AlexIndustrial/mesa
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
179,973 Commits 1 Branch 0 Tags
40656b6fa05faaa6582a3fc40abdc3a9d99dcf85
Commit Graph

13421 Commits

Author SHA1 Message Date
Samuel Pitoiset 790fabd38e radv: advertise VK_EXT_device_fault 
This is only exposed if the kernel supports GPUVM fault query.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25875>
 2023-11-06 10:35:05 +00:00
Samuel Pitoiset 8097becc7f radv: add initial VK_EXT_device_fault support 
This implementation only returns VM faults information for now, but
vendor crash dumps will be adder later.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25875>
 2023-11-06 10:35:05 +00:00
Marek Olšák ca1d37e1db radeonsi: adjust setting PA_SC_EDGERULE once more 
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/26055>
 2023-11-05 14:06:52 -05:00
Marek Olšák 44eaf50a34 ac/surface/tests: cosmetic changes 
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26055>
 2023-11-05 12:39:42 -05:00
Marek Olšák dfcc7f83a4 ac/surface: cosmetic changes 
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26055>
 2023-11-05 12:39:42 -05:00
Marek Olšák 355242f055 ac/gpu_info: adjust attribute ring size for gfx11 
these are better numbers

Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26055>
 2023-11-05 12:36:52 -05:00
Marek Olšák bd57630885 ac: add missing gfx11.5 bits 
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26055>
 2023-11-05 12:34:55 -05:00
Jesse Natalie 228329f4da vulkan: Consolidate common ICD methods 
Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25998>
 2023-11-03 20:01:14 +00:00
Chia-I Wu 227300345e radv: stop using vk_render_pass_state::render_pass 
vk_render_pass_state::pipeline_flags is derived from
vk_get_pipeline_rendering_flags and has the info we need.

Cc: mesa-stable
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/10074
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26000>
 2023-11-03 17:23:30 +00:00
Eric Engestrom 47398c65ee ci/radeonsi: add another flake 
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26020>
 2023-11-03 09:20:32 +00:00
Samuel Pitoiset fc9bab73a9 radv/ci: document one more flake test 
This test consistently fails on some GPUs (already documented) but on
some others it's a flake. It's a known issue that should be fixed soon
in RADV.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26019>
 2023-11-03 09:16:54 +01:00
Konstantin Seurer 11282598e6 radv: Add radv_nir_lower_hit_attrib_derefs_tests 
Tests hit attrib lowering for various variable/type configurations.

Reviewed-by: Friedrich Vock <friedrich.vock@gmx.de>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24271>
 2023-11-02 15:48:36 +00:00
Konstantin Seurer f51227d253 radv/clang-format: Do not indent C++ modifiers 
Turns

class asd {
 private:
};

into

class asd {
private:
};

Reviewed-by: Friedrich Vock <friedrich.vock@gmx.de>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24271>
 2023-11-02 15:48:36 +00:00
Konstantin Seurer ba8d3afa56 radv/nir: Handle boolean hit attribs 
Reviewed-by: Friedrich Vock <friedrich.vock@gmx.de>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24271>
 2023-11-02 15:48:36 +00:00
Konstantin Seurer 3a69424e09 radv/nir: Add radv_nir_lower_hit_attrib_derefs 
Move out the pass so it can be unit tested.

Reviewed-by: Friedrich Vock <friedrich.vock@gmx.de>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24271>
 2023-11-02 15:48:36 +00:00
Konstantin Seurer b7c582e5c7 radv: Add RADV_MAX_HIT_ATTRIB_DWORDS 
Reviewed-by: Friedrich Vock <friedrich.vock@gmx.de>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24271>
 2023-11-02 15:48:36 +00:00
Rhys Perry 0a418561da radv: skip radv_remove_varyings for mesh shaders 
Fixes compilation of a Talos Principle 2 shader.

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Fixes: 9fa9782c17 ("radv: stop compiling a noop FS when the application doesn't provide a FS")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25659>
 2023-11-02 12:44:43 +00:00
Rhys Perry ed12be533e radv: call lower_array_deref_of_vec before lower_io_arrays_to_elements 
nir_lower_io_arrays_to_elements does not support array derefs of vectors,
even when nir_deref_instr_is_known_out_of_bounds is fixed.

They can occur with mesh shaders.

Found by inspection.

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Cc: mesa-stable
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25659>
 2023-11-02 12:44:43 +00:00
Chia-I Wu 796cba9bda radv: fix vkCmdCopyImage2 for emulated etc2/astc 
When the image copy is between size-compatible formats with different
block sizes, we need to fix up the extent.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25984>
 2023-11-01 20:02:14 +00:00
Rhys Perry b18f0dec41 aco: collect Pre-Sched SGPRs/VGPRs before spilling 
The usage after spilling is usually either the same as before or the
maximum.

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25559>
 2023-11-01 19:41:30 +00:00
Rhys Perry d200916ca2 aco: add VALU/SALU/VMEM/SMEM statistics 
This lets us measure optimizations without interference of waitcnt
instructions.

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25559>
 2023-11-01 19:41:30 +00:00
Rhys Perry 7de34ad3ef radv: use NIR_LOOP_PASS helpers 
A somewhat random collection of fossils:
    N           Min           Max        Median           Avg        Stddev
x   6         16.59         16.61        16.605     16.603333  0.0081649658
+   6         15.99            16            16     15.998333  0.0040824829
Difference at 95.0% confidence
	-0.605 +/- 0.00830327
	-3.64385% +/- 0.0485573%
	(Student's t, pooled s = 0.00645497)

I'm not sure if nir_opt_if and nir_opt_loop_unroll are actually idempotent
or not.

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24197>
 2023-11-01 14:16:37 +00:00
Timur Kristóf a19e46f5d0 radv: Implement workaround for unaligned buffer/image copies. 
When the pitch or slice pitch isn't properly aligned,
the SDMA HW is unable to copy between tiled images and buffers.

To work around this, we process the image chunk by chunk,
copying the data to a temporary buffer which uses supported
pitches, and then copy it to the intended destination.

The implementation assumes that at least one pixel row of the
image will fit into the temporary buffer, and will try to copy
as many rows at once as possible. Sadly, this still results in
a lot of packets being generated for large images.

A possibe future improvement is to copy the image slice by slice
when only the slice pitch is misaligned. However, that is out
of scope for this commit.

Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Tatsuyuki Ishi <ishitatsuyuki@gmail.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25831>
 2023-11-01 13:21:01 +00:00
Timur Kristóf ec0605ff72 radv: Add temporary BO for transfer queues. 
Some copy operations are poorly supported by the SDMA hardware,
meaning that the built-in packets don't support them, so we will
need to work around that by copying to and from a temporary BO.

The size of the temporary buffer was chosen so that it can fit
at least one full pixel row of the largest possible image.

Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Tatsuyuki Ishi <ishitatsuyuki@gmail.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25831>
 2023-11-01 13:21:01 +00:00
Timur Kristóf 8156c923ee radv: Implement buffer/image copies on transfer queues. 
Previously, RADV only had a simple implementation of
image to buffer copies using the SDMA for the PRIME copy.

This commit replaces that with a full-featured implementation
that includes buffer to image and image to buffer copies and
removes the assumptions that the PRIME copy had, as well as
adds new helper functions which will be shared with other copy
functions in upcoming commits.

Unaligned buffer/image copies require a workaround, which
will be implemented by a future commit.

Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Tatsuyuki Ishi <ishitatsuyuki@gmail.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25831>
 2023-11-01 13:21:01 +00:00
Timur Kristóf ed21f1c962 radv: Expose radv_get_dcc_max_uncompressed_block_size function. 
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25831>
 2023-11-01 13:21:01 +00:00
Timur Kristóf 848f2f2b99 radv: Remove always false tmz variables from SDMA functions. 
We can re-add them later as-needed.

Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25831>
 2023-11-01 13:21:00 +00:00
Samuel Pitoiset 17daa08dff radv: emit COMPUTE_PIPELINESTAT_ENABLE for CS invocations on ACE 
This register seems needed to enable compute shader shader invocations
on GFX7. On GFX8+ it's working fine without emitting this register but
I think it doesn't hurt.

This fixes dEQP-VK.query_pool.statistics_query.*_cq on GFX7.

Fixes: a9945216ba ("radv: fix COMPUTE_SHADER_INVOCATIONS query on compute queue")
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25957>
 2023-11-01 12:46:17 +00:00
Samuel Pitoiset 9a0a77cb53 radv: fix compute shader invocations query on compute queue on GFX6 
Looks like GFX6 always writes the number of compute shader invocations
at offset 0 when used on compute queue.

This fixes dEQP-VK.query_pool.statistics_query.*_cq on GFX6.

Fixes: a9945216ba ("radv: fix COMPUTE_SHADER_INVOCATIONS query on compute queue")
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25957>
 2023-11-01 12:46:17 +00:00
Samuel Pitoiset 46dc02354a radv: adjust binning settings to improve performance on GFX9 
This partially reverts 74ab940156 which
was a fix for random GPU hangs with binning on GFX10+. Though,
according to RadeonSI, only GFX10+ is affected and this reduced perf
on GFX9 chips.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25951>
 2023-11-01 12:24:45 +00:00
Samuel Pitoiset e4a1bc70dd radv: bind the non-dynamic graphics state from the pipeline unconditionally 
The following sequence is valid (although weird) but many other drivers
(including RADV) were broken:
    - bind pipeline with some static state
    - set state command for that static state (to a bad value)
    - bind the same pipeline again
    - draw

Fixes new dEQP-VK.dynamic_state.*.double_static_bind.

Cc: mesa-stable
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25954>
 2023-11-01 09:44:13 +00:00
Samuel Pitoiset 024dab650e radv/ci: enable RADV_DEBUG=nomeshshader for vkcts-navi31-valve 
To make VKCTS on NAVI31 stable until the task shader issue is
resolved.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25969>
 2023-11-01 08:50:06 +01:00
Samuel Pitoiset a97160cad8 radv: add RADV_DEBUG=nomeshshader 
This option will be used to disable VK_EXT_mesh_shader in Mesa CI
for GFX11 because running task shader tests in parallel can hang the
GPU.

Gitlab: https://gitlab.freedesktop.org/mesa/mesa/-/issues/10051
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25969>
 2023-11-01 08:49:35 +01:00
Samuel Pitoiset 15f92c3b2c radv/ci: update list of expected failures/flakes for NAVI31 
Let's clean the flakes list, it might be needed to re-add some of them
but we will track those.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25969>
 2023-11-01 08:49:35 +01:00
Bas Nieuwenhuizen da7e6f303b radv: Add some initial graphics DGC preprocessing support. 
Just the bits that obviously need no adjustment in the DGC
preprocessing code.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25835>
 2023-11-01 00:06:10 +00:00
Bas Nieuwenhuizen c4fb827441 radv: Add compute DGC preprocessing support. 
This should reduce the overhead due to reduced syncs.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25835>
 2023-11-01 00:06:10 +00:00
Bas Nieuwenhuizen 108227a84e radv: Add DGC preprocessing barrier support. 
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25835>
 2023-11-01 00:06:09 +00:00
Samuel Pitoiset 0477346c0b aco: remove dead code in nir_intrinsic_xfb_counter_{add,sub}_amd 
This code path is only used by GFX11 now.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25903>
 2023-10-30 08:51:51 +00:00
Samuel Pitoiset d390cd7c5d ac/nir: remove dead code in nir_intrinsic_xfb_counter_{add,sub}_amd 
This code path is only used by GFX11 now.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25903>
 2023-10-30 08:51:51 +00:00
Samuel Pitoiset 5176f75e0d radv: remove unnecessary VS_PARTIAL_FLUSH for NGG streamout 
This used to be a PS_PARTIAL_FLUSH to fix synchronization issues with
NGG streamout on RDNA2 but it's no longer needed for RDNA3. It's
already synchronized in CmdEndTransformFeedbackEXT().

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25903>
 2023-10-30 08:51:51 +00:00
Samuel Pitoiset eb47e07782 radv: remove NGG streamout support for RDNA1-2 
This was useful for experimenting it on RDNA2 and during RNDA3 bringup,
but now the support is rock solid on RDNA3 and it's useless to keep the
RADV_PERFTEST=ngg_streamout option.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25903>
 2023-10-30 08:51:51 +00:00
Samuel Pitoiset 7beddd4f5c radv: use the GPUVM fault protection status helper 
To print more useful information when a fault happens.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25855>
 2023-10-30 08:10:23 +00:00
Samuel Pitoiset e18c4bca31 ac/debug: add a helper to print GPUVM fault protection status 
This basically prints the same info as dmesg.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25855>
 2023-10-30 08:10:22 +00:00
Samuel Pitoiset bc09932ec3 ac/registers: allow to parse GCVM_L2_PROTECTION_FAULT_STATUS 
To have defined bitfields for this register.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25855>
 2023-10-30 08:10:22 +00:00
Samuel Pitoiset 7d96fe853e radv: fix a synchronization issue with primitives generated query on RDNA1-2 
Only RDNA1-2 are affected because RADV needs to handle the legacy vs
NGG path for this query, and the NGG results are stored with 2 extra
64-bit values.

Fixes flakes with
dEQP-VK.transform_feedback.primitives_generated_query.* since VKCTS
1.3.7.0.

Cc: mesa-stable
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25862>
 2023-10-30 07:33:06 +00:00
Mike Blumenkrantz c4283e32e3 radv: correctly return oom from the device when failing to create a cs 
cc: mesa-stable

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25927>
 2023-10-27 15:29:32 +00:00
Samuel Pitoiset 7ec2544a9b radv: dump the pipeline hash to the gpu hang report 
It can be useful to verify that a specific pipeline causes a hang.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25908>
 2023-10-27 12:10:51 +00:00
Karol Herbst 1c619b668d radv: fix buffers in vkGetDescriptorEXT with size not aligned to 4 
The range alignment didn't happen through GetDescriptorEXT as it called
write_buffer_descriptor directly. So simply move the align
from write_buffer_descriptor_impl into write_buffer_descriptor.

Fixes: 46e0c77582 ("radv: implement VK_EXT_descriptor_buffer")
Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Georg Lehmann <dadschoorse@gmail.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25837>
 2023-10-27 10:52:55 +00:00
Faith Ekstrand d5c310899a nir: Split nir_lower_subgroup_options::lower_vote_eq into two bits 
On NVIDIA, we can do a vote_ieq on bool in one hardware op so we don't
want that lowered.  We do want to lower vote_feq and other vote_ieq,
though.

Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25894>
 2023-10-26 23:05:44 +00:00
Samuel Pitoiset 4f8a225387 ac/gpu_info: remove bogus assertion about number of COMPUTE/SDMA queues 
For example, my polaris10 GPU now returns 3 compute queues.

Cc: mesa-stable
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25876>
 2023-10-26 13:44:21 +00:00