mesa.git
14 months agointel/perf: export performance counters sorted by [group|set] and name
Marcin Ślusarz [Wed, 10 Jun 2020 13:58:59 +0000 (15:58 +0200)]
intel/perf: export performance counters sorted by [group|set] and name

It's a lot easier to deal with them in RenderDoc when they are
in some meaningful order.

Signed-off-by: Marcin Ślusarz <marcin.slusarz@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5788>

14 months agointel/perf: split load_oa_metrics
Marcin Ślusarz [Thu, 9 Jul 2020 17:49:55 +0000 (19:49 +0200)]
intel/perf: split load_oa_metrics

Move oa_metrics_available out of load_oa_metrics and call
build_unique_counter_list outside.

This change is a preparation for the next patch. It should
not have any functional impact.

Signed-off-by: Marcin Ślusarz <marcin.slusarz@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5788>

14 months agointel/perf: fix performance counters availability after glFinish
Marcin Ślusarz [Fri, 10 Jul 2020 17:25:10 +0000 (19:25 +0200)]
intel/perf: fix performance counters availability after glFinish

Currently Linux kernel gathers performance counters at fixed
intervals (~5-10ms), so if application uses AMD_performance_monitor
extension and immediately after glFinish() asks GL driver for HW
performance counter values it might not get any data (values == 0).

Fix this by moving the "read counters from kernel" code from
"is query ready" to "get counter values" callback with a loop around
it. Unfortunately it means that the "read counters from kernel"
code can spin for up to 10ms.

Ideally kernel should gather performance counters whenever we ask
it for counter values, but for now we have deal with what we have.

Signed-off-by: Marcin Ślusarz <marcin.slusarz@intel.com>
Cc: <mesa-stable@lists.freedesktop.org>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5788>

14 months agointel/perf: streamline error handling in read_oa_samples_until
Marcin Ślusarz [Fri, 10 Jul 2020 14:24:31 +0000 (16:24 +0200)]
intel/perf: streamline error handling in read_oa_samples_until

No functional changes.

Signed-off-by: Marcin Ślusarz <marcin.slusarz@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5788>

14 months agointel/perf: fix how pipeline stats are stored
Marcin Ślusarz [Thu, 9 Jul 2020 18:54:41 +0000 (20:54 +0200)]
intel/perf: fix how pipeline stats are stored

It matters only when counters are not ordered by offset.

Signed-off-by: Marcin Ślusarz <marcin.slusarz@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5788>

14 months agointel/perf: fix calculation of used counter space
Marcin Ślusarz [Wed, 10 Jun 2020 13:29:53 +0000 (15:29 +0200)]
intel/perf: fix calculation of used counter space

It matters only when counters are not ordered by offset.

Signed-off-by: Marcin Ślusarz <marcin.slusarz@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5788>

14 months agoradv: report a better error message when QueueWaitIdle() failed
Samuel Pitoiset [Fri, 7 Aug 2020 15:18:41 +0000 (17:18 +0200)]
radv: report a better error message when QueueWaitIdle() failed

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6230>

14 months agoradv: report errors back to the application via VK_EXT_debug_report
Samuel Pitoiset [Fri, 7 Aug 2020 14:21:31 +0000 (16:21 +0200)]
radv: report errors back to the application via VK_EXT_debug_report

Help for debugging.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6230>

14 months agoradv: rework the error function helpers a bit
Samuel Pitoiset [Fri, 7 Aug 2020 14:05:46 +0000 (16:05 +0200)]
radv: rework the error function helpers a bit

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6230>

14 months agoradv: report the spirv-nir logs back to the application
Samuel Pitoiset [Fri, 7 Aug 2020 13:23:03 +0000 (15:23 +0200)]
radv: report the spirv-nir logs back to the application

Via VK_EXT_debug_report to help debugging various SPIRV->NIR issues.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6223>

14 months agoradv: Fix assert that is too strict.
Bas Nieuwenhuizen [Fri, 7 Aug 2020 18:29:07 +0000 (20:29 +0200)]
radv: Fix assert that is too strict.

The added assert fails on MSAA images if we disable FMASK ....

Reordered things.

Fixes: c6aadbae715 "radv: Don't use both DCC and CMASK for single sample images."
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/3385
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6232>

14 months agoradeon/radeon_vce: fix out of target bitrate in CBR mode (H.264)
Louis Li [Tue, 31 Mar 2020 02:20:21 +0000 (10:20 +0800)]
radeon/radeon_vce: fix out of target bitrate in CBR mode (H.264)

StoneyRidge may not comply to required target bitrate
when generating H.264 stream in CBR mode.

Signed-off-by: Louis Li <Ching-shih.Li@amd.com>
Reviewed-by: Leo Liu <leo.liu@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4384>

14 months agor600: Enable compute shaders for NIR code path
Gert Wollny [Sun, 5 Jul 2020 14:56:37 +0000 (16:56 +0200)]
r600: Enable compute shaders for NIR code path

There are still regessions compared to TGSI, but there are also many
fixes.

Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6025>

14 months agor600/sfn: Force a minimum of 4 GPRs, it seems to fix atomics
Gert Wollny [Wed, 20 May 2020 22:03:34 +0000 (00:03 +0200)]
r600/sfn: Force a minimum of 4 GPRs, it seems to fix atomics

This fixes spec@arb_compute_shader@execution@atomic-counter on
HD 5450

Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6025>

14 months agor600/sfn: handle querying SSBO size
Gert Wollny [Tue, 21 Jul 2020 09:57:52 +0000 (11:57 +0200)]
r600/sfn: handle querying SSBO size

Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6025>

14 months agor600/sfn: Correct ssbo instruction handling
Gert Wollny [Tue, 21 Jul 2020 10:00:00 +0000 (12:00 +0200)]
r600/sfn: Correct ssbo instruction handling

Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6025>

14 months agor600/sfn: correct allocating and emitting of atomics
Gert Wollny [Tue, 21 Jul 2020 09:56:51 +0000 (11:56 +0200)]
r600/sfn: correct allocating and emitting of atomics

Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6025>

14 months agor600/sfn: Add a mapping table for atomics
Gert Wollny [Tue, 21 Jul 2020 09:52:41 +0000 (11:52 +0200)]
r600/sfn: Add a mapping table for atomics

Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6025>

14 months agor600/sfn: add r600 specific lowering pass for atomics and use it
Gert Wollny [Tue, 21 Jul 2020 09:50:38 +0000 (11:50 +0200)]
r600/sfn: add r600 specific lowering pass for atomics and use it

v2: rebase to use global variable lists

Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6025>

14 months agor600/sfn: Sort uniforms by binding and offset
Gert Wollny [Tue, 21 Jul 2020 09:43:08 +0000 (11:43 +0200)]
r600/sfn: Sort uniforms by binding and offset

This is required to get atomics right.

v2: rebase to use global variable lists

Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6025>

14 months agor600: Set PIPE_CAP_NIR_ATOMICS_AS_DEREF to true
Gert Wollny [Tue, 21 Jul 2020 11:39:11 +0000 (13:39 +0200)]
r600: Set PIPE_CAP_NIR_ATOMICS_AS_DEREF to true

Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6025>

14 months agogallium + mesa/st: Add PIPE_CAP_NIR_ATOMICS_AS_DEREF and use it
Gert Wollny [Sat, 27 Jun 2020 20:40:56 +0000 (22:40 +0200)]
gallium + mesa/st: Add PIPE_CAP_NIR_ATOMICS_AS_DEREF and use it

This cap is useful for drivers that support hardware atomics and need
special handling to resolve their addresses.

Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6025>

14 months agomeson: bump required glvnd version
Eric Engestrom [Thu, 9 Jan 2020 15:10:53 +0000 (15:10 +0000)]
meson: bump required glvnd version

https://github.com/KhronosGroup/EGL-Registry/pull/95 has moved
a couple of extensions defines and functions to the upstream `eglext.h`,
but when 9a74746bd1f3bd28d4c4 sync'ed these files we broke compilation
of apps that require these symbols on systems that don't have the
updated Khronos headers.

On non-GLVND builds, we still provide these headers, so everything's
fine, but on GLVND builds the Khronos headers are external so we need to
make sure we have a libglvnd version that's recent enough.

Fixes: 9a74746bd1f3bd28d4c4 ("EGL: sync headers with Khronos")
Signed-off-by: Eric Engestrom <eric@engestrom.ch>
Acked-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6069>

14 months agodriconf: fix force_gl_vendor description
Eric Engestrom [Thu, 6 Aug 2020 14:05:51 +0000 (16:05 +0200)]
driconf: fix force_gl_vendor description

The option is not a toggle to "allow GPU vendor to be overridden", it
*is* the override.

Fixes: dca119f12c291d7665d7 ("mesa/gallium: add dric option to allow overriding GL vendor string")
Signed-off-by: Eric Engestrom <eric@engestrom.ch>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6207>

14 months agoegl/entrypoint-check: add check that GLVND and plain EGL have the same entrypoints
Eric Engestrom [Fri, 3 Apr 2020 10:29:04 +0000 (12:29 +0200)]
egl/entrypoint-check: add check that GLVND and plain EGL have the same entrypoints

Cc: mesa-stable
Signed-off-by: Eric Engestrom <eric@engestrom.ch>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4448>

14 months agoegl/entrypoint-check: split sort-check into a function
Eric Engestrom [Fri, 3 Apr 2020 10:23:27 +0000 (12:23 +0200)]
egl/entrypoint-check: split sort-check into a function

Cc: mesa-stable
Signed-off-by: Eric Engestrom <eric@engestrom.ch>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4448>

14 months agoCorrect a typo in threads_win32.h
Felix Yan [Sat, 8 Aug 2020 05:35:35 +0000 (05:35 +0000)]
Correct a typo in threads_win32.h

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6238>

14 months agopick-ui: specify git commands in "resolve cherry pick" message
Eric Engestrom [Wed, 29 Jul 2020 11:27:51 +0000 (13:27 +0200)]
pick-ui: specify git commands in "resolve cherry pick" message

Cc: mesa-stable
Signed-off-by: Eric Engestrom <eric@engestrom.ch>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6134>

14 months agozink: Fix 32-bit compilation
Joshua Ashton [Sat, 8 Aug 2020 06:24:56 +0000 (07:24 +0100)]
zink: Fix 32-bit compilation

`sizeof(void*) != sizeof(VkShaderModule)` on x86 hence this fails with `-Werror=int-conversion`

Fixes: 0f059d550bd "zink: split up creating zink_shader objects and VkShaderModule objects"
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6239>

14 months agoi965: add support for force_gl_vendor
Timothy Arceri [Thu, 6 Aug 2020 04:25:42 +0000 (14:25 +1000)]
i965: add support for force_gl_vendor

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/3363
Reviewed-by: Danylo Piliaiev <danylo.piliaiev@globallogic.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6198>

14 months agofreedreno/ir3: don't install ir3_compiler cmdline tool
Rob Clark [Fri, 7 Aug 2020 23:14:38 +0000 (16:14 -0700)]
freedreno/ir3: don't install ir3_compiler cmdline tool

It is mostly just useful to us.. and it is big since it links in nir and
most of the rest of gallium.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6234>

14 months agofreedreno/registers: add some missing regs to build
Rob Clark [Fri, 7 Aug 2020 23:04:22 +0000 (16:04 -0700)]
freedreno/registers: add some missing regs to build

Needed for installed version of crashdec/cffdump.

Fixes: 9c33c538981 ("freedreno/registers: install gzip'd register database")
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6234>

14 months agomeson/freedreno: Fix lua requirement
Dylan Baker [Fri, 7 Aug 2020 16:14:03 +0000 (09:14 -0700)]
meson/freedreno: Fix lua requirement

Freedreno needs at least Lua 5.2, but the current code will report found
for 5.1, which doesn't actually work.

Fixes: caa107cb8db930bc59cd557a325a6a0bc4a86565
       ("freedreno/decode: move dependencies up a level")

Reviewed-by: Rob Clark <robclark@freedesktop.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6229>

14 months agoradeonsi: various fixes for gfx10.3
Marek Olšák [Thu, 30 Jul 2020 09:27:47 +0000 (05:27 -0400)]
radeonsi: various fixes for gfx10.3

The magic numbers fix sample shading.
The bypass flag is optional.

Fixes: a23802bcb9a - ac,radeonsi: start adding support for gfx10.3
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6137>

14 months agoradeonsi: remove the NGG hack decreasing LDS usage to deal with overflows
Marek Olšák [Thu, 30 Jul 2020 12:32:15 +0000 (08:32 -0400)]
radeonsi: remove the NGG hack decreasing LDS usage to deal with overflows

The LDS size can't overflow anymore, so we can use the correct max LDS size.

Fixes: a23802bcb9a - ac,radeonsi: start adding support for gfx10.3
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6137>

14 months agoradeonsi: add a common function for getting the size of gs_ngg_scratch
Marek Olšák [Thu, 30 Jul 2020 12:31:22 +0000 (08:31 -0400)]
radeonsi: add a common function for getting the size of gs_ngg_scratch

The next commit will use it.

Fixes: a23802bcb9a - ac,radeonsi: start adding support for gfx10.3
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6137>

14 months agoradeonsi: don't count unusable vertices to the NGG LDS size
Marek Olšák [Thu, 30 Jul 2020 12:19:48 +0000 (08:19 -0400)]
radeonsi: don't count unusable vertices to the NGG LDS size

Now we get optimal LDS usage.

Fixes: a23802bcb9a - ac,radeonsi: start adding support for gfx10.3
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6137>

14 months agoradeonsi: fix applying the NGG minimum vertex count requirement
Marek Olšák [Thu, 30 Jul 2020 12:19:48 +0000 (08:19 -0400)]
radeonsi: fix applying the NGG minimum vertex count requirement

The code applied the restriction too late, which could overflow LDS size,
which started happening more often after the minimum vertex count was
increased for Sienna.

Incorporate the clamping into the previous code for rounding up the counts.
Now the LDS size can never overflow, but it may use vector lanes less
efficiently (max_gsprims can be decreased more), which will be addressed
in the next commit.

Fixes: 4ecc39e1aa1 ("radeonsi/gfx10: NGG geometry shader PM4 and upload")
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6137>

14 months agoradeonsi: increase minimum NGG vertex count requirement per workgroup on gfx 10.3
Marek Olšák [Wed, 29 Jul 2020 23:04:38 +0000 (19:04 -0400)]
radeonsi: increase minimum NGG vertex count requirement per workgroup on gfx 10.3

Fixes: a23802bcb9a - ac,radeonsi: start adding support for gfx10.3
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6137>

14 months agoradeonsi: use the same units for esgs_ring_size and ngg_emit_size
Marek Olšák [Thu, 30 Jul 2020 12:04:28 +0000 (08:04 -0400)]
radeonsi: use the same units for esgs_ring_size and ngg_emit_size

for consistency

Fixes: a23802bcb9a - ac,radeonsi: start adding support for gfx10.3
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6137>

14 months agoradeonsi: use correct wave size in gfx10_ngg_calculate_subgroup_info
Marek Olšák [Thu, 30 Jul 2020 11:47:42 +0000 (07:47 -0400)]
radeonsi: use correct wave size in gfx10_ngg_calculate_subgroup_info

Fixes: 88efb63cafc ("radeonsi/gfx10: implement Wave32")
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6137>

14 months agoRevert "radeonsi: honor a user-specified pitch on gfx10.3"
Marek Olšák [Wed, 29 Jul 2020 23:06:44 +0000 (19:06 -0400)]
Revert "radeonsi: honor a user-specified pitch on gfx10.3"

This reverts commit c4b5fd9ab096a0bb5106b93191b13c81cc32243b.

It breaks mipmapping. This is only meant to be used by OpenCL, which allows
setting a user pitch for linear images. In all other cases, don't support
a custom pitch.

Fixes: c4b5fd9ab096a0bb51 "radeonsi: honor a user-specified pitch on gfx10.3"
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6137>

14 months agoac/gpu_info: set num_tiles_pipes on gfx10+ too
Marek Olšák [Thu, 30 Jul 2020 01:43:04 +0000 (21:43 -0400)]
ac/gpu_info: set num_tiles_pipes on gfx10+ too

Based on PAL.

Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6137>

14 months agoradeonsi: enable ETC2 hw acceleration on Raven2
Marek Olšák [Wed, 5 Aug 2020 17:12:06 +0000 (13:12 -0400)]
radeonsi: enable ETC2 hw acceleration on Raven2

Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6213>

14 months agozink: pre-hash gfx-pipeline-state
Antonio Caggiano [Fri, 24 Jul 2020 14:50:15 +0000 (16:50 +0200)]
zink: pre-hash gfx-pipeline-state

Store a hash in `zink_gfx_pipeline_state` to keep track of state
changes and avoid to recompute it when the state has not changed.

Signed-off-by: Antonio Caggiano <antonio.caggiano@collabora.com>
Reviewed-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6061>

14 months agozink: rename zink_gfx_program::stages to 'modules'
Mike Blumenkrantz [Tue, 30 Jun 2020 17:01:09 +0000 (13:01 -0400)]
zink: rename zink_gfx_program::stages to 'modules'

we've been confusing 'stages' and 'shaders' over and over for a long time,
so maybe having a totally different name will help here

Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5970>

14 months agozink: always compile shaders in pipeline order
Mike Blumenkrantz [Tue, 30 Jun 2020 13:59:57 +0000 (09:59 -0400)]
zink: always compile shaders in pipeline order

in order to accurately perform slot/location mapping that's consistent across
stages, we need to go through the stages in order so that we can pass each successive
slot map allocation along to the next compiled stage

Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5970>

14 months agozink: start using per-stage flags for new shaders, refcount shader modules
Mike Blumenkrantz [Tue, 30 Jun 2020 13:57:19 +0000 (09:57 -0400)]
zink: start using per-stage flags for new shaders, refcount shader modules

we don't want to recompile shaders if we don't have to, so we can set bitflags
upon receiving new shader states and then compile only the stages that have
changed while refcounting the unchanged stages

Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5970>

14 months agozink: use ZINK_SHADER_COUNT instead of PIPE_SHADER_TYPES - 1 everywhere
Mike Blumenkrantz [Tue, 30 Jun 2020 12:55:07 +0000 (08:55 -0400)]
zink: use ZINK_SHADER_COUNT instead of PIPE_SHADER_TYPES - 1 everywhere

this is just for convenience and consistency

Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5970>

14 months agozink: split up creating zink_shader objects and VkShaderModule objects
Mike Blumenkrantz [Sat, 13 Jun 2020 16:54:36 +0000 (12:54 -0400)]
zink: split up creating zink_shader objects and VkShaderModule objects

the actual VkShaderModule is only needed when we're creating a program to
draw with, so this can be split off for "uncompiled" and "compiled" shader
objects which will facilitate implementing shader keys

Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5970>

14 months agozink: refcount zink_gfx_program objects
Mike Blumenkrantz [Sat, 13 Jun 2020 19:53:29 +0000 (15:53 -0400)]
zink: refcount zink_gfx_program objects

now that we're tracking these by shader, we want to ensure that they live through
each render pass successfully if there's no flush regardless of the timing when the
shader objects are destroyed. this becomes useful when we split up shader create and compile
functionality in future patches, at which point program refcounts can be changed
during successive draw calls, potentially resulting in a program being destroyed at that
point when it shouldn't be

with this patch, each shader used by the program gets a reference, with the renderpass
batch itself becoming the owner of the program such that it will be deleted
when the draw state gets invalidated and a new program is created

Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5970>

14 months agozink: adjust zink_shader struct to contain full streamout info
Mike Blumenkrantz [Sat, 13 Jun 2020 15:39:00 +0000 (11:39 -0400)]
zink: adjust zink_shader struct to contain full streamout info

move remapped register_index -> slot into a separate value on the struct
in order to preserve the register_index value

Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5970>

14 months agozink: move shader state methods for pipe_context into zink_program.c
Mike Blumenkrantz [Sat, 13 Jun 2020 14:45:22 +0000 (10:45 -0400)]
zink: move shader state methods for pipe_context into zink_program.c

just moving these so all the shader code can be in one place

Acked-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5970>

14 months agointel/dump_gpu: add an option to capture a single frame
Lionel Landwerlin [Wed, 4 Mar 2020 15:55:46 +0000 (17:55 +0200)]
intel/dump_gpu: add an option to capture a single frame

We use the driver identifier buffer to get the information about the
current frame.

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/2201>

14 months agointel/dump_gpu: fix --platform option
Lionel Landwerlin [Wed, 4 Mar 2020 15:38:48 +0000 (17:38 +0200)]
intel/dump_gpu: fix --platform option

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/2201>

14 months agoanv: track the current frame and write it into the driver identifier BO
Lionel Landwerlin [Wed, 4 Mar 2020 23:15:57 +0000 (01:15 +0200)]
anv: track the current frame and write it into the driver identifier BO

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/2201>

14 months agointel/dump_gpu: only map in GTT buffers not previously mapped
Lionel Landwerlin [Sun, 13 Oct 2019 21:42:27 +0000 (00:42 +0300)]
intel/dump_gpu: only map in GTT buffers not previously mapped

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/2201>

14 months agointel/dump_gpu: add an only-capture option
Lionel Landwerlin [Thu, 3 Oct 2019 21:55:43 +0000 (00:55 +0300)]
intel/dump_gpu: add an only-capture option

This option allows for smaller aub files to be generated by only
storing the BOs flagged with EXEC_OBJECT_CAPTURE.

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/2201>

14 months agointel/dump_gpu: set default device_override
Lionel Landwerlin [Thu, 3 Oct 2019 21:55:02 +0000 (00:55 +0300)]
intel/dump_gpu: set default device_override

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/2201>

14 months agointel/dump_gpu: further track mapping of BOs
Lionel Landwerlin [Thu, 3 Oct 2019 08:05:45 +0000 (11:05 +0300)]
intel/dump_gpu: further track mapping of BOs

We can go further in tracking what BOs are written to by the driver by
tracking when a buffer in unmapped. A BO could be mmap, written, unmap
and never be written to again. In such case we can just write the BO's
content on the first exec buf after unmap and never write it again.

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/2201>

14 months agointel/dump_gpu: only write BOs mapped by the driver
Lionel Landwerlin [Wed, 2 Oct 2019 23:13:32 +0000 (02:13 +0300)]
intel/dump_gpu: only write BOs mapped by the driver

Track what BOs are mapped by the driver and only write those. We can
safely assume that when not mapped there is no data to save.

v2: Don't forget to return the ret (Rafael)

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/2201>

14 months agozink: set primitive restart cap
Mike Blumenkrantz [Fri, 31 Jul 2020 13:32:23 +0000 (09:32 -0400)]
zink: set primitive restart cap

Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5912>

14 months agozink: use util_draw_vbo_without_prim_restart for unsupported prim modes
Mike Blumenkrantz [Wed, 1 Jul 2020 21:45:18 +0000 (17:45 -0400)]
zink: use util_draw_vbo_without_prim_restart for unsupported prim modes

this fixes up primitive restart functionality for the primitive types that
vulkan doesn't support using primitive restart with

fixes #2873

Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5912>

14 months agozink: move 8bit index handling out of u_primconvert path
Mike Blumenkrantz [Mon, 20 Jul 2020 13:04:21 +0000 (09:04 -0400)]
zink: move 8bit index handling out of u_primconvert path

putting this through util_translate_prim_restart_ib reduces our reliance on
u_primconvert, which ideally we don't want to be using

Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5912>

14 months agozink: basic primitive restart support for strip/fan topologies
Mike Blumenkrantz [Wed, 1 Jul 2020 19:57:43 +0000 (15:57 -0400)]
zink: basic primitive restart support for strip/fan topologies

this conditionally handles rewriting the index buffer to use vk-compatible
restart indexes and then enables it in the pipeline for supported draw
modes

fixes #3174

Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5912>

14 months agoradv: Add ETC2 support on RAVEN2.
Bas Nieuwenhuizen [Thu, 6 Aug 2020 22:08:21 +0000 (00:08 +0200)]
radv: Add ETC2 support on RAVEN2.

I did a whole bunch of validation and it essentially works.

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6215>

14 months agoradv: allow to force-enable LLVM internally for a specific shader stage
Samuel Pitoiset [Thu, 6 Aug 2020 15:50:33 +0000 (17:50 +0200)]
radv: allow to force-enable LLVM internally for a specific shader stage

For ACO debugging purposes, developers only.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6208>

14 months agoetnaviv: call nir_lower_bool_to_bitsize
Christian Gmeiner [Tue, 4 Aug 2020 20:44:21 +0000 (22:44 +0200)]
etnaviv: call nir_lower_bool_to_bitsize

Starting with commit 6f394343b1f ("nir/algebraic: i2f(f2i()) -> trunc()")
dEQP-GLES2.functional.shaders.operator.binary_operator.div.lowp_int_vertex
ends with an unsuppored flt instruction. Use nir_lower_bool_to_bitsize to
convert this flt to a flt32 which is supported. This fixes the introduced
regression.

Cc: 20.2 <mesa-stable@lists.freedesktop.org>
Signed-off-by: Christian Gmeiner <christian.gmeiner@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6182>

14 months agopanfrost: Drop compiler cmdstream deps
Alyssa Rosenzweig [Wed, 5 Aug 2020 22:12:00 +0000 (18:12 -0400)]
panfrost: Drop compiler cmdstream deps

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6196>

14 months agopanfrost: Treat texture dimension as first-class
Alyssa Rosenzweig [Wed, 5 Aug 2020 22:11:15 +0000 (18:11 -0400)]
panfrost: Treat texture dimension as first-class

Instead of hiding behind a cmdstream enum. The raw values have a very
simple interpretation as dimension.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6196>

14 months agopanfrost: Inline max rt into compilers
Alyssa Rosenzweig [Wed, 5 Aug 2020 22:10:41 +0000 (18:10 -0400)]
panfrost: Inline max rt into compilers

Another dep.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6196>

14 months agopan/bi: Drop use of MALI_POSITIVE
Alyssa Rosenzweig [Wed, 5 Aug 2020 22:09:51 +0000 (18:09 -0400)]
pan/bi: Drop use of MALI_POSITIVE

Contributing to a dep on the main cmdstream. It's a fancy word for
subtract one...

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6196>

14 months agopanfrost: Add panloader/ to .gitignore
Alyssa Rosenzweig [Wed, 5 Aug 2020 21:36:47 +0000 (17:36 -0400)]
panfrost: Add panloader/ to .gitignore

For out-of-tree trace infrastructure.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6196>

14 months agopanfrost: Reduce bit dependency to disassembly only
Alyssa Rosenzweig [Wed, 5 Aug 2020 21:20:25 +0000 (17:20 -0400)]
panfrost: Reduce bit dependency to disassembly only

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6196>

14 months agopan/bi: Separate disasm/compiler targets
Alyssa Rosenzweig [Wed, 5 Aug 2020 21:25:27 +0000 (17:25 -0400)]
pan/bi: Separate disasm/compiler targets

Likewise.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6196>

14 months agopan/mdg: Separate disassembler and compiler targets
Alyssa Rosenzweig [Wed, 5 Aug 2020 21:16:27 +0000 (17:16 -0400)]
pan/mdg: Separate disassembler and compiler targets

The compiler depends on the disassembler, which is okay. But the
disassembler now no longer depends on the compiler circularly, avoiding
the need to being in GLSL/NIR.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6196>

14 months agopanfrost: Move pandecode into lib/
Alyssa Rosenzweig [Wed, 5 Aug 2020 20:25:52 +0000 (16:25 -0400)]
panfrost: Move pandecode into lib/

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6196>

14 months agopanfrost: Rename encoder/ to lib/
Alyssa Rosenzweig [Wed, 5 Aug 2020 20:16:00 +0000 (16:16 -0400)]
panfrost: Rename encoder/ to lib/

We'll want both encoding and decoding here, as a generic hardware
interface library based on GenXML.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6196>

14 months agopanfrost: Don't export exception_status
Alyssa Rosenzweig [Wed, 5 Aug 2020 20:20:17 +0000 (16:20 -0400)]
panfrost: Don't export exception_status

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6196>

14 months agopanfrost: Remove panfrost-misc.h
Alyssa Rosenzweig [Wed, 5 Aug 2020 20:07:16 +0000 (16:07 -0400)]
panfrost: Remove panfrost-misc.h

Now unused.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6196>

14 months agopanfrost: Inline panfrost-misc.h into panfrost-job.h
Alyssa Rosenzweig [Wed, 5 Aug 2020 20:05:12 +0000 (16:05 -0400)]
panfrost: Inline panfrost-misc.h into panfrost-job.h

We only need a small subset of the defines here.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6196>

14 months agopan/decode: Remove shader replacement artefact
Alyssa Rosenzweig [Wed, 5 Aug 2020 20:03:09 +0000 (16:03 -0400)]
pan/decode: Remove shader replacement artefact

Did this ever work?

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6196>

14 months agopanfrost: Move format stringify to decode.c
Alyssa Rosenzweig [Wed, 5 Aug 2020 20:01:27 +0000 (16:01 -0400)]
panfrost: Move format stringify to decode.c

Let's simplify pandecode before moving it around. We don't need to share
this routine right now, and we'll be generating it soon.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6196>

14 months agopanfrost: Remove blend prettyprinters
Alyssa Rosenzweig [Wed, 5 Aug 2020 19:45:52 +0000 (15:45 -0400)]
panfrost: Remove blend prettyprinters

Unused and mostly wrong too.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6196>

14 months agoanv: fix incorrect realloc failure handling
Lionel Landwerlin [Thu, 6 Aug 2020 08:13:06 +0000 (11:13 +0300)]
anv: fix incorrect realloc failure handling

We don't want to leak in case of failure.

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reported-by: Ivan Briano <ivan.briano@intel.com>
Fixes: 246261f0addf ("anv: prepare the driver for delayed submissions")
Fixes: 34f32a6d6648 ("anv: implement VK_KHR_timeline_semaphore")
Reviewed-by: Ivan Briano <ivan.briano@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6200>

14 months agoaco: fix C++11/C++14 compilation
Rhys Perry [Thu, 6 Aug 2020 22:21:47 +0000 (23:21 +0100)]
aco: fix C++11/C++14 compilation

static_assert without a message is only available since C++17.

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Fixes: d1f992f3c2d ('aco: rework barriers and replace can_reorder')
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/3374
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6216>

14 months agost/mesa: Treat vertex outputs absent in outputMapping as zero in mesa_to_tgsi
Danylo Piliaiev [Wed, 5 Aug 2020 15:07:06 +0000 (18:07 +0300)]
st/mesa: Treat vertex outputs absent in outputMapping as zero in mesa_to_tgsi

After updating vertex outputs being written based on optimized NIR, they may
go out of sync with outputs in mesa IR. Which is translated to TGSI and used
together with NIR if draw doesn't have llvm.

It's much easier to treat such outputs as zero because there is no pass to
entirely get rid of them.

Similar to eeab9c93db84e5759145891e8fdde66a5cdcf917 but now for outputs.

Fixes: d684fb37bfbc47d098158cb03c0672119a4469fe
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/3365
Signed-off-by: Danylo Piliaiev <danylo.piliaiev@globallogic.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6187>

14 months agoBump development version and clear new_features
Dylan Baker [Thu, 6 Aug 2020 16:34:42 +0000 (09:34 -0700)]
Bump development version and clear new_features

Signed-off-by: Dylan Baker <dylan.c.baker@intel.com>
14 months agoanv: Advertise shaderIntegerFunctions2
Jason Ekstrand [Wed, 22 Jul 2020 20:23:46 +0000 (15:23 -0500)]
anv: Advertise shaderIntegerFunctions2

We advertised the extension string but never the feature bit.  Doh!

Fixes: c57338b92471 "anv: Enable SPV_INTEL_shader_integer_functions2..."
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6033>

14 months agopanfrost: signal syncobj if nothing is going to be flushed
Icenowy Zheng [Wed, 5 Aug 2020 20:48:05 +0000 (04:48 +0800)]
panfrost: signal syncobj if nothing is going to be flushed

When nothing is going to be flushed, the kernel will get no job that
signals the syncobj.

Signal it by ourselves, otherwise it will never get signaled.

Closes: #3371
Signed-off-by: Icenowy Zheng <icenowy@aosc.io>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Cc: mesa-stable
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6190>

14 months agoradv: Don't use both DCC and CMASK for single sample images.
Bas Nieuwenhuizen [Wed, 5 Aug 2020 22:51:38 +0000 (00:51 +0200)]
radv: Don't use both DCC and CMASK for single sample images.

Fixes: c67ef7695a2 "radv: Use ac_surface to allocate aux surfaces."
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6194>

14 months agoappveyor: Use Python3.
Jose Fonseca [Tue, 4 Aug 2020 13:16:37 +0000 (14:16 +0100)]
appveyor: Use Python3.

This implied upgrading to the Visual Studio 2019 image, not for VS
itself, but for the newer Python 3.8.5 version it contains, to avoid
UnicodeDecodeError inside modulefinder module when attempting to decode
our UTF-8 encoded Python scripts with cp1252 encoding.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6184>

14 months agoappveyor: Upgrade pip.
Jose Fonseca [Wed, 5 Aug 2020 09:01:24 +0000 (10:01 +0100)]
appveyor: Upgrade pip.

To avoid all those warnings.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6184>

14 months agoutil: Fix SCons build.
Vinson Lee [Thu, 6 Aug 2020 05:44:12 +0000 (22:44 -0700)]
util: Fix SCons build.

Fixes: 848e7b947d0d ("util: Move stack debug functions to src/util")
Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Reviewed-by: Kristian H. Kristensen <hoegsberg@google.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6199>

14 months agofreedreno: Add more asserts for DST_OFF/NUM_UNIT in indirect const uploads.
Eric Anholt [Fri, 31 Jul 2020 16:54:09 +0000 (09:54 -0700)]
freedreno: Add more asserts for DST_OFF/NUM_UNIT in indirect const uploads.

These are just empirical alignment numbers from looking at dEQP traces of
the blob driver (a330, a418, a540, a618, a630), with one exception noted
in the comments.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5990>

14 months agofreedreno: Increase the NUM_UNIT on compute's consts in indirect dispatch.
Eric Anholt [Fri, 31 Jul 2020 23:18:41 +0000 (16:18 -0700)]
freedreno: Increase the NUM_UNIT on compute's consts in indirect dispatch.

Avoids tripping the assert in the next commit -- the blob never uses
num_unit % 4 != 0 for indirect const uploads.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5990>

14 months agofreedreno/ir3: Clean up instrlen setup.
Eric Anholt [Fri, 17 Jul 2020 21:08:43 +0000 (14:08 -0700)]
freedreno/ir3: Clean up instrlen setup.

We were calculating it with the gpu_id check in two places, do it once and
use ir3_compiler for the gpu_id dependency.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5990>

14 months agofreedreno: Split ir3_const's user buffer and indirect upload APIs.
Eric Anholt [Thu, 9 Jul 2020 23:57:29 +0000 (16:57 -0700)]
freedreno: Split ir3_const's user buffer and indirect upload APIs.

They're almost entirely split by whether you're uploading user buffer or
from a BO.  While I'm rewriting the API, drop the emit_const ->
fdN_emit_const wrapper in favor of a #define before the header and a
little helper for the asserts.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5990>

14 months agofreedreno: Rename emit_const_bo() to emit_const_ptrs().
Eric Anholt [Thu, 9 Jul 2020 23:23:49 +0000 (16:23 -0700)]
freedreno: Rename emit_const_bo() to emit_const_ptrs().

I keep thinking it's the "upload from inside a BO" path when it's not.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5990>

14 months agofreedreno/ir3: Simpify the immediates from an array of vec4 to array of dwords.
Eric Anholt [Thu, 9 Jul 2020 22:00:33 +0000 (15:00 -0700)]
freedreno/ir3: Simpify the immediates from an array of vec4 to array of dwords.

We usually had to split the idx/swiz out of the dword index anyway.  Note
that incidentally, immediates_size now increments in vec4s instad of
4*vec4s.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5990>