vulkan/wsi/radv: add initial prime support (v1.1)

This is a complete rewrite of my previous rfc patches.

This adds the ability to present to a different GPU that rendering
using a driver side operation that can copy from the tiled to
linear shared image.

This does prime support completely in the swapchain present code,
and each queue has a precreated command buffer for each image
and for the each queue family. This means presenting should work
on graphics and compute queues and transfer in the future.

v1.1: initialise needs_linear_copy in swapchain.

Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Tested-by: Mike Lothian <mike@fireburn.co.uk>
Signed-off-by: Dave Airlie <airlied@redhat.com>

radv/ac: Add integer->integer casts.

Signed-off-by: Bas Nieuwenhuizen <basni@google.com>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Acked-by: Edward O'Callaghan <funfunctor@folklore1984.net>

check: add support for running test as standalone

Signed-off-by: Eric Engestrom <eric@engestrom.ch>

check: make any failure fatal

Previously, only the last error code was returned.
Using `set -e` makes the script quit on any unhandled error.

Signed-off-by: Eric Engestrom <eric@engestrom.ch>

check: mark two tests are requiring bash

Requirement was removed just before pushing, but it's actually needed
for heredocs (`<<<`).

Signed-off-by: Eric Engestrom <eric@engestrom.ch>

st/nine: Drop USER_INDEX_BUFFERS check

This fixes 4a883966c1f74f43afc145d2c3d27af7b8c5e01a where the
PIPE_CAP was removed.

Now USER_INDEX_BUFFERS are always enabled remove the check and only
check for cmst_active directly.

v2: Axel pointed out the code was still needed when cmst was inactive,
    Rebase on master too
v3: Drop struct member user_ibufs also && fixup shortlog (Edward).
v4: Fix negation
v5: Use the right variable name csmt != cmst

Fixes: 4a883966c1f7 ("gallium: remove PIPE_CAP_USER_INDEX_BUFFERS")
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=99953
Reported-and-tested-by: Vinson Lee <vlee@freedesktop.org> (v1)
Cc: Marek Olšák <marek.olsak@amd.com>
Cc: Axel Davy <axel.davy@ens.fr>
Signed-off-by: Edward O'Callaghan <funfunctor@folklore1984.net>
Signed-off-by: Mike Lothian <mike@fireburn.co.uk>
Reviewed-by: Edward O'Callaghan <funfunctor@folklore1984.net>

st/nine: make use of common uploaders v4

Make use of common uploaders that landed recently to Mesa

v2: fixed formatting, broken due to thunderbird configuration

v3: per Axel comment: added a comment into NineDevice9_DrawPrimitiveUP

v4: per Axel comment: changed style of the comment

compiler: style clean-ups in blob.h

Reviewed-by: Elie Tournier <tournier.elie@gmail.com>

svga: fix MSVC build error after PIPE_CAP_USER_INDEX_BUFFERS removal

Need to specify the zero for the struct initializer.  My earlier test
of the patch series was with MinGW, not MSVC.

Trivial.

vc4: Lazily emit our FS/VS input loads.

This reduces register pressure in both types of shaders, by reordering the
input loads from the var->data.driver_location order to whatever order
they appear first in the NIR shader.  These instructions aren't
reorderable at our QIR scheduling level because the FS takes two in
lockstep to do an interpolation, and the VS takes multiple read
instructions in a row to get a whole vec4-level attribute read.

shader-db impact:
total instructions in shared programs: 76666 -> 76590 (-0.10%)
instructions in affected programs:     42945 -> 42869 (-0.18%)
total max temps in shared programs: 9395 -> 9208 (-1.99%)
max temps in affected programs:     2951 -> 2764 (-6.34%)

Some programs get their max temps hurt, depending on the order that the
load_input intrinsics appear, because we end up being unable to copy
propagate an older VPM read into its only use.

vc4: Refactor the load_input code out of the intrinsic code.

It's going gain most of ntq_setup_inputs(), so simplify it first.

vc4: Track the last block we emitted at the top level.

This will be used for delaying our VPM reads (which must be unconditional)
until just before they're used.

vc4: Emit max number of temps in the shader-db output.

We need to be paying attention to optimization's impact on this -- even if
we reduce instruction count, increasing max temps in general is likely to
cause us to fail to register allocate on some shaders, which means that
those won't run at all.

util/disk_cache: Use backward compatible st_mtime.

Fix Mac OS X build error.

  CC       libmesautil_la-disk_cache.lo
In file included from disk_cache.c:46:
./disk_cache.h:57:20: error: no member named 'st_mtim' in 'struct stat'
   *timestamp = st.st_mtim.tv_sec;
                ~~ ^

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=99918
Fixes: 207e3a6e4b ("util/radv: move *_get_function_timestamp() to utils")
Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>

glsl: Fix missing-braces warning.

  CXX    glsl/ast_to_hir.lo
glsl/ast_to_hir.cpp: In member function 'virtual ir_rvalue* ast_declarator_list::hir(exec_list*, _mesa_glsl_parse_state*)':
glsl/ast_to_hir.cpp:4846:42: warning: missing braces around initializer for 'unsigned int [16]' [-Wmissing-braces]

Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Reviewed-by: Andres Gomez <agomez@igalia.com>

ac: silence a warning

trivial

radeonsi: fix broken tessellation on Carrizo and Stoney

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=99850

Cc: 13.0 17.0 <mesa-stable@lists.freedesktop.org>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>

st/mesa: don't update unrelated states in non-draw calls such as Clear

If a VAO isn't bound and u_vbuf isn't enabled because of the Core profile,
we'll get user vertex buffers in drivers if we update vertex buffers
in glClear. So don't do that.

This fixes a regression since disabling u_vbuf for Core profiles.

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

st/mesa: set blend state for PBO readbacks

v2: restore the state

Cc: 13.0 17.0 <mesa-stable@lists.freedesktop.org>
Reviewed-by: Brian Paul <brianp@vmware.com>

st/mesa: reset sample_mask, min_sample, and render_condition for PBO ops

Cc: 13.0 17.0 <mesa-stable@lists.freedesktop.org>
Reviewed-by: Brian Paul <brianp@vmware.com>

st/mesa: don't check st->vp in update_clip

The clip state is updated before VS, so it can be NULL for the first draw
call. Just remove the unnecessary dependency on st->vp.

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

trace: remove pipe_resource wrapping

Not needed. ddebug does the same thing. The limitation is that drivers
can only use pipe_resource::screen through pipe_resource_reference.

This unbreaks trace, because pipe_context uploaders aren't wrapped,
so trace doesn't understand buffers returned by them.

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

gallium: remove PIPE_CAP_USER_INDEX_BUFFERS

all drivers support it

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Tested-by: Brian Paul <brianp@vmware.com> (VMware driver only)

st/mesa: assume all drivers support user index buffers

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Tested-by: Brian Paul <brianp@vmware.com> (VMware driver only)

svga: implement user index buffers

Reviewed-by: Brian Paul <brianp@vmware.com>
Tested-by: Brian Paul <brianp@vmware.com> (VMware driver only)

freedreno: add support for user index buffers

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

etnaviv: add support for user index buffers

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

gallium/util: add new helpers for user index buffer uploading

v3: split from the etnaviv patch; fix new_ib.buffer leak

Reviewed-by: Brian Paul <brianp@vmware.com>
Tested-by: Brian Paul <brianp@vmware.com> (VMware driver only)

nir: delete magic number

Signed-off-by: Elie Tournier <tournier.elie@gmail.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>

gallium/util: (trivial) fix util_clear_render_target

the format of the rt can be different than the one of the texture, so must
propagate the format explicitly to the helper. Broken since
3f9c5d62441eba38e8b1592aba965ed5db6fd89b (but unused by st/mesa).

util: automake: add sha1/README to the tarball

Suggested-by: Andreas Boll <andreas.boll.dev@gmail.com>
Signed-off-by: Emil Velikov <emil.velikov@collabora.com>

mapi: remove unused mapi.[ch]

The final user of it was st/vega.

Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Andreas Boll <andreas.boll.dev@gmail.com>

blorp: automake: add TODO to the tarball

Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Andreas Boll <andreas.boll.dev@gmail.com>

anv: automake: add TODO to the tarball

Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Andreas Boll <andreas.boll.dev@gmail.com>

vc4: automake: add the kernel/README to the tarball

Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Andreas Boll <andreas.boll.dev@gmail.com>

nir: automake: add the README to the tarball

Similar to other accompanying documentation we have in-tree.
For example glsl/README.

Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Andreas Boll <andreas.boll.dev@gmail.com>

radv/entrypoints: Only generate entrypoints for supported features

This changes the way radv_entrypoints_gen.py works from generating a
table containing every single entrypoint in the XML to just the ones
that we actually need.  There's no reason for us to burn entrypoint
table space on a bunch of NV extensions we never plan to implement.

RADV implements VK_AMD_draw_indirect_count, so add that to the list.

Port of 114c281e707919a1420d3a20219ce4392d8b360d
"and/entrypoints: Only generate entrypoints for supported features"

Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
Acked-by: Dave Airlie <airlied@redhat.com>

main/performance_query: s/GLboolean/bool/

Ideally would have caught these when adding the interface but this just
switches a few return types for the INTEL_performance_query backend
interface to bool instead of GLboolean.

Signed-off-by: Robert Bragg <robert@sixbynine.org>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>

eglapi: replace linear entrypoint search with binary search

Tested with dEQP-EGL.functional.get_proc_address.*

Signed-off-by: Eric Engestrom <eric@engestrom.ch>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>

egl: make sure entrypoints list is always sorted

Starting with the next commit, badly sorting this list will break the
eglGetProcAddress().

Signed-off-by: Eric Engestrom <eric@engestrom.ch>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>

egl: distribute all tests

Signed-off-by: Eric Engestrom <eric@engestrom.ch>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>

eglapi: move entrypoints list out to its own file

This will allow us to make sure the list is always sorted in the next
commit.

Signed-off-by: Eric Engestrom <eric@engestrom.ch>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>

eglapi: sort entrypoints list

Let's make that comment true.
If will also be necessary in a couple commits (using bsearch).

Signed-off-by: Eric Engestrom <eric@engestrom.ch>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>

eglapi: use macro to map entrypoints to functions

As of the last 3 commits, there's a function for each entrypoint.

Signed-off-by: Eric Engestrom <eric@engestrom.ch>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>

eglapi: add entrypoint for eglClientWaitSyncKHR

Signed-off-by: Eric Engestrom <eric@engestrom.ch>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>

eglapi: add entrypoint for eglDestroySyncKHR

Signed-off-by: Eric Engestrom <eric@engestrom.ch>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>

eglapi: add entrypoint for eglDestroyImageKHR

Signed-off-by: Eric Engestrom <eric@engestrom.ch>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>

st/va: Fix up YV12 to NV12 putImage conversion

Use the utility u_copy_nv12_from_yv12 to implement this similarly to
how it's been done in the VPAU state tracker. The old code mixed up
planes and fields and didn't correctly handle video surfaces in
interlaced format.

Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>

st/vdpau: Provide YV12 to NV12 putBits conversion v2

mplayer likes putting YV12 data, and if there is a buffer format mismatch,
the vdpau state tracker would try to reallocate the video surface as an
YV12 surface. A virtual driver doesn't like reallocating and doesn't like YV12
surfaces, so if we can't support YV12, try an YV12 to NV12 conversion
instead.

Also advertize that we actually can do the getBits and putBits conversion.

v2: A previous version of this patch prioritized conversion before
reallocating. This has been changed to prioritize reallocating in this version.

Cc: Christian König <christian.koenig@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>

configure.ac: check require_basic_egl only if egl enabled

Otherwise the configuration fails when building independant libs
like vdpau, vaapi or omx

Fixes: 1ac40173c2a ("configure.ac: simplify EGL requirements for
drivers dependent on EGL")

Signed-off-by: Leo Liu <leo.liu@amd.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Cc: "17.0" <mesa-stable@lists.freedesktop.org>

glx: add GLXdispatchIndex sort check

Signed-off-by: Eric Engestrom <eric.engestrom@imgtec.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>

docs: update features.txt for GL_ARB_clear_texture with llvmpipe and softpipe

Signed-off-by: Lars Hamre <chemecse@gmail.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Reviewed-by: Edward O'Callaghan <funfunctor@folklore1984.net>

softpipe: enable clear_texture with util_clear_texture

Passes all corresponding piglit tests.

Signed-off-by: Lars Hamre <chemecse@gmail.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Reviewed-by: Edward O'Callaghan <funfunctor@folklore1984.net>

llvmpipe: enable clear_texture with util_clear_texture

Passes all corresponding piglit tests.

Signed-off-by: Lars Hamre <chemecse@gmail.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Reviewed-by: Edward O'Callaghan <funfunctor@folklore1984.net>

gallium: implement util_clear_texture

v3: have util_clear_texture mirror the pipe function (Roland Scheidegger)
v2: rework util clear functions such that they operate on a resource
    instead of a surface (Roland Scheidegger)

Creates a util_clear_texture function for implementing the GL_ARB_clear_texture
in softpipe and llvmpipe.

Signed-off-by: Lars Hamre <chemecse@gmail.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Reviewed-by: Edward O'Callaghan <funfunctor@folklore1984.net>

haiku/winsys: fix dt prototype args

haiku: build fixes around debug defines

radv: add sample mask output support

This adds support to write to sample mask from the fragment shader.

We can optimise this later like radeonsi.

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

radv/ac: refactor our fmask sample index fixup.

This refactors out the sample index fixup between
txf and image load.

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

radv: fetch sample index via fmask for image coord as well.

This follows the txf_ms code, I can't figure out why amdgpu-pro
doesn't do this in their shaders, they must know someone we don't.

This fixes:
dEQP-VK.pipeline.multisample_shader_builtin.sample_id.*

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

radv: add sample mask input support

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

radv: enable location at sample when persample is forced.

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

radv: fix interpolation at wrong place for offset interp

The code was interpolating at the offset from the sample,
not the offset from the center. Also fix for persample interpolation
modes we should force the pixel center to be at the sample.

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

swr: fix index buffers with non-zero indices

Fix issue with index buffers that do not contain a 0 index.  0 index
can be a non-valid index if the (copied) vertex buffers are a subset of the
user's (which happens because we only copy the range between min & max).
Core will use an index passed in from the driver to replace invalid indices.

Only do this for calls that contain non-zero indices, to minimize performance

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

swr: add fetch shader cache

For now, the cache key is all of FETCH_COMPILE_STATE.

Use new/delete for swr_vertex_element_state, since we have to call the
constructors/destructors of the struct elements.

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

st/mesa: free shader cache buffer on fallback

Reviewed-by: Edward O'Callaghan <funfunctor@folklore1984.net>
Tested-by: Michel Dänzer <michel.daenzer@amd.com>

st/mesa: fix crash in shader cache cased by race condition

If a thread doesn't load GLSL IR from cache but does load TGSI
from cache (that was created by another thread) than it will
crash due to expecting gl_program_parameter_list to have been
restored from the GLSL IR cache and not be null.

Reviewed-by: Edward O'Callaghan <funfunctor@folklore1984.net>
Tested-by: Michel Dänzer <michel.daenzer@amd.com>

anv: Enable MSAA compression

This just enables basic MSAA compression (no fast clears) for all
multisampled surfaces.  This improves the framerate of the Sascha
"multisampling" demo by 76% on my Sky Lake laptop.  Running Talos on
medium settings with 8x MSAA, this improves the framerate in the
benchmark by 80%.

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Chad Versace <chadversary@chromium.org>

anv/blorp/clear_subpass: Only set surface clear color for fast clears

Not all clear colors are valid.  In particular, on Broadwell and
earlier, only 0/1 colors are allowed in surface state.  No CTS tests are
affected outright by this because, apparently, the CTS coverage for
different clear colors is pretty terrible.  However, when multisample
compression is enabled, we do hit it with CTS tests and this commit
prevents regressions when enabling MCS on Broadwell and earlier.

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Cc: "13.0 17.0" <mesa-stable@lists.freedesktop.org>

intel/isl: Apply render target alignment constraints for MCS

v2: Instead of having the same block in isl_gen7,8,9.c add it
    once into isl.c::isl_choose_image_alignment_el() instead.

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Signed-off-by: Topi Pohjolainen <topi.pohjolainen@intel.com>

intel/isl: add MCS width constraint 16 samples

v3 (Jason Ekstrand): Add a comment explaining why

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Chad Versace <chadversary@chromium.org>

intel/isl: Return surface creation success from aux helpers

The isl_surf_init call that each of these helpers make can, in theory,
fail.  We should propagate that up to the caller rather than just
silently ignoring it.

Reviewed-by: Topi Pohjolainen <topi.pohjolainen@intel.com>
Reviewed-by: Chad Versace <chadversary@chromium.org>

glsl: Raise a link error for non-SSO ES programs with a TES but no TCS.

OpenGL allows the TCS to be missing and supplies an implicit passthrough
shader, but OpenGL ES does not (see section 7.3 of the ES 3.2 spec,
cited above in the code).

One open question is how to handle this for ARB_ES3_2_compatibility.
This patch raises the link error for all ES shading language programs,
but it might make sense to base it on the API.  The approach taken in
this patch is more restrictive, but should still allow any valid ES
programs to work in GL.

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Andres Gomez <agomez@igalia.com>

isl/state: fix assert on raw buffer surface state minimum size

From IVB PRM, SURFACE_STATE::Height:

"For typed buffer and structured buffer surfaces, the number of
 entries in the buffer ranges from 1 to 2^27 . For raw buffer
 surfaces, the number of entries in the buffer is the number of bytes
 which can range from 1 to 2^30."

The minimum value is 1, according to the spec. The spec quote
was already added into the code by 028f6d8317f00.

Fixes crashing tests under:

dEQP-VK.robustness.buffer_access.*

Signed-off-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>

glsl: enable early_fragment_tests implicitly with post_depth_coverage

From ARB_post_depth_coverage:

   "This extension allows the fragment shader to control whether values in
    gl_SampleMaskIn[] reflect the coverage after application of the early
    depth and stencil tests.  This feature can be enabled with the following
    layout qualifier in the fragment shader:

       layout(post_depth_coverage) in;

    Use of this feature implicitly enables early fragment tests."

And a bit later it also adds:

   "early_fragment_tests" requests that fragment tests be performed before
    fragment shader execution, as described in section 15.2.4 "Early Fragment
    Tests" of the OpenGL Specification. If neither this nor post_depth_coverage
    are declared, per-fragment tests will be performed after fragment shader
    execution."

Fixes:
GL45-CTS.post_depth_coverage_tests.PostDepthSampleMask

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

glsl: refactor get_variable_being_redeclared() to return always an ir_variable pointer

It will return the current variable ('var') or the earlier declaration ('earlier') in
case of redeclaration of that variable.

In order to distinguish between both, 'is_redeclaration' boolean will indicate in which
case we are.

Signed-off-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>

glsl: fix heap-use-after-free in ast_declarator_list::hir()

The get_variable_being_redeclared() function can free 'var' because
a re-declaration of an unsized array variable can establish the size, so
we set the array type to the 'earlier' declaration and free 'var' as it is
not needed anymore.

However, the same 'var' is referenced later in ast_declarator_list::hir().

This patch fixes it by picking the ir_variable_mode from the proper
ir_variable.

This error was detected by Address Sanitizer.

Signed-off-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com>
Suggested-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=99677
Cc: "17.0" <mesa-stable@lists.freedesktop.org>
Cc: "13.0" <mesa-stable@lists.freedesktop.org>

st/wgl: flush with ST_FLUSH_WAIT before releasing shared contexts

Before releasing a shared context, flush the context
with ST_FLUSH_WAIT to make sure all commands are executed.
This ensures that rendering to any shared resources is completed
before they will be referenced by another context.

Fixes an intermittent flickering with Photoshop. (VMware bug# 1779340)

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

st: add ST_FLUSH_WAIT to st_context_flush()

When st_context_flush() is called with ST_FLUSH_WAIT,
the function will return after the fence is completed.

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

radv/ac: handle gs->copy shader clip distances.

This fixes up the clip distance passing between the geometry
shader and the copy shader. It packs the clip and cull distances
into one or two consecutive slots, and avoids wasting space and
make sure the gs output and copy shader input agree on where
things are stored.

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

radv/ac: pass clips properly from vertex->geometry shader stages.

This works out the geometry shader clip/cull inputs separately
to the outputs, and uses that information to read from the ES->GS
ring buffer. It stores the clip/cull distances packed into one
or two slots. It fixes the es output emission and gs input
reading to match.

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

radv/ac: rename num clips/cull to output clips/culls

As geom shaders can have different ones on entry and exit.

also move to uint8_t as these are never that big.

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

vulkan/wsi: move image count to shared structure.

For prime support I need to access this, so move it in advance.

[airlied: fix int->uint32_t]

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>

radeon: fix r600 builds when old version of llvm is present

Reviewed-by: Edward O'Callaghan <funfunctor@folklore1984.net>

vulkan: Fix gen_enum_to_str in out of tree builds

In some configurations the util directory is created when building out
of tree, but not others. This patch ensures that it's created.

Signed-off-by: Dylan Baker <dylanx.c.baker@intel.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Reviewed-and-Tested-by: Mike Lothian <mike@fireburn.co.uk>

anv/Makefile: Gather all the genX files into one place

While we're here, we also fix the alphabetization of the list of
genx_* files.

Reviewed-by: Emil Velikov <emil.velikov@collabora.com>

r600/radeonsi: enable glsl/tgsi on-disk cache

For gpu generations that use LLVM we create a timestamp string
containing both the LLVM and Mesa build times, otherwise we just
use the Mesa build time.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Edward O'Callaghan <funfunctor@folklore1984.net>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>

st/mesa: get on-disk shader cache

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Edward O'Callaghan <funfunctor@folklore1984.net>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>

ddebug/rbug/trace: add get_disk_shader_cache() to pass-throughs

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

gallium: add get_disk_shader_cache() callback

V2: Provide more detail in callback description and add description to
    screen.rst

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

st/mesa: implement a tgsi on-disk shader cache

Implements a tgsi cache for the OpenGL state tracker.

V2: add support for compute shaders

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

st/mesa: add sha1 field to st program structs

This will be used to share the sha1 computed by the tgsi load
function with the tgsi write function.

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

st/mesa: move set_prog_affected_state_flags() to st_program.c

We want to use this in the new tgsi shader cache so we move it here
and make it available externally.

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

util/disk_cache: fix bug with deleting old cache dirs

If there was more than a single directory in the .cache/mesa dir
then it would only remove one (or none) of the directories.

Apparently Valgrind was also reporting:
Conditional jump or move depends on uninitialised value

Reviewed-by: Edward O'Callaghan <funfunctor@folklore1984.net>

vulkan: Combine wsi and util makefiles

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

vulkan/util: Add generator for enum_to_str functions

This adds a python generator to produce enum_to_str functions for
Vulkan from the vk.xml API description. It supports extensions as well
as core API features, and the generator works with both python2 and
python3.

Signed-off-by: Dylan Baker <dylanx.c.baker@intel.com>
Acked-by: Matt Turner <mattst88@gmail.com>
Acked-by: Jason Ekstrand <jason@jlekstrand.net>

Revert "st/vdpau: Fix multithreading"

This reverts commit f1e5dfbe3c8951a6c8acf41bf5e6c2d090098b2c.

For a detailed discussion see
https://lists.freedesktop.org/archives/mesa-dev/2017-February/145283.html

Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>

vl: u_upload_alloc might fail to allocate buffer in bicubic filter

Signed-off-by: Nayan Deshmukh <nayan26deshmukh@gmail.com>
Signed-off-by: Marek Olšák <marek.olsak@amd.com>

gallium: reorder fields in pipe_draw_info

sizeof(struct pipe_draw_info) = 104 -> 88

Also, vertices_per_patch is switched to ubyte, because it can't be more
than 32.

Seemed-reasonable-to: Roland Scheidegger

gallium/hud: handle a thread switch for API-thread-busy monitoring

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