AlexIndustrial/mesa
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
155,563 Commits 1 Branch 0 Tags
515112eabd9903ed0570535fbdb18fe168eedcb9
Commit Graph

143932 Commits

Author SHA1 Message Date
Ruijing Dong 515112eabd radeonsi/vcn: prepare for unified queue in vcn4 
- apply unified queue ib headers to vcn4
  - re-use encoding queue as unified queue
  - define unified queue functions and structures

Reviewed-by: Boyuan Zhang <Boyuan.Zhang@amd.com>
Reviewed-by: Mihai Preda <mhpreda@gmail.com>
Signed-off-by: Ruijing Dong <ruijing.dong@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16911>
 2022-06-16 03:30:47 +00:00
Mark Janes 7b74747854 iris: provide a callback to INTEL_MEASURE to clean up snapshots 
Snapshots are processed asynchronously by INTEL_MEASURE, but snapshot
memory is allocated and associated with an iris batch.  Provide a
callback that will free snapshot memory after a batch is fully
processed.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16571>
 2022-06-16 02:58:08 +00:00
Mark Janes a1a5f6430a intel: provide a callback to clean up after intel_measure_gather 
The caller may have passed ownership of intel_measure_batch structures
to intel_measure until they are ready to be gathered.  The caller
needs a notification when rendering is complete and snapshots have
been processed, so it can free the resources that measure the batch.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16571>
 2022-06-16 02:58:08 +00:00
Mark Janes 34a130fedf anv: fix performance bug in INTEL_MEASURE 
Re-allocating the buffer object for snapshots carries a heavy penalty
at run-time.  When resetting a command buffer, the buffer object that
is allocated for snapshots may be re-used directly on subsequent
renders.

Stale snapshot data will persist in the buffer object.  To verify that
rendering is complete, zero the final timestamp value and check that
it has been written before gathering data.

Reviewed-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/16571>
 2022-06-16 02:58:08 +00:00
Mark Janes c4c096e66e intel: relax assertion in INTEL_MEASURE 
It is possible that a secondary command buffer was submitted with no
renders in it.  For that case, no timestamp will be collected.  Only
verify that timestamps if the index is nonzero.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16571>
 2022-06-16 02:58:08 +00:00
Mark Janes 3c53c6b247 intel: parse intel_measure environment without side effects 
If an application links agaist both iris and anv, they will clash when
parsing the INTEL_MEASURE environment variable.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16571>
 2022-06-16 02:58:08 +00:00
Jan Beich 78523eea2f vulkan/wsi: treat EBADF as missing DMA_BUF_IOCTL_{EXPORT,IMPORT}_SYNC_FILE 
FreeBSD supports DMA-BUF but not DMA_BUF_IOCTL_SYNC and similar yet.
As it returns EBADF instead of ENOTTY all consumers fail due to
VK_ERROR_OUT_OF_HOST_MEMORY.

Fixes: 30b57f10b3 ("vulkan/wsi: Signal semaphores and fences from the dma-buf")
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17070>
 2022-06-16 02:33:23 +00:00
Mike Blumenkrantz 1783a261f6 zink: only enable vtn debugging if validation is active 
this is otherwise incredibly cumbersome and eats up minutes in ci

Acked-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17069>
 2022-06-16 02:18:53 +00:00
Mike Blumenkrantz 1bd94b07f6 zink: always bind on the sparse queue 
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17066>
 2022-06-16 02:01:53 +00:00
Mike Blumenkrantz 0f9bc1ebe3 zink: detect sparse-only binding queue 
still prefer binding on the gfx queue when available, but store a separate
sparse binding queue if necessary

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17066>
 2022-06-16 02:01:53 +00:00
Mike Blumenkrantz d1456a6b0a zink: add semaphore handling for sparse binds 
this is a little tricky since it's possible to need a truly unreasonable
number of sparse binds in one go, so the current implementation will
daisy-chain through the sparse submits such that:
* sparse bind A creates signal semaphore X
* sparse bind B waits on X, creates signal semaphore Y
* sparse bind C waits on Y, creates signal semaphore Z

thus, any number of sparse binds will produce exactly one semaphore
that can be waited on by the following submit

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17066>
 2022-06-16 02:01:53 +00:00
Mike Blumenkrantz 285a5144cc zink: add batch api for adding a wait semaphore 
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17066>
 2022-06-16 02:01:53 +00:00
Mike Blumenkrantz 48b56d2efd zink: use a dynarray for semaphore wait flags 
ensure that any number of semaphore waits is handled as expected

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17066>
 2022-06-16 02:01:53 +00:00
Mike Blumenkrantz f5ac058812 zink: remove thread_queue 
I had originally intended to hook this up properly with synchronization et al,
but I didn't do it, and now it's seeming less smart anyway, so let's
remove the footgun

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17066>
 2022-06-16 02:01:53 +00:00
Connor Abbott cba6da2b21 tu: Save/restore *_BIN_CONTROL in 3d GMEM store path 
These are normally only set once because it's constant across the entire
renderpass, but they're trashed by the 3d store path because it needs to
store to CCU instead of GMEM. Therefore we need to save/restore them. Do
it in a way compatible with #5181.

Fixes: b157a5d ("tu: Implement non-aligned multisample GMEM STORE_OP_STORE")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17058>
 2022-06-16 01:10:24 +00:00
Connor Abbott a9d7b47613 tu: Add missing WFI to the 3d GMEM store path 
Similar to the 2d path.

Fixes: b157a5d ("tu: Implement non-aligned multisample GMEM STORE_OP_STORE")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17058>
 2022-06-16 01:10:24 +00:00
Connor Abbott 705c0d0373 tu: Fix 3d GMEM store swizzle in texture descriptor 
Even though image views for attachments must use the identity swizzle,
there are cases where we have to add in our own swizzle, in particular
for D24S8 when the view is depth-only/stencil-only. Therefore we have to
reset it to the identity, similar to what we do with input attachments.

Fixes: b157a5d ("tu: Implement non-aligned multisample GMEM STORE_OP_STORE")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17058>
 2022-06-16 01:10:24 +00:00
Rhys Perry 33e7ba2e3e aco: update SMEM offset workaround for LLVM 15 
This isn't needed since LLVM 15's b0ccf38b018.

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-By: Tatsuyuki Ishi <ishitatsuyuki@gmail.com>
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/6663
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17036>
 2022-06-16 00:47:51 +00:00
Mike Blumenkrantz d301883aba zink: use uint64_t for batch_id 
this maps directly to the vulkan api and allows removal of timeline
wrapping code

the consequence of this is a ~0.26% reduction in drawoverhead performance
on base cases (n=1000), but the simplification and deletions seem worth it

Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17045>
 2022-06-16 00:30:45 +00:00
Tiago Koji Castro Shibata 4622e99e10 d3d12: fallback to integrated adapter instead of arbitrary one 
Using the integrated adapter when none is specified uses less power by
default and doesn't break scenarios on Optimus systems (for example, on
Surface Books, detaching the screen gets prohibited because the GPU on
the performance base is in use)

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17005>
 2022-06-16 00:19:54 +00:00
Mike Blumenkrantz 3b867ae69b zink: anv ci updates 
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17061>
 2022-06-16 00:05:52 +00:00
Mike Blumenkrantz bd859e1f96 zink: clamp renderpass layers better 
this needs to be the minimum layer, not the maximum

cc: mesa-stable

Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17061>
 2022-06-16 00:05:52 +00:00
Mike Blumenkrantz 7edecbf4c2 zink: clamp dynamic fb layers better 
this needs to be the minimum layer, not the maximum

Fixes: c81cd989c8 ("zink: use dynamic rendering (most of the time)")

Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17061>
 2022-06-16 00:05:52 +00:00
Mike Blumenkrantz f74df6205d zink: add a function for getting the minimum framebuffer layers 
this clamps the layer count to the smallest number of layers rather
than the largest

cc: mesa-stable

Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17061>
 2022-06-16 00:05:52 +00:00
Emma Anholt 979f213110 ci/iris: Disable blender-demo-cube_diorama on APL. 
It has timed out on 3 jobs today.

Fixes: 96f0944a69 ("ci/panfrost: add Blender, Warzone2100, Freedoom and Unvanquished traces")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17041>
 2022-06-15 23:28:23 +00:00
Connor Abbott 6dfabab08f freedreno/a6xx: Support more 3-component formats 
These should be trivially enableable. This gets us some test coverage on
turnip because there are no tests of RGB32 texel buffers.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16980>
 2022-06-15 22:48:38 +00:00
Connor Abbott 2e63c570dd tu: Enable uniform texture buffers with NPOT formats 
This is enough for zink to expose ARB_texture_buffer_object_rgb32 and
therefore GL 4.0. We could enable sampled images with a few more
workarounds, but the blob doesn't bother and there isn't any need at the
moment.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16980>
 2022-06-15 22:48:38 +00:00
Boris Brezillon db2d264989 dzn: Enable support for geometry shaders 
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17039>
 2022-06-15 22:04:01 +00:00
Boris Brezillon e9066ff2be dzn: Disable rasterization if the last geometry stage doesn't write the position 
If we don't do that, and we get passed a dummy geometry shader (one
that has no EmitVertex() calls) we get a DXIL validation error.

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17039>
 2022-06-15 22:04:00 +00:00
Boris Brezillon e11035bf6d microsoft/compiler: Support gl_Layer 
Maps pretty nicely to SV_RenderTargetArrayIndex, so nothing complicated
here.

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17039>
 2022-06-15 22:04:00 +00:00
Boris Brezillon 9c45ee3f86 microsoft/compiler: Fix Layer type 
DXIL wants a uint, but we get passed an int.

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17039>
 2022-06-15 22:04:00 +00:00
Boris Brezillon afb64e10c1 microsoft/compiler: Move d3d12_fix_io_uint_type() to dxil_nir.c 
We currently have two implementations of the same logic. Let's pick
the d3d12 one, move it to dxil_nir.c and let nir_to_dxil() call it
when appropriate.

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17039>
 2022-06-15 22:04:00 +00:00
Boris Brezillon c9b2c79d16 d3d12: Keep access masks up-to-date when removing vars in d3d12_disable_multisampling() 
Fixes: e5cf19fced ("d3d12: Modify shaders when MSAA is disabled")
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17039>
 2022-06-15 22:04:00 +00:00
Mike Lothian e1accb1c4c util: Fix invalid usage of alignas in u_cpu_detect.c 
This fixes the following errors when compiling Mesa with Clang 14:
../mesa-9999/src/util/u_cpu_detect.c:368:5: error: expected ';' after struct
   } alignas(16) fxarea;
    ^
    ;

This has been tested with Clang 14.0.5 and GCC 12.1

Fixes: e3bc78b8e3 ("util: use c11 alignas instead of rolling our own")
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/6667
Signed-off-by: Mike Lothian <mike@fireburn.co.uk>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Reviewed-by: Yonggang Luo <luoyonggang@gmail.com>
Acked-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17035>
 2022-06-15 21:27:57 +00:00
Marek Olšák e24354c1b2 radeonsi/gfx11: rework GDS streamout code to single-lane and enable streamout 
GDS is basically scalar in gfx11.

This is not exactly how it's supposed to be done (we should be using
the GDS_STRMOUT registers), but it works.

Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16990>
 2022-06-15 20:52:42 +00:00
Marek Olšák 44e4d42c23 radeonsi/gfx11: add missing register shadowing code 
it doesn't work yet

Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16990>
 2022-06-15 20:52:42 +00:00
Marek Olšák fbd68a3839 radeonsi/gfx11: drop the ES vertex count requirement 
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16990>
 2022-06-15 20:52:42 +00:00
Marek Olšák 99fd408946 radeonsi/gfx11: don't allocate unused wait_mem_scratch 
We sync using PWS instead of memory.

Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16990>
 2022-06-15 20:52:42 +00:00
Marek Olšák 98d6a3d6c6 radeonsi/gfx11: don't use memory for waiting for cache flushes 
There is a new flush/wait mechanism called PixelWaitSync that uses
an internal counter.

Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16990>
 2022-06-15 20:52:42 +00:00
Marek Olšák 56d4e0be86 radeonsi/gfx11: synchronize correctly before setting SPI_ATTRIBUTE_RING_* 
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16990>
 2022-06-15 20:52:42 +00:00
Marek Olšák fa25eba744 radeonsi/gfx11: allocate more space for pipeline statistics 
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16990>
 2022-06-15 20:52:42 +00:00
Marek Olšák 0e8beb1eed radeonsi/gfx11: compile monolithic PS if it writes memory 
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16990>
 2022-06-15 20:52:42 +00:00
Marek Olšák ee7150da79 amd/gfx11: add PixelWaitSync packet fields 
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16990>
 2022-06-15 20:52:42 +00:00
Marek Olšák 7d8847d963 ac/llvm: fix tcs_wave_id unpacking on gfx11 
no effect on behavior because the higher bits are always zero

Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16990>
 2022-06-15 20:52:42 +00:00
Emma Anholt ae432b67b7 ci/dzn: Skip a test that intermittently times out at a minute. 
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17054>
 2022-06-15 20:30:46 +00:00
Christian Gmeiner 15f394cc7a nir: Fix unused-variable compile warnings 
Fixes: 8492e78f9d ("nir/deref: Handle SSBO array bindings specially")

Signed-off-by: Christian Gmeiner <christian.gmeiner@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17021>
 2022-06-15 19:43:27 +00:00
Timur Kristóf 08435b1c6f radv: Always use 3D block ID and grid size in task shaders. 
These are needed to address the task draw and payload ring buffers
when the task shader dispatch is 3 dimensional.

Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17023>
 2022-06-15 18:28:58 +00:00
Timur Kristóf 0fb9573c57 ac/nir/taskmesh: Preserve workgroup ID Y and Z when applying firstTask. 
NV_mesh_shader workgroups are only 1 dimensional, so it's OK to
only add firstTask to the X dimension.
However, let's keep the Y and Z dimensions so this doesn't mess up
the workgroup ID for future 3 dimensional task shader dispatches.

Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17023>
 2022-06-15 18:28:58 +00:00
Timur Kristóf b243e94f07 radv: Lower mesh shader 3D workgroup ID to 1D index. 
This allows future mesh shaders to use a 3D workgroup ID.
Also changes how the NV_mesh_shader first_task is emulated.
The new code moves the responsibility from ac_nir into radv.

Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17023>
 2022-06-15 18:28:58 +00:00
Timur Kristóf e05f63f56c ac/nir/ngg: Add mesh shader workgroup index. 
This will allow lowering 3D workgroup ID to a 1D index.

Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17023>
 2022-06-15 18:28:58 +00:00