <mesa-stable@lists.freedesktop.org>
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

radv: fix DCC enablement since partial MSAA implementation

dcc_msaa_allowed is always false on GFX9+ and only true on VI
if RADV_PERFTEST=dccmsaa is set. This means DCC was disabled
in some situations where it should not.

This is likely going to fix a performance regression.

Fixes: 2f63b3dd09 ("radv: enable DCC for MSAA 2x textures on VI under an option")
Cc: 18.1 <mesa-stable@lists.freedesktop.org>
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

nir/opt_constant_folding: fix folding of 8 and 16 bit ints

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Jose Maria Casanova Crespo <jmcasanova@igalia.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>

nir: print 8 and 16 bit constants correctly

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Rob Clark <robdclark@gmail.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>

nir: support converting to 8-bit integers in nir_type_conversion_op

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Jose Maria Casanova Crespo <jmcasanova@igalia.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>

spirv: Don’t check for NaN for most OpFOrd* comparisons

For all of the OpFOrd* comparisons except OpFOrdNotEqual the hardware
should probably already return false if one of the operands is NaN so
we don’t need to have an explicit check for it. This seems to at least
work on Intel hardware. This should reduce the number of instructions
generated for the most common comparisons.

For what it’s worth, the original code to handle this was added in
e062eb6415de3a. The commit message for that says that it was to fix
some CTS tests for OpFUnord* opcodes. Even if the hardware doesn’t
handle NaNs this patch shouldn’t affect those tests. At any rate they
have since been moved out of the mustpass list. Incidentally those
tests fail on the nvidia proprietary driver so it doesn’t seem like
handling NaNs correctly is a priority.

Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>

Intel: Add a Kaby Lake PCI ID

v2: Branding changed

Signed-off-by: Matt Atwood <matthew.s.atwood@intel.com>
Reviewed-by: Rafael Antognolli <rafael.antognolli@intel.com>

gallium/util: Fix incorrect refcounting of separate stencil.

The driver may have a reference on the separate stencil buffer for some
reason (like an unflushed job using it), so we can't directly free the
resource and should instead just decrement the refcount that we own.
Fixes double-free in KHR-GLES3.packed_depth_stencil.blit.depth32f_stencil8
on vc5.

Fixes: e94eb5e6000e ("gallium/util: add u_transfer_helper")
Reviewed-by: Rob Clark <robdclark@gmail.com>

broadcom/vc5: Fix reloads of separate stencil buffers.

Like for stores, we need to emit a separate load_general packet.

broadcom/vc5: Fix cpp of MSAA surfaces on 4.x.

The internal-type-bpp path is for surfaces that get stored in the raw TLB
format.  For 4.x, we're storing MSAA as just 2x width/height at the
original format.

broadcom/vc5: Implement stencil blits using RGBA.

Fixes piglit fbo-depthstencil blit default_fb

broadcom/vc5: Remove leftover vc4 MSAA lowering setup in the FS key.

broadcom/vc5: Fix tile load/store of MSAA surfaces on 4.x.

For single-sample we have to always program SAMPLE_0, but for multisample
we want to store all the samples.

travis: update libva required version

Commit fa328456e8f29 added VP9 config support, but this needs a newer
libva version, 1.7.0 or above.

Fixes: fa328456e8f ("st/va: add VP9 config to enable profile2")
CC: 18.1 <mesa-stable@lists.freedesktop.org>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>

mesa: GL_EXT_texture_norm16 extension plumbing

Patch enables use of short and unsigned short data for texture uploads,
rendering and reading of framebuffers within the restrictions specified
in GL_EXT_texture_norm16 spec.

Patch also enables those 16bit format layout qualifiers listed in
GL_NV_image_formats that depend on EXT_texture_norm16.

v2: expose extension with dummy_true
    fix layout qualifier map changes (Ilia Mirkin)

v3: use _mesa_has_EXT_texture_norm16, other fixes
    and cleanup (Ilia Mirkin)

v4: fix rest of the issues found

Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Acked-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

meson: Fix with_intel_vk and with_amd_vk variables

Fixes: 5608d0a2cee "meson: use array type options"
Cc: Dylan Baker <dylan@pnwbakers.com>
Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>

draw: fix different sign logic when clipping

The logic was flawed, since mul(x,y) will be <= 0 (exactly 0) when
the sign is the same but both numbers are sufficiently small
(if the product is smaller than 2^-128).
This could apparently lead to emitting a sufficient amount of
additional bogus vertices to overflow the allocated array for them,
hitting an assertion (still safe with release builds since we just
aborted clipping after the assertion in this case - I'm however unsure
if this is now really no longer possible, so that code stays).
Not sure if the additional vertices could cause other grief, I didn't
see anything wrong even when hitting the assertion.

Essentially, both +-0 are treated as positive (the vertex is considered
to be inside the clip volume for this plane), so integrate the logic
determining different sign into the branch there.

Reviewed-by: Jose Fonseca <jfonseca@vmware.com>

draw: simplify clip null tri logic

Simplifies the logic when to emit null tris (albeit the reasons why we
have to do this remain unclear).
This is strictly just logic simplification, the behavior doesn't change
at all.

Reviewed-by: Jose Fonseca <jfonseca@vmware.com>

nvc0/ir: all short immediates are sign-extended, adjust LIMM test

Some analysis suggests that all short immediates are sign-extended. The
insnCanLoad logic already accounted for this, but we could still pick
the wrong form when emitting actual instructions that support both short
and long immediates (with the long form usually having additional
restrictions that insnCanLoad should be aware of).

This also reverses a bunch of commits that had previously "worked
around" this issue in various emitters:

9c63224540ef: gm107/ir: make use of ADD32I for all immediates
83a4f28dc27b: gm107/ir: make use of LOP32I for all immediates
b84c97587b4a: gm107/ir: make use of IMUL32I for all immediates
d30768025a22: gk110/ir: make use of IMUL32I for all immediates

as well as the original import for UMUL in the nvc0 emitter.

Reported-by: Karol Herbst <kherbst@redhat.com>
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Tested-by: Karol Herbst <kherbst@redhat.com>

mesa: call DrawBufferAllocate driver hook in update_framebuffer for windows-system FB

When draw buffers are changed on a bound framebuffer, DrawBufferAllocate()
hook should be called. However, it is missing in update_framebuffer with
window-system framebuffer, in which FB's draw buffer state should match
context state, potentially resulting in a change.

Note: This is needed because gallium delays creating the front buffer,
      i965 works fine without this change.

V2 (Timothy Arceri):
 - Rebased on merged/simplified DrawBuffer driver function
 - Move DrawBuffer call outside fb->ColorDrawBuffer[0] !=
   ctx->Color.DrawBuffer[0] check to make piglit pass.

v3 (Timothy Arceri):
 - Call new DrawBuffaerAllocate() driver function.

Tested-by: Dieter Nützel <Dieter@nuetzel-hh.de> (v2)
Reviewed-by: Brian Paul <brianp@vmware.com> (v2)
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=99116

st/mesa: add new driver function DrawBufferAllocate

Unlike some of the classic drivers the st was only using DrawBuffer()
to allocated some buffers on-demand. Creating a separate function
will allow us to call it from update_framebuffer() in the following
patch without regressing some of the older classic drivers.

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

mesa: some C99 tidy ups for framebuffer.c

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>

meson: Fix no-rtti in llvm detection

Because I clearly wasn't thinking and clearly didn't do a good job
testing. Sigh

Fixes: c5a97d658ec19cc02719d7f86c1b0715e3d9ffc4
       ("meson: fix builds against LLVM built without rtti")
Signed-off-by: Dylan Baker <dylan.c.baker@intel.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>

meson: use new warning function

Instead of emulating it with message.

Signed-off-by: Dylan Baker <dylan.c.baker@intel.com>
Reviewed-by: Eric Anholt <eric@anholt.net>

meson: use array type options

This option type is nice since it involves less converting strings into
lists, and because it validates the values that are provided.

v2: - Set with_any_vk to true if any vulkan driver is built (Eric)

Signed-off-by: Dylan Baker <dylan.c.baker@intel.com>
Reviewed-by: Eric Anholt <eric@anholt.net>

meson: fix builds against LLVM built without rtti

Building without rtti is a frought with peril, but it's something that
autotools supports so we need to support it too.

Since we've moved to version 0.44 as a whole we can use the meson
functionality for accessing random llvm-config options we can check for
rtti and add -fno-rtti to all C++ code accordingly.

Signed-off-by: Dylan Baker <dylan.c.baker@intel.com>

meson: remove dummy_cpp

meson has gotten pretty smart about tracking C and C++ dependencies
(internal and external), and using the right linker. This wasn't always
the case and we created empty c++ files to force the use of the c++
linker. We don't need that any more.

Signed-off-by: Dylan Baker <dylan.c.baker@intel.com>
Reviewed-by: Eric Anholt <eric@anholt.net>

meson: allow empty sources when using link_whole

meson used to get grumpy if the sources list was empty, even when using
--whole-archive (link_whole). In more recent versions that's not true,
so remove the workaround.

Signed-off-by: Dylan Baker <dylan.c.baker@intel.com>
Reviewed-by: Eric Anholt <eric@anholt.net>

meson: remove workaround for custom target creating .h and .c files

In more modern versions of meson a custom_target returns an index-able
object. This allows us to create accurate dependency models for targets
that rely only on the header and not on the code from anv_entrypoints.

Signed-off-by: Dylan Baker <dylan.c.baker@intel.com>
Reviewed-by: Eric Anholt <eric@anholt.net>

meson: raise required version to 0.44.1

We have already required 0.44 for building clover and swr, so it was
already partially required. This just makes it required across the board
instead of just for clover and swr.

There is a bug in 0.44 which makes it impossible to build mesa in some
configurations, so require 0.44.1 which fixes this.

Signed-off-by: Dylan Baker <dylan.c.baker@intel.com>
Reviewed-by: Eric Anholt <eric@anholt.net>

meson: fix graw-xlib after auxiliary consolidation

This one's completely my fault, I didn't do good enough testing after
rebasing and this got missed.

Fixes: d28c24650110c130008be3d3fe584520ff00ceb1
       ("meson: build graw tests")
Signed-off-by: Dylan Baker <dylan.c.baker@intel.com>
Reviewed-by: Eric Anholt <eric@anholt.net>

meson: only build mesa_st tests when build-tests is true

Since we have an option to turn test building on and off, we should
honor that.

Fixes: 34cb4d0ebc14663113705beae63dd52b9d1b2d87
       ("meson: build tests for gallium mesa state tracker")
Signed-off-by: Dylan Baker <dylan.c.baker@intel.com>
Reviewed-by: Eric Anholt <eric@anholt.net>

meson: don't build classic mesa tests without dri_drivers

Since mesa_classic is build-on-demand the tests will create a demand and
add a bunch of extra compilation.

Fixes: 43a6e84927e3b1290f6f211f5dfb184dfe5a719e
       ("meson: build mesa test.")
Signed-off-by: Dylan Baker <dylan.c.baker@intel.com>
Reviewed-by: Eric Anholt <eric@anholt.net>

i965/meta_util: Re-enable sRGB-encoded fast-clears on CNL

The paths which sample with the clear color are now using a getter which
performs the sRGB decode needed to enable this fast clear.

This path can be exercised by fast-clearing a texture, then performing
an operation which requires sRGB decoding. Test coverage for this
feature is provided with the following tests:

* Shader texture calls:
  - spec@ext_texture_srgb@tex-srgb

* Shader texelfetch calls:
  - spec@arb_framebuffer_srgb@fbo-fast-clear
  - spec@arb_framebuffer_srgb@msaa-fast-clear

* Blending:
  - spec@arb_framebuffer_srgb@arb_framebuffer_srgb-fast-clear-blend

* Blitting:
  - spec@arb_framebuffer_srgb@blit texture srgb msaa enabled clear

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

i965/miptree: Extend the sRGB-blending WA to future platforms

The blending issue seems to be present on CNL as well.

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

i965: Add and use a getter for the clear color

It returns both the inline clear color and a clear address which points
to the indirect clear color buffer (or NULL if unused/non-existent).
This getter allows CNL to sample from fast-cleared sRGB textures
correctly by doing the needed sRGB-decode on the clear color (inline)
and making the indirect clear color buffer unused.

v2 (Rafael):
* Have a more detailed commit message.
* Add a comment on the sRGB conversion process.

Reviewed-by: Rafael Antognolli <rafael.antognolli@intel.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>

util/srgb: Add a float sRGB -> linear helper

Reviewed-by: Nanley Chery <nanley.g.chery@intel.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>

i965/wm_surface_state: Use the clear address if clear_bo is non-NULL

We want to add and use a getter that turns off the indirect path by
returning zero for the clear color bo and offset.

v2: Fix usage of "clear address" in commit message (Jason).

Reviewed-by: Rafael Antognolli <rafael.antognolli@intel.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>

i965: Add and use a single miptree aux_buf field

We want to add and use a function that accesses the auxiliary buffer's
clear_color_bo and doesn't care if it has an MCS or HiZ buffer
specifically.

v2 (Jason Ekstrand):
* Drop intel_miptree_get_aux_buffer().
* Mention CCS in the aux_buf field.

Reviewed-by: Rafael Antognolli <rafael.antognolli@intel.com> (v1)
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>

i965: Add and use a getter for the miptree aux buffer

Make the next patch easier to read by eliminating most of the would-be
duplicate field accesses now.

v2: Update the HiZ comment instead of deleting it (Rafael).

Reviewed-by: Rafael Antognolli <rafael.antognolli@intel.com>

gm107/ir/lib: fix sched in div u32 builtin

Imad needs to set a read barrier.

With significant big work groups I was getting wrong results for div u32. Turns
out the issue was with the sched opcodes.

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>

intel/compiler: Add scheduler deps for instructions that implicitly read g0

Otherwise the scheduler can move the writes after the reads.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=95009
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=95012
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Tested-by: Mark Janes <mark.a.janes@intel.com>
Cc: Clayton A Craft <clayton.a.craft@intel.com>
Cc: mesa-stable@lists.freedesktop.org

intel/compiler: Silence unused parameter warnings in empty vec4_instruction_scheduler methods

src/intel/compiler/brw_schedule_instructions.cpp: In member function ‘virtual void vec4_instruction_scheduler::count_reads_remaining(backend_instruction*)’:
src/intel/compiler/brw_schedule_instructions.cpp:764:72: warning: unused parameter ‘be’ [-Wunused-parameter]
 vec4_instruction_scheduler::count_reads_remaining(backend_instruction *be)
                                                                        ^~
src/intel/compiler/brw_schedule_instructions.cpp: In member function ‘virtual void vec4_instruction_scheduler::setup_liveness(cfg_t*)’:
src/intel/compiler/brw_schedule_instructions.cpp:769:51: warning: unused parameter ‘cfg’ [-Wunused-parameter]
 vec4_instruction_scheduler::setup_liveness(cfg_t *cfg)
                                                   ^~~
src/intel/compiler/brw_schedule_instructions.cpp: In member function ‘virtual void vec4_instruction_scheduler::update_register_pressure(backend_instruction*)’:
src/intel/compiler/brw_schedule_instructions.cpp:774:75: warning: unused parameter ‘be’ [-Wunused-parameter]
 vec4_instruction_scheduler::update_register_pressure(backend_instruction *be)
                                                                           ^~
src/intel/compiler/brw_schedule_instructions.cpp: In member function ‘virtual int vec4_instruction_scheduler::get_register_pressure_benefit(backend_instruction*)’:
src/intel/compiler/brw_schedule_instructions.cpp:779:80: warning: unused parameter ‘be’ [-Wunused-parameter]
 vec4_instruction_scheduler::get_register_pressure_benefit(backend_instruction *be)
                                                                                ^~
src/intel/compiler/brw_schedule_instructions.cpp: In member function ‘virtual int vec4_instruction_scheduler::issue_time(backend_instruction*)’:
src/intel/compiler/brw_schedule_instructions.cpp:1550:61: warning: unused parameter ‘inst’ [-Wunused-parameter]
 vec4_instruction_scheduler::issue_time(backend_instruction *inst)
                                                             ^~~~

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>

intel/compiler: Silence unused parameter warning in compile_cs_to_nir

src/intel/compiler/brw_fs.cpp: In function ‘nir_shader* compile_cs_to_nir(const brw_compiler*, void*, const brw_cs_prog_key*, brw_cs_prog_data*, const nir_shader*, unsigned int)’:
src/intel/compiler/brw_fs.cpp:7205:44: warning: unused parameter ‘prog_data’ [-Wunused-parameter]
                   struct brw_cs_prog_data *prog_data,
                                            ^~~~~~~~~

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>

intel/compiler: Silence unused parameter warnings in generate_foo methods

Since all of the fs_generator::generate_foo methods take a fs_inst * as
the first parameter, just remove the name to quiet the compiler.

src/intel/compiler/brw_fs_generator.cpp: In member function ‘void fs_generator::generate_barrier(fs_inst*, brw_reg)’:
src/intel/compiler/brw_fs_generator.cpp:743:41: warning: unused parameter ‘inst’ [-Wunused-parameter]
 fs_generator::generate_barrier(fs_inst *inst, struct brw_reg src)
                                         ^~~~
src/intel/compiler/brw_fs_generator.cpp: In member function ‘void fs_generator::generate_discard_jump(fs_inst*)’:
src/intel/compiler/brw_fs_generator.cpp:1326:46: warning: unused parameter ‘inst’ [-Wunused-parameter]
 fs_generator::generate_discard_jump(fs_inst *inst)
                                              ^~~~
src/intel/compiler/brw_fs_generator.cpp: In member function ‘void fs_generator::generate_pack_half_2x16_split(fs_inst*, brw_reg, brw_reg, brw_reg)’:
src/intel/compiler/brw_fs_generator.cpp:1675:54: warning: unused parameter ‘inst’ [-Wunused-parameter]
 fs_generator::generate_pack_half_2x16_split(fs_inst *inst,
                                                      ^~~~
src/intel/compiler/brw_fs_generator.cpp: In member function ‘void fs_generator::generate_shader_time_add(fs_inst*, brw_reg, brw_reg, brw_reg)’:
src/intel/compiler/brw_fs_generator.cpp:1743:49: warning: unused parameter ‘inst’ [-Wunused-parameter]
 fs_generator::generate_shader_time_add(fs_inst *inst,
                                                 ^~~~
src/intel/compiler/brw_vec4_generator.cpp: In function ‘void generate_set_simd4x2_header_gen9(brw_codegen*, brw::vec4_instruction*, brw_reg)’:
src/intel/compiler/brw_vec4_generator.cpp:1412:52: warning: unused parameter ‘inst’ [-Wunused-parameter]
                                  vec4_instruction *inst,
                                                    ^~~~
src/intel/compiler/brw_vec4_generator.cpp: In function ‘void generate_mov_indirect(brw_codegen*, brw::vec4_instruction*, brw_reg, brw_reg, brw_reg, brw_reg)’:
src/intel/compiler/brw_vec4_generator.cpp:1430:41: warning: unused parameter ‘inst’ [-Wunused-parameter]
                       vec4_instruction *inst,
                                         ^~~~
src/intel/compiler/brw_vec4_generator.cpp:1432:63: warning: unused parameter ‘length’ [-Wunused-parameter]
                       struct brw_reg indirect, struct brw_reg length)
                                                               ^~~~~~
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>

broadcom/vc5: Set up internal_format for imported resources.

Without this, we'd assertion fail in u_transfer_helper when mapping an
imported resource.

broadcom/vc5: Assert that created BOs have offset != 0.

The kernel shouldn't return a bo at NULL, and the HW special-cases NULL
address values for things like OQs.

broadcom/vc5: Don't allocate simulator BOs at offset 0.

The kernel won't return us BOs at offset 0 (because things like OQs
wouldn't work there), so we shouldn't in the simulator either.

broadcom/vc5: Add sim support for the GET_BO_OFFSET ioctl.

Otherwise we'd crash immediately upon importing a BO through EGL
interfaces.

broadcom/vc5: Treat imports of DRM_FORMAT_MOD_INVALID BOs as linear.

We don't have any kernel metadata about BO tiling, so this probably is all
we should do for the moment.

i965: expose MESA_FORMAT_R8G8B8A8_SRGB visual

Exposing the visual makes following dEQP tests pass on Android:

   dEQP-EGL.functional.wide_color.window_8888_colorspace_srgb
   dEQP-EGL.functional.wide_color.pbuffer_8888_colorspace_srgb

Visual is exposed only when DRI_LOADER_CAP_RGBA_ORDERING is set.

Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

dri: Add __DRI_IMAGE_FORMAT_SABGR8

Add format definition and required plumbing to create images.
Note that there is no match to drm_fourcc definition, just like
with existing _DRI_IMAGE_FOURCC_SARGB8888.

Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

Revert "st/dri: Fix dangling pointer to a destroyed dri_drawable"

This reverts commit dab02dea3411d325a5aee6cda5b581e61396ecc6.

It causes crashes of qtcreator and firefox.

Fixes: dab02de "st/dri: Fix dangling pointer to a destroyed dri_drawable"
Cc: 18.0 18.1 <mesa-stable@lists.freedesktop.org>

gallivm: dump bitcode before optimization

If we dump the bitcode for off-line debug purposes, we really want the
pre-optimized bitcode, otherwise it's useless in identifying problems
with IR optimization (if you have a shader which takes an hour to do
IR optimization, it's also nice you don't have to wait that hour...).
Also, print out the function passes for opt which correspond to what
was used for jit compilation (and also the opt level for codegen).
Using opt/llc this way should then pretty much mimic what was done
for jit. (When specifying something like -time-passes
-debug-pass=[Structure|Arguments] (for either opt or llc) that also
gives very useful information in which passes all the time was spent,
and which passes are really run along with the order - llvm will add
passes due to dependencies on its own, and of course -O2 for llc
comes with a ~100 pass list.)

Reviewed-by: Jose Fonseca <jfonseca@vmware.com>

gallivm: (trivial) do division by 1000 with int64

Conversion to int can otherwise overflow if compile times are over
~71min. (Yes this can happen...)

Reviewed-by: Jose Fonseca <jfonseca@vmware.com>