mesa.git
6 years agoi965: Make intelEmitCopyBlit static
Ian Romanick [Wed, 3 Jan 2018 20:40:00 +0000 (12:40 -0800)]
i965: Make intelEmitCopyBlit static

And rename to emit_copy_blit.

v2: sed --in-place -e 's/color_logic_ops/gl_logicop_mode/g' $(grep -lr
color_logic_ops src/) suggested by Brian.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com> [v1]
6 years agoi915: Use enum color_logic_ops for blits
Ian Romanick [Wed, 3 Jan 2018 20:33:20 +0000 (12:33 -0800)]
i915: Use enum color_logic_ops for blits

v2: sed --in-place -e 's/color_logic_ops/gl_logicop_mode/g' $(grep -lr
color_logic_ops src/) suggested by Brian.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com> [v1]
6 years agoi965: Use enum color_logic_ops for blits
Ian Romanick [Wed, 3 Jan 2018 20:03:58 +0000 (12:03 -0800)]
i965: Use enum color_logic_ops for blits

v2: sed --in-place -e 's/color_logic_ops/gl_logicop_mode/g' $(grep -lr
color_logic_ops src/) suggested by Brian.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com> [v1]
6 years agomesa: Pass the translated color logic op dd_function_table::LogicOpcode
Ian Romanick [Wed, 3 Jan 2018 19:45:58 +0000 (11:45 -0800)]
mesa: Pass the translated color logic op dd_function_table::LogicOpcode

And delete the resulting dead code.  This has only been compile-tested.

v2: sed --in-place -e 's/color_logic_ops/gl_logicop_mode/g' $(grep -lr
color_logic_ops src/) suggested by Brian.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
6 years agost/mesa: Use the translated color logic op from the context
Ian Romanick [Wed, 3 Jan 2018 19:44:17 +0000 (11:44 -0800)]
st/mesa: Use the translated color logic op from the context

And delete the resulting dead code.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
6 years agoi965: Use the translated color logic op from the context
Ian Romanick [Wed, 3 Jan 2018 19:43:10 +0000 (11:43 -0800)]
i965: Use the translated color logic op from the context

And delete the resulting dead code.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com>
6 years agomesa: Also track a remapped version of the color logic op
Ian Romanick [Wed, 3 Jan 2018 19:26:41 +0000 (11:26 -0800)]
mesa: Also track a remapped version of the color logic op

With the exception of NVIDIA hardware, these are is the values that all
hardware and Gallium want.  The remapping is currently implemented in at
least 6 places.  This starts the process of consolidating to a single
place.

v2: sed --in-place -e 's/color_logic_ops/gl_logicop_mode/g' $(grep -lr
color_logic_ops src/) suggested by Brian.  Added some comments about the
selection of bit patterns for gl_logicop_mode and the GLenums.
Suggested by Nicolai.  Folded the GLenum_to_color_logicop macro into its
only users.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Brian Paul <brianp@vmware.com> [v1]
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
6 years agoradeonsi: Export signalled sync file instead of -1.
Bas Nieuwenhuizen [Thu, 25 Jan 2018 15:48:17 +0000 (16:48 +0100)]
radeonsi: Export signalled sync file instead of -1.

-1 is considered an error for EGL_ANDROID_native_fence_sync, so
we need to actually create a sync file.

Fixes: f536f45250 "radeonsi: implement sync_file import/export"
Reviewed-by: Dave Airlie <airlied@redhat.com>
6 years agoi965/fs: Reset the register file to VGRF in lower_integer_multiplication
Jason Ekstrand [Fri, 15 Dec 2017 06:10:10 +0000 (22:10 -0800)]
i965/fs: Reset the register file to VGRF in lower_integer_multiplication

18fde36ced4279f2577097a1a7d31b55f2f5f141 changed the way temporary
registers were allocated in lower_integer_multiplication so that we
allocate regs_written(inst) space and keep the stride of the original
destination register.  This was to ensure that any MUL which originally
followed the CHV/BXT integer multiply regioning restrictions would
continue to follow those restrictions even after lowering.  This works
fine except that I forgot to reset the register file to VGRF so, even
though they were assigned a number from alloc.allocate(), they had the
wrong register file.  This caused some GLES 3.0 CTS tests to start
failing on Sandy Bridge due to attempted reads from the MRF:

    ES3-CTS.functional.shaders.precision.int.highp_mul_fragment.snbm64
    ES3-CTS.functional.shaders.precision.int.mediump_mul_fragment.snbm64
    ES3-CTS.functional.shaders.precision.int.lowp_mul_fragment.snbm64
    ES3-CTS.functional.shaders.precision.uint.highp_mul_fragment.snbm64
    ES3-CTS.functional.shaders.precision.uint.mediump_mul_fragment.snbm64
    ES3-CTS.functional.shaders.precision.uint.lowp_mul_fragment.snbm64

This commit remedies this problem by, instead of copying inst->dst and
overwriting nr, just make a new register and set the region to match
inst->dst.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=103626
Fixes: 18fde36ced4279f2577097a1a7d31b55f2f5f141
Cc: "17.3" <mesa-stable@lists.freedesktop.org>
Reviewed-by: Matt Turner <mattst88@gmail.com>
6 years agovulkan: Update the XML and headers to 1.0.68
Jason Ekstrand [Thu, 25 Jan 2018 03:01:47 +0000 (19:01 -0800)]
vulkan: Update the XML and headers to 1.0.68

Acked-by: Dave Airlie <airlied@redhat.com>
Acked-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Acked-by: Chad Versace <chadversary@chromium.org>
6 years agoradv: don't enable tc compat for d32s8 + 4/8 samples (v1.1)
Dave Airlie [Wed, 24 Jan 2018 23:29:55 +0000 (09:29 +1000)]
radv: don't enable tc compat for d32s8 + 4/8 samples (v1.1)

This seems to be broken, at least the cts tests fail.

This fixes:
dEQP-VK.renderpass.suballocation.multisample.d32_sfloat_s8_uint.samples_4
dEQP-VK.renderpass.suballocation.multisample.d32_sfloat_s8_uint.samples_8

2 samples seems to pass fine, amdvlk doesn't appear to enable TC for
possibly some other reasons here.

This is most likely a hack.

v1.1: add a bit of explaination text. (Samuel)
Fixes: ad3d98da9 (radv: enable tc compatible htile for d32s8 also.)
Signed-off-by: Dave Airlie <airlied@redhat.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
6 years agoconfigure.ac: add missing llvm dependencies to .pc files
Chuck Atkins [Tue, 23 Jan 2018 13:52:46 +0000 (08:52 -0500)]
configure.ac: add missing llvm dependencies to .pc files

v2: Only add as dependencies for gallium-osmesa and gallium-xlib

CC: <mesa-stable@lists.freedesktop.org>
Signed-of-by: Chuck Atkins <chuck.atkins@kitware.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
6 years agoswr/rast: Optimize DumpToFile output size
George Kyriazis [Wed, 24 Jan 2018 20:15:02 +0000 (14:15 -0600)]
swr/rast: Optimize DumpToFile output size

Modify DumpToFile to only dump the function, not the entire module.
Reduces file sizes and speeds up the dumping.

Reviewed-by: Bruce Cherniak <bruce.cherniak@intel.com>
6 years agoswr/rast: Updated copyright dates
George Kyriazis [Wed, 24 Jan 2018 01:29:05 +0000 (19:29 -0600)]
swr/rast: Updated copyright dates

on knob-related files.

Reviewed-by: Bruce Cherniak <bruce.cherniak@intel.com>
6 years agoswr/rast: Move memory-related JIT functions
George Kyriazis [Tue, 23 Jan 2018 01:23:32 +0000 (19:23 -0600)]
swr/rast: Move memory-related JIT functions

Move them to their own file (builder_mem.{h|cpp}).  Add builder_mem.cpp
to the build system.

Reviewed-by: Bruce Cherniak <bruce.cherniak@intel.com>
6 years agoswr/rast: Add extra (optional) parameter in GATHERPS
George Kyriazis [Tue, 23 Jan 2018 07:51:00 +0000 (01:51 -0600)]
swr/rast: Add extra (optional) parameter in GATHERPS

Now also takes in an additional parameter (draw context) for future
expansion.

Reviewed-by: Bruce Cherniak <bruce.cherniak@intel.com>
6 years agoswr/rast: Better ExecCmd (i.e. system()) implmentation
George Kyriazis [Sat, 20 Jan 2018 00:03:16 +0000 (18:03 -0600)]
swr/rast: Better ExecCmd (i.e. system()) implmentation

Hides console window creation during JIT linker execution in apps that
don't have a console.  Remove hooking of CreateProcessInternalA - the
MSFT implementation just turns around and calls CreateProcessInternalW
which, we do hook.

Reviewed-by: Bruce Cherniak <bruce.cherniak@intel.com>
6 years agoswr/rast: Support USE_SIMD16_FRONTEND=0 for EarlyRast
George Kyriazis [Fri, 19 Jan 2018 23:58:33 +0000 (17:58 -0600)]
swr/rast: Support USE_SIMD16_FRONTEND=0 for EarlyRast

Early Rasterization did not initially work with USE_SIMD16_FRONTEND=0.
Fix it so it works there, too.  Please note that the default setting
is USE_SIMD16_FRONTEND=1.

Reviewed-by: Bruce Cherniak <bruce.cherniak@intel.com>
6 years agomesa: whitespace fixes in attrib.c
Brian Paul [Thu, 25 Jan 2018 16:56:06 +0000 (09:56 -0700)]
mesa: whitespace fixes in attrib.c

Trivial.

6 years agomesa: whitespace fixes in varray.h
Brian Paul [Thu, 25 Jan 2018 16:19:10 +0000 (09:19 -0700)]
mesa: whitespace fixes in varray.h

Reviewed-by: Charmaine Lee <charmainel@vmware.com>
6 years agomesa: include mtypes.h in varray.h
Brian Paul [Thu, 25 Jan 2018 16:16:54 +0000 (09:16 -0700)]
mesa: include mtypes.h in varray.h

We actually use some of the types from mtypes.h so include it directly
instead of relying on indirectly including it via bufferobj.h

Reviewed-by: Charmaine Lee <charmainel@vmware.com>
6 years agomesa: s/gl_vertex_attrib_array/gl_array_attributes/ in comments
Brian Paul [Thu, 25 Jan 2018 16:12:59 +0000 (09:12 -0700)]
mesa: s/gl_vertex_attrib_array/gl_array_attributes/ in comments

The structure type was renamed some time ago, but some comments
were not updated.

Reviewed-by: Charmaine Lee <charmainel@vmware.com>
6 years agomesa: simplify _mesa_delete_list() a bit, add some assertions
Brian Paul [Sat, 20 Jan 2018 03:57:24 +0000 (20:57 -0700)]
mesa: simplify _mesa_delete_list() a bit, add some assertions

All but two cases of the switch did the same n += InstSize[n[0].opcode]
instruction.  Just move it after the switch.

Add some sanity check assertions.

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
6 years agost/mesa: expand glDrawPixels cache to handle multiple images
Brian Paul [Fri, 12 Jan 2018 16:32:41 +0000 (09:32 -0700)]
st/mesa: expand glDrawPixels cache to handle multiple images

The newest version of WSI Fusion makes several glDrawPixels calls
per frame.  By caching more than one image, we get better performance
when panning/zooming the map.

v2: move pixel unpack param checking out of cache search loop, per Roland
v3: also move unpack->BufferObj check out of loop, per Roland.

6 years agost/mesa: add some debug code in st_choose_format()
Brian Paul [Thu, 11 Jan 2018 18:53:02 +0000 (11:53 -0700)]
st/mesa: add some debug code in st_choose_format()

To aid in debugging gallium surface format selection issues.

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
6 years agosvga: s/Bool/SVGA3dBool/ in SVGA3dDevCapResult
Brian Paul [Thu, 25 Jan 2018 18:36:53 +0000 (11:36 -0700)]
svga: s/Bool/SVGA3dBool/ in SVGA3dDevCapResult

And fix whitespace.  To sync up with in-house code.

Reviewed-by: Charmaine Lee <charmainel@vmware.com>
6 years agoconfigure.ac: correct driglx-direct help text
Emil Velikov [Wed, 20 Dec 2017 17:34:55 +0000 (17:34 +0000)]
configure.ac: correct driglx-direct help text

The default was toggled a while back, but the text wasn't updated.

Fixes: bd526ec9e1b ("configure: Always default to
--enable-driglx-direct")
Cc: Jon TURNEY <jon.turney@dronecode.org.uk>
Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
6 years agoswrast: remove non-applicable GLX_SWAP_COPY_OML comment
Emil Velikov [Thu, 7 Dec 2017 17:07:39 +0000 (17:07 +0000)]
swrast: remove non-applicable GLX_SWAP_COPY_OML comment

Noticed while skimming for GLX_ instances in the dri codebase.
Comment is completely off and was in such a state since day 1.

Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
6 years agomapi: remove duplicate GL typedefs
Emil Velikov [Thu, 7 Dec 2017 17:07:38 +0000 (17:07 +0000)]
mapi: remove duplicate GL typedefs

Remove the instances already available in gl.h or glext.h.
Sadly GLclampx is only available in GLES(1) so we need to keep that one.

Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
6 years agomapi: remove non applicable HAVE_DIX_CONFIG_H hunk
Emil Velikov [Thu, 7 Dec 2017 17:07:37 +0000 (17:07 +0000)]
mapi: remove non applicable HAVE_DIX_CONFIG_H hunk

Seeming artefact from when the xserver build was diving directly into
mesa's tree.

Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
6 years agomapi: autotools: remove unused MAPI_FILES file list
Emil Velikov [Thu, 7 Dec 2017 17:07:36 +0000 (17:07 +0000)]
mapi: autotools: remove unused MAPI_FILES file list

The sole user was OpenVG, which was removed couple of years ago.

Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
6 years agoautomake: st/mesa/tests: add st_tests_common.h to the tarball
Emil Velikov [Thu, 25 Jan 2018 15:31:10 +0000 (15:31 +0000)]
automake: st/mesa/tests: add st_tests_common.h to the tarball

Fixes: 6569b33b6ee ("mesa/st/tests: unify MockCodeLine* classes")
Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
6 years agoautomake: mesa: include vbo_private.h in the tarball
Emil Velikov [Thu, 25 Jan 2018 15:27:20 +0000 (15:27 +0000)]
automake: mesa: include vbo_private.h in the tarball

Fixes: a7cfec3be0e ("vbo: move VBO-private types, prototypes, etc. into
new vbo_private.h header")
Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
6 years agoautomake: small cleanup after the meson.build inclusion
Emil Velikov [Tue, 23 Jan 2018 13:04:57 +0000 (13:04 +0000)]
automake: small cleanup after the meson.build inclusion

Namely extend the EXTRA_DIST list, instead of re-assigning it and bring
back a file dropped by mistake.

Fixes: 436ed65d38d ("autotools: include meson build files in tarball")
Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
6 years agoautomake: anv: ship anv_extensions_gen.py in the tarball
Emil Velikov [Tue, 23 Jan 2018 13:02:50 +0000 (13:02 +0000)]
automake: anv: ship anv_extensions_gen.py in the tarball

Fixes: dd088d4bec7 ("anv/extensions: Generate a header file with
extension tables")
Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
6 years agoautomake: vc5: remove non-applicable v3dx_simulator.h
Emil Velikov [Tue, 23 Jan 2018 11:52:49 +0000 (11:52 +0000)]
automake: vc5: remove non-applicable v3dx_simulator.h

Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
6 years agogallivm: fix crash with seamless cube filtering with different min/mag filter
Roland Scheidegger [Thu, 25 Jan 2018 03:30:41 +0000 (04:30 +0100)]
gallivm: fix crash with seamless cube filtering with different min/mag filter

We are not allowed to modify the incoming coords values, or things may
crash (as we may be inside a llvm conditional and the values may be used
in another branch).
I recently broke this when fixing an issue with NaNs and seamless cube
map filtering, and it causes crashes when doing cubemap filtering
if the min and mag filters are different.
Add const to the pointers passed in to prevent this mishap in the future.

Fixes: a485ad0bcd ("gallivm: fix an issue with NaNs with seamless cube filtering")
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
6 years agoegl: keep extension list sorted, per comment at the top
Eric Engestrom [Mon, 8 Jan 2018 15:47:09 +0000 (15:47 +0000)]
egl: keep extension list sorted, per comment at the top

Signed-off-by: Eric Engestrom <eric.engestrom@imgtec.com>
Reviewed-by: Adam Jackson <ajax@redhat.com>
6 years agoswr/rast: support llvm 3.9 type declarations
George Kyriazis [Tue, 23 Jan 2018 22:12:42 +0000 (16:12 -0600)]
swr/rast: support llvm 3.9 type declarations

LLVM 3.9 was not taken into account in initial check-in.

Fixes: 01ab218bbc ("swr/rast: Initial work for debugging support.")
cc: mesa-stable@lists.freedesktop.org
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=104749
Acked-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Bruce Cherniak <bruce.cherniak@intel.com>
6 years agoac/nir: add break statements in needs_view_index_sgpr()
Samuel Pitoiset [Thu, 25 Jan 2018 09:49:23 +0000 (10:49 +0100)]
ac/nir: add break statements in needs_view_index_sgpr()

Previous code is correct but as the first case statement uses
a break, keep it consistent.

CID: 1428579
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
6 years agoloader: let compiler figure out the length of the string
Eric Engestrom [Tue, 19 Dec 2017 14:12:44 +0000 (14:12 +0000)]
loader: let compiler figure out the length of the string

Basically, turn comment into code

Signed-off-by: Eric Engestrom <eric.engestrom@imgtec.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
6 years agomeson: simplify dri3 logic
Eric Engestrom [Thu, 7 Dec 2017 16:04:14 +0000 (16:04 +0000)]
meson: simplify dri3 logic

Signed-off-by: Eric Engestrom <eric.engestrom@imgtec.com>
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>
6 years agomesa: add missing RGB9_E5 format in _mesa_base_fbo_format
Juan A. Suarez Romero [Mon, 15 Jan 2018 10:58:50 +0000 (10:58 +0000)]
mesa: add missing RGB9_E5 format in _mesa_base_fbo_format

This fixes KHR-GL45.internalformat.renderbuffer.rgb9_e5.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
6 years agoi965: Stop disabling aux during texture preparation
Jason Ekstrand [Wed, 24 Jan 2018 19:51:08 +0000 (11:51 -0800)]
i965: Stop disabling aux during texture preparation

Previously, we were handling self-dependencies by marking the render
buffer and then passing disable_aux=true to prepare_texture so that it
would do a resolve.  This works but ends us up doing to much resolving
in some cases.  Specifically, if we're doing something such as mipmap
generation, this would cause us to resolve all levels of the texture if
even one of them is overlapping.

Instead, this commit makes us wait until we process the framebuffer to
do these resolves and we only resolve the slices needed for rendering.
Doing this resolve puts them into the pass-through state so, even if we
do texture using CCS_E, the CCS data will effectively be ignored and the
real surface contents read.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
6 years agoi965/draw: Set NEW_AUX_STATE when draw aux changes
Jason Ekstrand [Wed, 24 Jan 2018 07:47:26 +0000 (23:47 -0800)]
i965/draw: Set NEW_AUX_STATE when draw aux changes

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=104411
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=104383
Fixes: ea0d2e98ecb369ab84e78c84709c0930ea8c293a
Cc: mesa-stable@lists.freedesktop.org
Reviewed-by: Topi Pohjolainen <topi.pohjolainen@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
6 years agoi965: Replace draw_aux_buffer_disabled with draw_aux_usage
Jason Ekstrand [Wed, 24 Jan 2018 07:13:10 +0000 (23:13 -0800)]
i965: Replace draw_aux_buffer_disabled with draw_aux_usage

Instead of keeping an array of booleans, we now hang onto an array of
isl_aux_usage enums.  This means that the thing we are passing from
brw_draw.c to surface state setup is the thing that surface state setup
actually needs instead of an input to compute what it needs.

Cc: mesa-stable@lists.freedesktop.org
Reviewed-by: Topi Pohjolainen <topi.pohjolainen@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
6 years agoi965/surface_state: Drop brw_aux_surface_disabled
Jason Ekstrand [Wed, 24 Jan 2018 19:43:55 +0000 (11:43 -0800)]
i965/surface_state: Drop brw_aux_surface_disabled

The only purpose of this function is to disable aux on texture surfaces
when the corresponding renderbuffer has aux disabled.  However, the act
of disabling aux on the renderbuffer will cause it to be resolved and
intel_miptree_texture_aux_usage will already check the resolved status
of a texture and return ISL_AUX_USAGE_NONE for it.  Even if we used CCS
for it, that wouldn't really be a problem because the CCS will be in the
pass-through state and so it would effectively be ignored.

Cc: mesa-stable@lists.freedesktop.org
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
6 years agoi965/miptree: Add an aux_disabled parameter to render_aux_usage
Jason Ekstrand [Tue, 23 Jan 2018 07:40:48 +0000 (23:40 -0800)]
i965/miptree: Add an aux_disabled parameter to render_aux_usage

Only one of the callers of intel_miptree_render_aux_usage actually took
brw->draw_aux_buffer_disabled into account.  This was causing us to
ignore draw_aux_buffer_disabled for the intel_miptree_prepare_render.
This isn't a problem because the draw_aux_buffer_disabled entry was set
during texture preparation and we already did the resolve at that time.
However, this also meant that the aux_usage we were passing to
brw_cache_flush_for_render and brw_render_cache_add_bo was wrong so our
automatic cache flushing around aux_usage changes wasn't happening.
This was causing GPU hangs in Oxenfree.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=104711
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=104411
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=104383
Fixes: ea0d2e98ecb369ab84e78c84709c0930ea8c293a
Cc: mesa-stable@lists.freedesktop.org
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
6 years agoi965/miptree: Take an aux_usage in prepare/finish_render
Jason Ekstrand [Tue, 23 Jan 2018 07:33:53 +0000 (23:33 -0800)]
i965/miptree: Take an aux_usage in prepare/finish_render

Both callers of intel_miptree_prepare/finish_render have to call
intel_miptree_render_aux_usage anyway for other reasons.  They may as
well pass the result in instead of us calling it again.

Cc: mesa-stable@lists.freedesktop.org
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
6 years agoaubinator: Multiply count by 4 to compute buffer sizes
Jason Ekstrand [Tue, 23 Jan 2018 06:45:14 +0000 (22:45 -0800)]
aubinator: Multiply count by 4 to compute buffer sizes

The count field is in terms of dwords and not bytes.

6 years agost/glsl_to_nir: remove reallocation of sampler/image location
Timothy Arceri [Tue, 23 Jan 2018 21:48:49 +0000 (08:48 +1100)]
st/glsl_to_nir: remove reallocation of sampler/image location

As far as I can tell this always just reassigns the same value.

Also as we don't curretly store UniformHash in the shader cache
removing this will help with adding a shader cache to gallium
nir drivers.

Reviewed-by: Rob Clark <robdclark@gmail.com>
6 years agodocs: add 18.1.0-devel release notes template
Jordan Justen [Thu, 25 Jan 2018 01:07:30 +0000 (17:07 -0800)]
docs: add 18.1.0-devel release notes template

Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
6 years agomesa: bump version to 18.1.0-devel
Jordan Justen [Thu, 25 Jan 2018 01:02:55 +0000 (17:02 -0800)]
mesa: bump version to 18.1.0-devel

Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
6 years agomeson: handle LLVM 'x.x.xgit-revision' versions
Greg V [Wed, 24 Jan 2018 18:02:43 +0000 (21:02 +0300)]
meson: handle LLVM 'x.x.xgit-revision' versions

When LLVM is built inside of a git repo (even way below, e.g. /usr/ports/.git
exists, and LLVM is built in /usr/ports/devel/llvm50/work), its version
becomes something like 5.0.0git-f8ab206b2176.

New meson versions already handle this, but we support older versions too.

Fixes: 673dda8330769 ("meson: build "radv" vulkan driver for radeon hardware")
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>
6 years agomeson: fix getting cflags from pkg-config
Greg V [Wed, 24 Jan 2018 18:02:42 +0000 (21:02 +0300)]
meson: fix getting cflags from pkg-config

get_pkgconfig_variable('cflags') always returns an empty list, it's a
function for getting *custom* variables.

Meson does not yet support asking for cflags, so explicitly invoke
pkg-config for now.

Fixes: 68076b87474e ("meson: build gallium vdpau state tracker")
Fixes: a817af8a89eb ("meson: build gallium xvmc state tracker")
Fixes: 1d36dc674d52 ("meson: build gallium omx state tracker")
Fixes: 5a785d51a6d6 ("meson: build gallium va state tracker")
Reviewed-by: Dylan Baker <dylan.c.baker@intel.com>
6 years agomeson: fix BSD build
Greg V [Wed, 24 Jan 2018 18:02:40 +0000 (21:02 +0300)]
meson: fix BSD build

CC: 18.0 <mesa-stable@lists.freedesktop.org>
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>
Reviewed-by: Eric Engestrom <eric.engestrom@imgtec.com>
6 years agomeson: fix missing dependencies
Greg V [Wed, 24 Jan 2018 18:02:39 +0000 (21:02 +0300)]
meson: fix missing dependencies

Fixes: 66f97f6640f5 ("meson: build radeonsi")
Reviewed-by: Emil Velikov <emil.velikov@colalbora.com>
Reviewed-by: Dylan Baker <dylan.c.baker@intel.com>
6 years agoanv: correct a duplicate check in an assert
Grazvydas Ignotas [Mon, 22 Jan 2018 22:44:36 +0000 (00:44 +0200)]
anv: correct a duplicate check in an assert

Looks like checking both sources was intended, instead of the first one
twice. Found with Coccinelle, coccinellery/xand/xand.cocci semantic patch.

Signed-off-by: Grazvydas Ignotas <notasas@gmail.com>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
6 years agomeson: fix HAVE_LLVM version define in meson build
Marc Dietrich [Wed, 24 Jan 2018 21:03:51 +0000 (22:03 +0100)]
meson: fix HAVE_LLVM version define in meson build

LLVM patch level is not included in HAVE_LLVM.

Fixes: e6418ab1566d ("meson: build "radv" vulkan driver for radeon hardware")
Reviewed-by: Eric Engestrom <eric.engestrom@imgtec.com>
Reviewed-by: Dylan Baker <dylan.c.baker@intel.com>
Signed-off-by: Marc Dietrich <marvin24@gmx.de>
6 years agomeson: correctly set SYSCONFDIR for loading dirrc
Dylan Baker [Tue, 23 Jan 2018 18:28:08 +0000 (10:28 -0800)]
meson: correctly set SYSCONFDIR for loading dirrc

Fixes: d1992255bb29 ("meson: Add build Intel "anv" vulkan driver")
Reported-by: Marc Dietrich <marvin24@gmx.de>
Signed-off-by: Dylan Baker <dylan.c.baker@intel.com>
Reviewed-by: Eric Engestrom <eric.engestrom@imgtec.com>
6 years agoradv: add multisample Z optimisation from amdvlk
Dave Airlie [Wed, 24 Jan 2018 03:13:49 +0000 (13:13 +1000)]
radv: add multisample Z optimisation from amdvlk

This was just found while reading for other stuff,
src/core/hw/gfxip/gfx6/gfx6DepthStencilView.cpp.

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
6 years agoradv: move spi_baryc_cntl to pipeline
Dave Airlie [Wed, 24 Jan 2018 02:53:26 +0000 (12:53 +1000)]
radv: move spi_baryc_cntl to pipeline

We need to enable the pos float location 2 mode anytime we have
persample not just when forced by the frag shader.

This fixes:
dEQP-VK.pipeline.multisample.min_sample_shading*

Fixes: 58c97a079 (radv: enable location at sample when persample is forced.)
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
6 years agogallium/u_tests: add texture_barrier and FBFETCH tests
Marek Olšák [Wed, 10 Jan 2018 12:35:39 +0000 (13:35 +0100)]
gallium/u_tests: add texture_barrier and FBFETCH tests

Tested-by: Dieter Nützel <Dieter@nuetzel-hh.de>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
6 years agoradeonsi: don't ignore pitch for imported textures
Marek Olšák [Wed, 10 Jan 2018 12:37:08 +0000 (13:37 +0100)]
radeonsi: don't ignore pitch for imported textures

Cc: 17.2 17.3 <mesa-stable@lists.freedesktop.org>
Tested-by: Dieter Nützel <Dieter@nuetzel-hh.de>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
6 years agomeson: Fix define for USE_SSE41
Scott D Phillips [Wed, 24 Jan 2018 19:24:12 +0000 (11:24 -0800)]
meson: Fix define for USE_SSE41

Before we were adding -DHAVE_SSE41 which isn't what the code is
looking for, so some uses of the sse4.1 code were always being
skipped.

v2: Don't add any compile check for the quite old -msse4.1 option (Dylan)

Fixes: 84486f6462 ("meson: Enable SSE4.1 optimizations")
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>
6 years agomesa/st/glsl_to_tgsi: remove now unneeded assert.
Gert Wollny [Tue, 17 Oct 2017 10:36:50 +0000 (12:36 +0200)]
mesa/st/glsl_to_tgsi: remove now unneeded assert.

With the implementation of the tracking of the registers used in reladdr
asserting that a driver calling merge_register() uses the address register
is no longer needed.

Reviewed-by: Brian Paul <brianp@vmware.com>
Signed-off-by: Gert Wollny <gw.fossdev@gmail.com>
6 years agomesa/st/tests: Add tests for lifetime tracking with indirect addressing
Gert Wollny [Thu, 7 Dec 2017 12:57:21 +0000 (13:57 +0100)]
mesa/st/tests: Add tests for lifetime tracking with indirect addressing

Add a code line type that accepts one layer of indirect addressing and
add tests to check that temporary register access used for indirect
addressing is accounted for in the lifetime estimation.

Reviewed-by: Brian Paul <brianp@vmware.com>
Signed-off-by: Gert Wollny <gw.fossdev@gmail.com>
6 years agomesa/st/glsl_to_tgsi: Add tracking of indirect addressing registers
Gert Wollny [Thu, 7 Dec 2017 12:51:22 +0000 (13:51 +0100)]
mesa/st/glsl_to_tgsi: Add tracking of indirect addressing registers

So far indirect addressing was not tracked to estimate the temporary
life time, and it was not needed, because code to load the address
registers was always emitted eliminating the reladdr* handles in the
past glsl-to.tgsi stages. Now, with Mareks patch allowing any 1D register
to be used for addressing on some hardware this changed, and
the tracking becomes necessary.

Because the registers have no direct indication on whether the reladdr* was
already loaded into an address register, the temporaries in reladdr* are
always tracked as reads. This may result in a slight over-estimation of the
lifetime in the cases when the load to the address register was emitted.

v2: no changes
v3: Use debug_log variable instead of directly writing to std::err in debugging
    output.
v6: fix indention and typos

Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com> (v1)
Signed-off-by: Gert Wollny <gw.fossdev@gmail.com>
6 years agomesa/st/tests: Add tests for improved tracking of temporaries
Gert Wollny [Tue, 3 Oct 2017 07:49:37 +0000 (09:49 +0200)]
mesa/st/tests: Add tests for improved tracking of temporaries

Additional tests are added that check the tracking of access to temporaries
in if-else branches.

Reviewed-by: Brian Paul <brianp@vmware.com>
Signed-off-by: Gert Wollny <gw.fossdev@gmail.com>
6 years agomesa/st/glsl_to_tgsi: Add tracking of ifelse writes in register merging
Gert Wollny [Fri, 8 Sep 2017 11:40:09 +0000 (13:40 +0200)]
mesa/st/glsl_to_tgsi: Add tracking of ifelse writes in register merging

Improve the life-time evaluation of temporary registers by also tracking
writes in both if and else branches and in up to 32 nested scopes.
As a result the estimated required register life-times can be further
reduced enabling more registers to be merged.

Reviewed-by: Brian Paul <brianp@vmware.com>
Signed-off-by: Gert Wollny <gw.fossdev@gmail.com>
6 years agomesa/st/tests: cleanup whitespace usage and correct some comments
Gert Wollny [Thu, 12 Oct 2017 15:10:31 +0000 (17:10 +0200)]
mesa/st/tests: cleanup whitespace usage and correct some comments

Reviewed-by: Brian Paul <brianp@vmware.com>
Signed-off-by: Gert Wollny <gw.fossdev@gmail.com>
6 years agomesa/st/tests: unify MockCodeLine* classes
Gert Wollny [Wed, 11 Oct 2017 17:56:46 +0000 (19:56 +0200)]
mesa/st/tests: unify MockCodeLine* classes

 * Merge the classes MockCodeLine and MockCodelineWithSwizzle into
   one, and  refactor tests accordingly.
 * Change memory allocations to use ralloc* interface.

 v2:
 * move the test classes into a conveniance library
 * rename the Mock* classes to Fake* since they are not really
   Mocks
 * Base assertion of correct number of src and dst registers in tests
   on what the operatand actually expects
 * Fix number of destinations in one test

 v6:
 * fix local includes using "..." insteadof <...>

Reviewed-by: Brian Paul <brianp@vmware.com>
Signed-off-by: Gert Wollny <gw.fossdev@gmail.com>
6 years agomesa/st/tests: Fix zero-byte allocation leaks
Gert Wollny [Thu, 14 Sep 2017 10:49:07 +0000 (12:49 +0200)]
mesa/st/tests: Fix zero-byte allocation leaks

Don't allocate a zero-sized array, when no texture offsets are given.

v5: correct spaces and empty lines

Reviewed-by: Brian Paul <brianp@vmware.com>(v4)
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com> (v1)
Signed-off-by: Gert Wollny <gw.fossdev@gmail.com>
6 years agomesa/st/glsl_to_tgsi: Add some operators for glsl_to_tgsi related classes
Gert Wollny [Fri, 1 Dec 2017 14:46:20 +0000 (15:46 +0100)]
mesa/st/glsl_to_tgsi: Add some operators for glsl_to_tgsi related classes

Add the equal operator and the "<<" stream write operator for the
st_*_reg classes and the "<<" operator to the instruction class, and
make use of these operators in the debugging output.

v5: Fix empty lines

Reviewed-by: Brian Paul <brianp@vmware.com> (v4)
Signed-off-by: Gert Wollny <gw.fossdev@gmail.com>
6 years agomesa/program: Add missing file types to printout
Gert Wollny [Mon, 4 Dec 2017 22:01:08 +0000 (23:01 +0100)]
mesa/program: Add missing file types to printout

Reviewed-by: Brian Paul <brianp@vmware.com>
Signed-off-by: Gert Wollny <gw.fossdev@gmail.com>
6 years agovbo: fix incorrect min/max_index values in display list draw call
Brian Paul [Tue, 23 Jan 2018 17:48:51 +0000 (10:48 -0700)]
vbo: fix incorrect min/max_index values in display list draw call

This fixes another regression from commit 8e4efdc895ea ("vbo: optimize
some display list drawing").  The problem was the min_index, max_index
values passed to the vbo drawing function were not computed to compensate
for the biased prim::start values.

https://bugs.freedesktop.org/show_bug.cgi?id=104746
https://bugs.freedesktop.org/show_bug.cgi?id=104742
https://bugs.freedesktop.org/show_bug.cgi?id=104690
Tested-by: Clayton Craft <clayton.a.craft@intel.com>
Fixes: 8e4efdc895ea ("vbo: optimize some display list drawing")
Reviewed-by: Emil Velikov <emil.velikov@collabora.co.uk>
6 years agovbo: whitespace/formatting fixes in vbo_split_inplace.c
Brian Paul [Thu, 18 Jan 2018 23:50:43 +0000 (16:50 -0700)]
vbo: whitespace/formatting fixes in vbo_split_inplace.c

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
6 years agovbo: whitespace/formatting fixes in vbo.h
Brian Paul [Thu, 18 Jan 2018 23:43:43 +0000 (16:43 -0700)]
vbo: whitespace/formatting fixes in vbo.h

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
6 years agovbo/i965: move vbo_all_varyings_in_vbos() to brw_draw.c
Brian Paul [Thu, 18 Jan 2018 23:10:49 +0000 (16:10 -0700)]
vbo/i965: move vbo_all_varyings_in_vbos() to brw_draw.c

It's only used in brw_draw_prims().

s/GLboolean/bool/, etc.

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
6 years agovbo: remove unused vbo_any_varyings_in_vbos() function
Brian Paul [Thu, 18 Jan 2018 21:50:08 +0000 (14:50 -0700)]
vbo: remove unused vbo_any_varyings_in_vbos() function

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
6 years agovbo: remove unneeded #includes
Brian Paul [Thu, 18 Jan 2018 21:47:18 +0000 (14:47 -0700)]
vbo: remove unneeded #includes

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
6 years agovbo: remove vbo_context.h and change includes to use vbo.h instead
Brian Paul [Thu, 18 Jan 2018 21:39:22 +0000 (14:39 -0700)]
vbo: remove vbo_context.h and change includes to use vbo.h instead

Now vbo.h is the public interface to the VBO module.

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
6 years agovbo: move remaining items from vbo_context.h to vbo.h
Brian Paul [Thu, 18 Jan 2018 21:32:38 +0000 (14:32 -0700)]
vbo: move remaining items from vbo_context.h to vbo.h

Non-VBO sources files sometimes included vbo.h while others included
vbo_context.h.  We're moving all public types, functions to the former.

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
6 years agovbo: move VBO-private types, prototypes, etc. into new vbo_private.h header
Brian Paul [Thu, 18 Jan 2018 21:22:29 +0000 (14:22 -0700)]
vbo: move VBO-private types, prototypes, etc. into new vbo_private.h header

Things which should not be used outside the VBO module.
More public/private clean-ups coming.

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
6 years agomesa: use new _vbo_install_exec_vtxfmt() function
Brian Paul [Thu, 18 Jan 2018 21:13:34 +0000 (14:13 -0700)]
mesa: use new _vbo_install_exec_vtxfmt() function

Instead of reaching into the vbo_context object in vtxfmt.c

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
6 years agonouveau: remove vbo_context() call
Brian Paul [Thu, 18 Jan 2018 20:51:06 +0000 (13:51 -0700)]
nouveau: remove vbo_context() call

_vbo_DestroyContext() can be safely called even if there's no VBO
module.  Removes a dependency on the vbo_context() function.

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
6 years agoi965: use vbo_set_[indirect]_draw_func()
Brian Paul [Thu, 18 Jan 2018 20:50:04 +0000 (13:50 -0700)]
i965: use vbo_set_[indirect]_draw_func()

Instead of poking into the vbo_context object.

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
6 years agovbo: move vbo_sizeof_ib_type() into vbo_exec_array.c
Brian Paul [Thu, 18 Jan 2018 20:28:10 +0000 (13:28 -0700)]
vbo: move vbo_sizeof_ib_type() into vbo_exec_array.c

It's only used in this one file.

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
6 years agomesa: move vbo_count_tessellated_primitives() to api_validate.c
Brian Paul [Thu, 18 Jan 2018 20:21:56 +0000 (13:21 -0700)]
mesa: move vbo_count_tessellated_primitives() to api_validate.c

It's only used in this file and has nothing VBO-specific about it.

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
6 years agomesa: update comment on gl_display_list
Brian Paul [Thu, 18 Jan 2018 19:53:12 +0000 (12:53 -0700)]
mesa: update comment on gl_display_list

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
6 years agomesa: whitespace clean-ups in mtypes.h
Brian Paul [Thu, 18 Jan 2018 19:42:53 +0000 (12:42 -0700)]
mesa: whitespace clean-ups in mtypes.h

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
6 years agomesa: remove unused MAT_INDEX_AMBIENT/DIFFUSE/SPECULAR contants
Brian Paul [Thu, 18 Jan 2018 19:36:42 +0000 (12:36 -0700)]
mesa: remove unused MAT_INDEX_AMBIENT/DIFFUSE/SPECULAR contants

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
6 years agovbo: move DLIST_DANGLING_REFS from mtypes.h to vbo_save_api.c
Brian Paul [Thu, 18 Jan 2018 19:22:14 +0000 (12:22 -0700)]
vbo: move DLIST_DANGLING_REFS from mtypes.h to vbo_save_api.c

It's only used in this file.

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
6 years agovbo: replace assert(0) with unreachable()
Brian Paul [Wed, 17 Jan 2018 19:54:04 +0000 (12:54 -0700)]
vbo: replace assert(0) with unreachable()

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
6 years agovbo: fix, add comment in vbo_save.h
Brian Paul [Wed, 17 Jan 2018 22:46:28 +0000 (15:46 -0700)]
vbo: fix, add comment in vbo_save.h

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
6 years agovbo: whitespace, formatting fixes in vbo_split.[ch]
Brian Paul [Wed, 17 Jan 2018 19:20:49 +0000 (12:20 -0700)]
vbo: whitespace, formatting fixes in vbo_split.[ch]

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
6 years agoi965: Don't try to disable render aux buffers for compute
Topi Pohjolainen [Tue, 16 Jan 2018 12:17:00 +0000 (14:17 +0200)]
i965: Don't try to disable render aux buffers for compute

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=104546
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Signed-off-by: Topi Pohjolainen <topi.pohjolainen@intel.com>
6 years agoanv/cmd_buffer: Move gen7 index buffer state to graphics state
Jason Ekstrand [Sat, 16 Dec 2017 00:55:54 +0000 (16:55 -0800)]
anv/cmd_buffer: Move gen7 index buffer state to graphics state

Tested-by: Józef Kucia <joseph.kucia@gmail.com>
Reviewed-by: Topi Pohjolainen <topi.pohjolainen@intel.com>
Cc: "18.0" <mesa-stable@lists.freedesktop.org>
6 years agoanv/cmd_buffer: Move num_workgroups to compute state
Jason Ekstrand [Sat, 16 Dec 2017 00:53:03 +0000 (16:53 -0800)]
anv/cmd_buffer: Move num_workgroups to compute state

While we're here, make it an anv_address.

Tested-by: Józef Kucia <joseph.kucia@gmail.com>
Reviewed-by: Topi Pohjolainen <topi.pohjolainen@intel.com>
Cc: "18.0" <mesa-stable@lists.freedesktop.org>
6 years agoanv/cmd_buffer: Move dynamic state to graphics state
Jason Ekstrand [Sat, 16 Dec 2017 00:48:53 +0000 (16:48 -0800)]
anv/cmd_buffer: Move dynamic state to graphics state

Tested-by: Józef Kucia <joseph.kucia@gmail.com>
Reviewed-by: Topi Pohjolainen <topi.pohjolainen@intel.com>
Cc: "18.0" <mesa-stable@lists.freedesktop.org>