936ebd3370
panvk: Rename non_vs_attribs into img_attribs
...
Images attributes are the only non-VS attributes we might store in
the attribute table, so let's rename fields and functions to make that
clear.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com >
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com >
Acked-by: Erik Faye-Lund <erik.faye-lund@collabora.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28927 >
2024-05-22 14:13:16 +00:00
5a37a62bed
panvk: Set unused attribute buffers descriptors to zero
...
We assume two attribute buffer slot for each attribute buffer. Let's
fill unused slots with zeros so we don't risk unexpected read accesses
if the slot contains garbage.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com >
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com >
Acked-by: Erik Faye-Lund <erik.faye-lund@collabora.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28927 >
2024-05-22 14:13:16 +00:00
a195486e6f
panvk: Fully transition to vk_multisample_state
...
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com >
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com >
Acked-by: Erik Faye-Lund <erik.faye-lund@collabora.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28927 >
2024-05-22 14:13:16 +00:00
a3e024a59e
panvk: Fully transition to vk_depth_stencil_state
...
Replace our ZS state by the vk_depth_stencil_state provided by the
vulkan runtime lib.
This was the last bit using panvk_dynamic_state_bits enum, so we kill it
here along with the parse_dynamic_state() function.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com >
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com >
Acked-by: Erik Faye-Lund <erik.faye-lund@collabora.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28927 >
2024-05-22 14:13:16 +00:00
a58171fb3c
panvk: Use vk_color_blend_state to fill our blend constant
...
We need to keep some panvk-specific blend state at the panvk_pipeline
level until we've made RSD emission fully dynamic.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com >
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com >
Acked-by: Erik Faye-Lund <erik.faye-lund@collabora.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28927 >
2024-05-22 14:13:16 +00:00
bbcf505a08
panvk: Fully transition to vk_input_assembly_state
...
No need to have our own input assembly state.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com >
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com >
Acked-by: Erik Faye-Lund <erik.faye-lund@collabora.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28927 >
2024-05-22 14:13:16 +00:00
07afc7e3ed
panvk: Fully transition to vk_rasterization_state
...
There's no point storing the rasterizer state twice. Use
vk_rasterization_state everywhere, and let the core implement
CmdSetLineWidth() and CmdSetDepthBias() for us.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com >
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com >
Acked-by: Erik Faye-Lund <erik.faye-lund@collabora.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28927 >
2024-05-22 14:13:16 +00:00
f4ce783f0e
panvk: Fully transition to vk_viewport_state
...
Pre-emitting viewport descriptors at pipeline creation time makes things
more complex for little savings. Be dumb and emit viewport at draw time
based on the current command buffer dynamic state.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com >
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com >
Acked-by: Erik Faye-Lund <erik.faye-lund@collabora.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28927 >
2024-05-22 14:13:16 +00:00
6b1a8226fa
panvk: Transition the graphics pipeline logic to vk_graphics_pipeline_state
...
Fill a vk_graphics_pipeline_state and pass it around instead of passing
VkGraphicsPipelineCreateInfo.
We also call vk_cmd_set_dynamic_graphics_state() from CmdBindPipeline()
to prepare for a smooth 'make it all dynamic' transition.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com >
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com >
Acked-by: Erik Faye-Lund <erik.faye-lund@collabora.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28927 >
2024-05-22 14:13:16 +00:00
160db68bd7
panvk: Kill the panvk_pipeline_builder object
...
The pipeline builder object is not super useful, and it's now getting
in the way of the vk_graphics_state transition. Let's get rid of it and
pass CreateInfo around for now.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com >
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com >
Acked-by: Erik Faye-Lund <erik.faye-lund@collabora.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28927 >
2024-05-22 14:13:16 +00:00
2e0081d44c
panvk: Use memory pools to store pipeline shaders/descriptors
...
This greatly simplifies the pipeline creation logic, and the memory
footprint overhead of two panvk_pool objects should be negligible
compared to all the states we already have in a pipeline object.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com >
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com >
Acked-by: Erik Faye-Lund <erik.faye-lund@collabora.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28927 >
2024-05-22 14:13:16 +00:00
bd6e32ef9c
panvk: Split compute/graphics pipeline objects
...
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com >
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com >
Acked-by: Erik Faye-Lund <erik.faye-lund@collabora.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28927 >
2024-05-22 14:13:16 +00:00
513e0bd46b
panvk: Split the graphics and compute state at the cmd_buffer level
...
Prepare the ground for the transition to vk_graphics_state.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com >
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com >
Acked-by: Erik Faye-Lund <erik.faye-lund@collabora.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28927 >
2024-05-22 14:13:16 +00:00
5196aeacf6
panvk: Get rid of special attribute support
...
This is a leftover of the Midgard support removal.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com >
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com >
Acked-by: Erik Faye-Lund <erik.faye-lund@collabora.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28927 >
2024-05-22 14:13:16 +00:00
04b20157f8
panvk: Kill unused dynamic state bits
...
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com >
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com >
Acked-by: Erik Faye-Lund <erik.faye-lund@collabora.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28927 >
2024-05-22 14:13:16 +00:00
94e28d350d
panvk: Move panvk_cmd_state::batch to panvk_cmd_buffer::cur_batch
...
We are about to split the compute/graphics state, but before we do
that, move the batch field out of the cmd_state.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com >
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com >
Acked-by: Erik Faye-Lund <erik.faye-lund@collabora.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28927 >
2024-05-22 14:13:16 +00:00
330a75b9f3
panvk: Get rid of panvk_descriptor_state::dirty
...
This field is only set, never tested. When we want to flag a descriptor
dirty, we simply set the relevant mali_ptr in panvk_descriptor_state
to zero.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com >
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com >
Acked-by: Erik Faye-Lund <erik.faye-lund@collabora.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28927 >
2024-05-22 14:13:16 +00:00
a3fb990a86
panvk: Kill panvk_queue_get_device()
...
We use to_panvk_device() elsewhere.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com >
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com >
Acked-by: Erik Faye-Lund <erik.faye-lund@collabora.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28927 >
2024-05-22 14:13:16 +00:00
f57af8fe1c
panvk: clang-format fixups
...
Reconcile recent modifications with clang-format.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com >
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com >
Acked-by: Erik Faye-Lund <erik.faye-lund@collabora.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28927 >
2024-05-22 14:13:16 +00:00
90f8be9bda
ci: define SNMP base interface on runner
...
In order to turn on/off through SNMP DuT under PoE switch, the SNMP key
in some vendors don't directly use the interface number, but a number
shifted a base number.
Define this base number as BM_POE_BASE environment in the runner.
Reviewed-by: Jose Maria Casanova Crespo <jmcasanova@igalia.com >
Signed-off-by: Juan A. Suarez Romero <jasuarez@igalia.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29306 >
2024-05-22 12:09:55 +00:00
18c736bcfc
radeonsi: Use max_se instead of num_se where appropriate
...
Scratch allocation needs to happen using max_se, otherwise there can be
hangs.
Cc: mesa-stable
Reviewed-by: Marek Olšák <marek.olsak@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29202 >
2024-05-22 10:35:01 +00:00
db564a40b3
radv: Use max_se instead of num_se where appropriate
...
Scratch allocation needs to happen using max_se, otherwise there can be
hangs.
Cc: mesa-stable
Reviewed-by: Marek Olšák <marek.olsak@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29202 >
2024-05-22 10:35:01 +00:00
1966b6c887
zink+nvk/ci: update expected failures
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28207 >
2024-05-22 10:06:16 +00:00
acd395f1f8
zink+nvk/ci: document flakes seen during stress-testing
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28207 >
2024-05-22 10:06:16 +00:00
fc1db264ed
zink+nvk/ci: skip more tests that times out
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28207 >
2024-05-22 10:06:16 +00:00
0bd551f6ec
zink+nvk/ci: skip timing out test
...
Instead of spending our time running it to make sure it times out, which
is not very useful.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28207 >
2024-05-22 10:06:16 +00:00
393cd1ffdf
zink+nvk/ci: skip glx piglit tests as they all fail
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28207 >
2024-05-22 10:06:16 +00:00
5a6ffd1420
zink/ci: add zink+nvk glcts+piglit job on a GA106 (RTX 3060)
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28207 >
2024-05-22 10:06:16 +00:00
94c82cd938
nvk/ci: add nvk job on a GA106 (RTX 3060)
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28207 >
2024-05-22 10:06:16 +00:00
3d6957268b
aco: use new common helpers for building buffer descriptors
...
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29268 >
2024-05-22 08:31:39 +00:00
074f3cfe73
radv: use new common helpers for building buffer descriptor
...
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29268 >
2024-05-22 08:31:39 +00:00
d3b01fd95e
amd/common: add new helpers to build buffer descriptors
...
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29268 >
2024-05-22 08:31:39 +00:00
3224fd706c
amd/common: only pass gfx_level to ac_get_gfx10_format_table()
...
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29268 >
2024-05-22 08:31:39 +00:00
64fefc1179
ac,radv,radeonsi: add a common helper for translating swizzle
...
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29268 >
2024-05-22 08:31:39 +00:00
2487a87552
ac,radv,radeonsi: add function to determine if alpha should be on MSB
...
The only difference for RADV is that the helper now converts SRGB
formats to non-SRGB but that shouldn't change anything in practice.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29308 >
2024-05-22 08:17:31 +02:00
179f4ed414
radv: adjust determining if alpha should be on MSB
...
Ported from RadeonSI.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29308 >
2024-05-22 08:14:56 +02:00
68c4d26691
radv: only set ALPHA_IS_ON_MSB if the image has DCC on GFX6-9
...
This is technically incorrect to only check meta_offset which might be
non-zero for CMASK/FMASK but this applies to DCC only.
Cc: mesa-stable
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29308 >
2024-05-22 08:14:56 +02:00
3e8b2fe053
broadcom/simulator: Add DRM_IOCTL_V3D_GET_COUNTER to simulator
...
As this new IOCTL was introduced in the kernel, mirror the change in
the simulator.
Signed-off-by: Maíra Canal <mcanal@igalia.com >
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29154 >
2024-05-22 05:37:48 +00:00
e630812b43
broadcom/simulator: Add DRM_V3D_PARAM_MAX_PERF_COUNTERS parameter support
...
As this new parameter was introduced in the kernel, mirror the change in
the simulator.
Signed-off-by: Maíra Canal <mcanal@igalia.com >
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29154 >
2024-05-22 05:37:48 +00:00
017dde0d1c
v3d: Use DRM_IOCTL_V3D_GET_COUNTER to get perfcnt information
...
Currently, the information about the performance counters is duplicated
both in the kernel and in user space. Naturally, this leads to
inconsistency, as the user space might be updated and while the kernel
isn't.
Aiming to turn the kernel as the "single source of truth", use
DRM_IOCTL_V3D_GET_COUNTER, when available, to get performance counter
information.
Signed-off-by: Maíra Canal <mcanal@igalia.com >
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29154 >
2024-05-22 05:37:48 +00:00
c5b2d943ad
v3dv: Use DRM_IOCTL_V3D_GET_COUNTER to get perfcnt information
...
Currently, the information about the performance counters is duplicated
both in the kernel and in user space. Naturally, this leads to
inconsistency, as the user space might be updated and the kernel isn't.
Aiming to turn the kernel as the "single source of truth", use
DRM_IOCTL_V3D_GET_COUNTER, when available, to get the performance
counter information.
Signed-off-by: Maíra Canal <mcanal@igalia.com >
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29154 >
2024-05-22 05:37:48 +00:00
273ba51d7f
broadcom/common: Add maximum number of perf counters to v3d_device_info
...
Now, the kernel has the ability to inform about the maximum number of
performance counters of a V3D device. Let's add this information to the
`struct v3d_device_info` to use it when performing performance queries.
From now on, V3D_PERFCNT_NUM must not be used to retrieve the maximum
number of performance counters. We must use `devinfo->max_perfcnt`,
except on the case that the kernel doesn't support DRM_V3D_PARAM_MAX_PERF_COUNTERS.
Signed-off-by: Maíra Canal <mcanal@igalia.com >
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29154 >
2024-05-22 05:37:47 +00:00
ce7bca176f
drm-uapi: Update v3d_drm.h
...
From https://cgit.freedesktop.org/drm/drm-misc/
commit 673087d8b023faf34b84e8faf63bbeea3da87bab
Author: Maíra Canal <mcanal@igalia.com >
Date: Sun May 12 19:23:29 2024 -0300
drm/v3d: Deprecate the use of the Performance Counters enum
Signed-off-by: Maíra Canal <mcanal@igalia.com >
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29154 >
2024-05-22 05:37:47 +00:00
4b3e286d33
venus: add missing sTypes for vk_set_physical_device_properties_struct
...
The contents were previously copied with vk_copy_struct_guts(),
now that we use vk_set_physical_device_properties_struct, the sType
is needed.
Fixes: ("3c152a6e5dd venus: Use common physical device properties")
Signed-off-by: Juston Li <justonli@google.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29314 >
2024-05-21 23:01:37 +00:00
b99c48b011
aco/lower_phis: don't create boolean loop header phis in some situations
...
If we have a loop with continue_or_break and no divergent exits, there is
no need for a loop header phi.
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/29121 >
2024-05-21 21:28:13 +00:00
4ae8a558b2
aco: remove nir_to_aco
...
This isn't used anymore
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/29121 >
2024-05-21 21:28:13 +00:00
b1964f03e7
aco: use scalar phi lowering for lcssa workaround
...
This lets us use non-undef for the last operand, if necessary
(demonstrated in the test).
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/29121 >
2024-05-21 21:28:13 +00:00
bbe4652430
aco: create lcssa phis for continue_or_break loops when necessary
...
These might not exist because adding would decrease the quality of
divergence analysis. They are necessary for continue_or_break though, so
add them later, where they won't affect divergence analysis.
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com >
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev >
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/10623
Cc: mesa-stable
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29121 >
2024-05-21 21:28:13 +00:00
3fc7207f50
aco/lower_phis: create loop header phis for non-boolean loop exit phis
...
These might be necessary if continue_or_break and divergent breaks are both used:
loop {
if (divergent) {
a = loop_invariant_sgpr
break
}
discard_if
}
b = phi a
If we break because discard_if makes exec empty but only did so in
previous iterations, then the phi should use "a" from those previous
iterations.
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/29121 >
2024-05-21 21:28:13 +00:00
175d2d680a
wsi/wayland: Fix use after free from improperly stored VkAllocationCallbacks
...
These callbacks are no longer valid when cleaning up, and a use after free
occurs.
There's no need to store this at all anyway, so just stop doing that.
Fixes: 57c03fe4https://gitlab.freedesktop.org/mesa/mesa/-/issues/11184
Signed-off-by: Derek Foreman <derek.foreman@collabora.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29310 >
2024-05-21 21:01:25 +00:00
Boris Brezillon