tu: VK_EXT_extended_dynamic_state3 is finished

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18912>

tu: Implement extendedDynamicState3ColorWriteMask

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18912>

tu: Implement extendedDynamicState3ColorBlendEquation

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18912>

tu: Make tu_pipeline::dynamic_state_mask 64-bit

We're running out.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18912>

freedreno: Make BIT() 64-bit

In turnip we were using this a lot with the dynamic state enum, and
we're running out of space there because we're needing to add more and
more dynamic states that don't correspond to draw states. Make it
64-bit-safe so we don't need to rewrite everything in turnip. In the
case where the thing being operated on is 32-bit the compiler can
usually optimize it away, as can be seen with the release build size
before and after:

before:
   text    data     bss     dec     hex filename
5404913  293592   22744 5721249  574ca1 /home/cwabbott/build/mesa-release/lib64/libvulkan_freedreno.so
   text    data     bss     dec     hex filename
13981320         498550  205000 14684870         e012c6 /home/cwabbott/build/mesa-release/lib64/dri/msm_dri.so

after:
   text    data     bss     dec     hex filename
5404969  293592   22744 5721305  574cd9 /home/cwabbott/build/mesa-release/lib64/libvulkan_freedreno.so
   text    data     bss     dec     hex filename
13981320         498550  205000 14684870         e012c6 /home/cwabbott/build/mesa-release/lib64/dri/msm_dri.so

In the end the only changes is an additional ~50 bytes of text in
turnip.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18912>

tu: Implement extendedDynamicState3ColorBlendEnable

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18912>

tu: Remove blend dependency on attachment format

This matches freedreno, still passes all the tests, and will make
dynamic blend easier because we can compute more up-front.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18912>

tu: Implement extendedDynamicState3SampleLocationsEnable

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18912>

tu: Clamp sample locations

This was missed in the initial implementation and fixes extreme sample
locations like (1.0, 1.0).

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18912>

tu: Implement extendedDynamicState3ProvokingVertexMode

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18912>

tu: Implement extendedDynamicState3LineRasterizationMode

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18912>

tu: Impement extendedDynamicState3RasterizationStream

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18912>

tu: Rename RASTERIZER_DISCARD state to PC_RASTER_CNTL

It also contains the rasterization stream.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18912>

tu: Implement extendedDynamicState3DepthClipNegativeOneToOne

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18912>

tu: Implement extendedDynamicState3AlphaToOneEnable

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18912>

tu: Implement extendedDynamicState3AlphaToCoverageEnable

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18912>

tu: Implement extendedDynamicState3RasterizationSamples

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18912>

tu: Set default sample mask to all 1's

This avoids a dependency on the sample count in the blend state, and
seems to work. Otherwise, we'd need to make blend dynamic if samples is
dynamic and record whether the sample mask was NULL, which is a lot more
complicated.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18912>

tu: Split up tu6_update_msaa()

Most of the time we will only be updating either the number of samples
or whether it should be disabled, not both, and we don't need to compare
both. With pipelines we were comparing both, but with dynamic
rasterization samples we want to only update disable when binding the
pipeline and only update samples when calling
vkCmdSetRasterizationSamplesEXT(). Stop optimizing the uncommon case
where both are changed when binding a pipeline, and split it into 2
parts while sharing the common part that records and emits the state
packet.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18912>

tu: Don't use rasterizationSamples when enabling sample shading

From the spec language, it seems like this change wasn't strictly
required and is just an optimization for when minSampleShading would
be small enough to allow one sample per pixel. However
rasterizationSamples will soon possibly be dynamic, and I don't think we
should keep this around.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18912>

tu: Implement extendedDynamicState3SampleMask

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18912>

tu: Implement extendedDynamicState3LogicOpEnable

This is a little tricky because now we always have to store the
translated logicOp in the pipeline, regardless of whether it's enabled
or not, because the enable/disable may now be dynamic even if the
logicOp is not.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18912>

tu: Implement extendedDynamicState3Depth*Enable

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18912>

tu: Implement extendedDynamicState3TessellationDomainOrigin

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18912>

tu: Implement extendedDynamicState3PolygonMode

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18912>

tu: Combine GRAS_SU_CNTL drawstate with rast draw state

Emit GRAS_SU_CNTL, GRAS_CL_CNTL, the polygon mode, and the VRS registers
in one draw state. We're running out of draw states, and this saves a
draw state while preparing us for the rest of the rasterization state to
be dynamic.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18912>

tu: VK_EXT_extended_dynamic_state3 boilerplate

Enable the extension but don't enable anything yet.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18912>

tu: tu: Clear patchControlPoints dirty state with static patchControlPoints

Noticed by inspection after the previous issue.

Fixes: 68f3c38c801 ("tu: Implement extendedDynamicState2PatchControlPoints")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18912>

tu: Ignore line stipple dynamic state

At least in the current Vulkan spec there is no validation language
saying that it isn't valid to set this state if stippled lines aren't
supported, so it seems we have to just ignore it. Ignore it if the user
specifies a dynamic line stipple state and don't emit warnings if they
call CmdSetLineStippleEXT because zink will do this.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18912>

tu: Clear viewport dirty state with static viewports

Vulkan allows the user to set extraneous dynamic state which then gets
ignored if a pipeline with static state is bound. We weren't
implementing this correctly for viewports because we weren't clearing
the dirty bit, but it was happening to work until changes for dynamic
depth negative-one-to-one broke
dEQP-VK.pipeline.*.depth.depth_clip_control.d32_sfloat_less_viewport_before_static.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18912>

radv/rra: Print invalid node types

Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Friedrich Vock <friedrich.vock@gmx.de>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19465>

intel: Add and use intel_gem_get_context_param()

Again sharing the same function across all Intel drivers.

There is still two additional DRM_IOCTL_I915_GEM_CONTEXT_GETPARAM
calls, one in intel/dev and other in perf.
The first one can't call intel_gem_get_context_param() because of the
build order of libs and the second one because it sets the size
parameter.

Will revisit those calls in future but this is already an improvement.

v2:
- using intel_gem_get_context_param() for the recently added query for
I915_CONTEXT_PARAM_PROTECTED_CONTENT

Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18974>

intel: Add and use intel_gem_set_context_param()

Again sharing the same function across all Intel drivers.

Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18974>

intel: Add and use intel_gem_destroy_context()

Again sharing the same function across all Intel drivers.

Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18974>

intel: Add and use intel_gem_create_context()

Add intel_gem_create_context() to common/intel_gem.c/h and use it
on Iris, Crocus, ANV and HASVK.

Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18974>

intel: Refactor intel_gem_create_context_engines()

This function was returning a int but there was no meaninfull errno
code being returned, also context_id is a uint32_t what would be
problematic if i915 even returned 2147483648(-1).

So here changing the return type and add context_id pointer parameter.

Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18974>

iris: Do not set I915_CONTEXT_PARAM_RECOVERABLE twice

For the protected context path, it was already set.

Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18974>

intel/common: Add and use intel_gem_create_context_ext()

v2:
- added flag mask bit to enable context protected and recoverable
v3:
- added enum intel_gem_create_context_flags

Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18974>

etnaviv: warn when imported TS buffer is the same as color buffer

Color and TS buffers are allocated separately for each etnaviv resource, so
getting the same base and TS buffer at import time is unexpected and a strong
hint at the application doing something wrong, like passing in the same GEM
handle for all planes on a GBM import. Print a warning to give the user some
feedback.

Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Christian Gmeiner <christian.gmeiner@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9780>

etnaviv: add support for sharing the TS buffer

This adds support for sharing the TS buffer, which up until now has been
an internal implementation detail, with the outside world. This mainly
improves performance with a GPU compositor present, but on i.MX8M also
direct to display use-cases benefit.

The impact of this change depends on the GPU generation:
- old GPUs with a single pipe won't see any difference
- GC2000 can skip the TS resolve in the client and will benefit from a
  more efficient blit into the sampler compatible format when the client
  buffer contains cleared tiles
- GC3000 can directly sample with TS support, so saves both write and read
  memory bandwidth when the client buffer contains cleared tiles
- GC7000 with compression support can keep the client buffer in compressed
  format, thus saving both read and write bandwidth even for fully filled
  client buffers
- GC7000 coupled to a display unit supporting the compression format (DCSS
  on i.MX8M) does not even need to uncompress the render buffer for display
  so will see significant bandwidth saving even when GPU compositing is
  bypassed

There is a slight complication in that the tile clear color isn't part of
the TS buffer, but is programmed into state registers in the GPU. To handle
this externally shared TS buffers now contain a software metadata area,
where the clear color is stored by the driver, so the receiving end of the
TS buffer can retrieve the clear color from this area.

The compression format is handled in the same way by storing it in the SW
meta area. While we can derive the compression format from the color buffer
format in most cases, some users, like weston, expect that they can "upgrade"
ARGB to XRGB color formats. While this works with plain color formats, as
it's just masking a channel, the compression format differs when alpha is in
use. Receivers of the TS buffer should thus not try to infer the compression
format from the color buffer format, but instead fetch it from the SW meta.

The import/export handling of the TS buffer is modelled after the Intel iris
driver: we add a separate plane for the TS buffer and fold it into the base
resource after the import.

Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Tested-by: Guido Günther <agx@sigxcpu.org>
Reviewed-by: Christian Gmeiner <christian.gmeiner@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9780>

etnaviv: canonicalize modifier on import

Unknown modifiers are currently squashed down to linear when transforming
the modifier into our interal layout representation. However, the only real
modifier that we expect to see, which isn't Vivante specific or LINEAR, is
the INVALID modifier. Treat this modifier as linear and reject any other
unexpected modifiers.

Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Christian Gmeiner <christian.gmeiner@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9780>

drm-uapi: bump headers

From drm-next at the following commit:

    commit 7f7a942c0a338c4a2a7b359bdb2b68e9896122ec
    Merge: 0a20a3ea4259 ddcb8fa6514f
    Author: Dave Airlie <airlied@redhat.com>
    Date:   Thu Oct 27 14:44:02 2022 +1000

        Merge tag 'drm-next-20221025' of git://linuxtv.org/pinchartl/media into drm-next

Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Christian Gmeiner <christian.gmeiner@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9780>

etnaviv: allocate TS memory from KMS when resource bind is SCANOUT

Some display engines are able to resolve fast clear and/or compression
on the fly and need access to the TS buffer to do so. As they might
have restrictions on which memory they can access, allocate the TS
buffer memory from the KMS side when the resource should be SCANOUT
capable.

Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Tested-by: Guido Günther <agx@sigxcpu.org>
Reviewed-by: Christian Gmeiner <christian.gmeiner@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9780>

etnaviv: rework modifier queries

Rework the loops in the modifier queries to make them a bit more
straight-forward and use less local variables. Those loops get
much more complex as more variations of the modifiers get added,
so the simplification done here is an attempt to keep the complexity
at bay.

Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Tested-by: Guido Günther <agx@sigxcpu.org>
Reviewed-by: Christian Gmeiner <christian.gmeiner@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9780>

etnaviv: correct comment about tile status size

The earlier understanding of how tile status tracks the color/depth
surface was wrong and the TS size calculations have long been fixed,
as we now know that color/depth tiles have a fixed size in bytes and
thus don't depend on the surface format. But nobody changed the comment
explaining how TS works, so it's still around as a trap for the reader.

Reword the comment to not confuse people that aren't intimately
familiar with the inner workings of tile status.

Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Tested-by: Guido Günther <agx@sigxcpu.org>
Reviewed-by: Christian Gmeiner <christian.gmeiner@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9780>

docs: make the "amber" branch's name explicit

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19507>

mesa: Add test to prevent windows.h to be included in shared headers

This test is hand crafted by running command
find | grep .*h$ >../headers.txt
under src directory, and using find replace to creating
the file list and then removed the files should not be directly included

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Acked-by: Brian Paul brianp@vmware.com
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19472>

mesa: #include "util/glheader.h" instead GL/gl.h in shared code

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Acked-by: Brian Paul brianp@vmware.com
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19472>

mapi: #include "util/glheader.h" instead of #include "GL/gl.h"

Also add macro to guard windows.h won't be included

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Acked-by: Brian Paul brianp@vmware.com
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19472>

glapi: #include "util/glheader.h" in glapi_priv.h and glapitable.h directly

So that we do not need define GLAPIENTRY repeatedly, always using the
GLAPIENTRY macro comes from GL/gl.h

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Acked-by: Brian Paul brianp@vmware.com
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19472>

mesa: Move glheader.h from mesa/main/glheader.h to util/glheader.h

So it's can be accessed in broader places

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Acked-by: Brian Paul brianp@vmware.com
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19472>

mesa: sync GLAPIENTRY with KHRONOS_APIENTRY in GL/gl.h

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Acked-by: Brian Paul brianp@vmware.com
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19472>

mesa: BUILD_GL32 is not used anymore

Remove usage of BUILD_GL32 in GL/gl.h

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Acked-by: Brian Paul brianp@vmware.com
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19472>

mesa: Remove out-dated comment in common_x86.c

Follow up: 8e3696137f2 ("remove final imports.h and imports.c bits")

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Acked-by: Brian Paul brianp@vmware.com
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19472>

util: Prevent glheader.h from including <windows.h> by defining APIENTRY

When APIENTRY is not defined, GL/gl.h and GL/glext.h will automatically
include <windows.h>, so we save the macro APIENTRY by push_macro and then
define APIENTRY before include of GL/gl.h and GL/glext.h.
After that we use pop_macro to recover the previous macro again.

Because windows.h is not included by glheader.h, mesa/main/errors.c needs
to include <windows.h> directly to prevent the following error:
errors.c:98:10: error: implicit declaration of function 'OutputDebugStringA' [-Werror=implicit-function-declaration]
   98 |          OutputDebugStringA(buf);

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Acked-by: Brian Paul brianp@vmware.com
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19472>

glapi: Remove duplicated #include "main/glheader.h"

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Acked-by: Brian Paul brianp@vmware.com
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19472>

docs: XFree86 source files in libGL.txt are out date, remove it

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Acked-by: Brian Paul brianp@vmware.com
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19472>

pvr: Fix NULL pointer indirection in pvr_enable_z_replicate()

Check input_subpass->depth_stencil_attachment is non-NULL before indirecting
it.

Signed-off-by: Sarah Walker <sarah.walker@imgtec.com>
Reviewed-by: Frank Binns <frank.binns@imgtec.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19475>

mesa: Use DEBUG_NAMED_VALUE_END for const struct debug_named_value

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Eric Engestrom <eric@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19448>

docs: Fix features.txt typo

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19500>

v3dv: fix event synchronization

Since we now implement events in the GPU we need to be more careful
and insert barriers to honor the dependencies provided by the API
as well as ensuring we are synchronizing these with the compute
queue, since that is how we implement GPU event functionality.

Fixes: ecb01d53fdb413121ce2 ("v3dv: refactor events")
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19458>

v3dv: make the helper to emit pipeline barriers public to other files

Fixes: ecb01d53fdb413121ce2 ("v3dv: refactor events")
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Reviewed-by: Eric Engestrom <eric@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19458>

v3dv: always check VK_ACCESS_2_MEMORY_READ_BIT for read accesses

Fixes: a981ac053964 ('v3dv: skip binning sync if binning shaders don't access external resources')

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Reviewed-by: Eric Engestrom <eric@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19458>

v3dv: increase limit for active event objects

Fixes: ecb01d53fdb413121ce2 ("v3dv: refactor events")
Fixes: dEQP-VK.api.command_buffers.execute_large_primary

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Reviewed-by: Eric Engestrom <eric@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19458>

v3dv: do better cleanup on failure during pipeline cache operation

Fixes (with disk cache enabled):
dEQP-VK.api.device_init.create_instance_device_intentional_alloc_fail.basic
dEQP-VK.api.object_management.alloc_callback_fail.device
dEQP-VK.api.object_management.alloc_callback_fail.device_group

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Reviewed-by: Eric Engestrom <eric@igalia.com>
cc: mesa-stable

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19458>

v3dv: handle allocation failure during pipeline initialization

Fixes (with disk cache disabled):
dEQP-VK.api.device_init.create_instance_device_intentional_alloc_fail.basic
dEQP-VK.api.object_management.alloc_callback_fail.device
dEQP-VK.api.object_management.alloc_callback_fail.device_group

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Reviewed-by: Eric Engestrom <eric@igalia.com>
cc: mesa-stable

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19458>

v3dv: fix incorrect return type

Fixes: ecb01d53fdb413121ce2 ("v3dv: refactor events")

Partially fixes:
dEQP-VK.api.device_init.create_instance_device_intentional_alloc_fail.basic
dEQP-VK.api.object_management.alloc_callback_fail.device
dEQP-VK.api.object_management.alloc_callback_fail.device_group

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Reviewed-by: Eric Engestrom <eric@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19458>

v3dv: do a better job at cleaning up the device on init failure

These leaks on device creation failure have been there before, but
were only exposed as CTS failures after the recent event refactoring.

Partially fixes:
dEQP-VK.api.device_init.create_instance_device_intentional_alloc_fail.basic
dEQP-VK.api.object_management.alloc_callback_fail.device
dEQP-VK.api.object_management.alloc_callback_fail.device_group

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Reviewed-by: Eric Engestrom <eric@igalia.com>
cc: mesa-stable

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19458>

v3dv: avoid freeing already-freed memory

Fixes: ecb01d53fdb413121ce2 ("v3dv: refactor events")
Signed-off-by: Eric Engestrom <eric@engestrom.ch>
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19458>

v3dv: remove unnecessary check for NULL

We are initializing the device, so we know this will be NULL.

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Reviewed-by: Eric Engestrom <eric@igalia.com>
cc: mesa-stable

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19458>

vk/runtime: drop incorrect UNUSED annotation

Signed-off-by: Eric Engestrom <eric@engestrom.ch>
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
cc: mesa-stable

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19458>

radv/ci: document a recent regression on VanGogh

It seems like !19442 may have been introducing a lot of flakes in the
dEQP-VK.query_pool.statistics_query.* tests.

Samuel is looking into it right now, but let's document the flakes in
the mean time so that we can get back to a green baseline.

Signed-off-by: Martin Roukala (né Peres) <martin.roukala@mupuf.org>
Acked-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19470>

radv/ci: document a flake on VanGogh

Seen only once, so not really worth filing a bug for.

Signed-off-by: Martin Roukala (né Peres) <martin.roukala@mupuf.org>
Acked-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19470>

radv/ci: document some VanGogh flakes that came from splitting the job

These tests are also flaky on NAVI21, which were also introduced when
the job was split in two.

Signed-off-by: Martin Roukala (né Peres) <martin.roukala@mupuf.org>
Acked-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19470>

radv: fix fallback for extreme geometry with tessellation on GFX11

It would assert anyways. Found by inspection.

Cc: 22.3 mesa-stable
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/19495>

anv: Reduce RHWO optimization (Wa_1508744258)

Implement Wa_1508744258:

   Disable RHWO by setting 0x7010[14] by default except during resolve
   pass.

Disable the RCC RHWO optimization at all times except when resolving
single sampled color surfaces.

v2: Move stalling to genX(cmd_buffer_apply_pipe_flushes) for clarity (Mark)

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Mark Janes <markjanes@swizzler.org>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Cc: mesa-stable
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19450>

radv: Make the compute scratch waves per SE as well.

Fixes: 278e533ec9c ("radv: update scratch buffer registers on GFX11")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19488>

radv: Fix compute scratch buffer emission.

Copied wrong from radeonsi. The registers following the scratch
buffer address are the shader rsrc1/rsrc2. Not the user SGPR0
containing the ring resource word 1.

Fixes: 278e533ec9c ("radv: update scratch buffer registers on GFX11")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19488>

radv: Style cleanup.

Now have a local info ptr.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19488>

docs/isl: remove empty headings

Reviewed-by: Eric Engestrom <eric@igalia.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19461>

docs/isl: put emphasis on unit names

This convinces the spell-checker that this is something meaniningful, it
seems...

Reviewed-by: Eric Engestrom <eric@igalia.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19461>

docs/isl: correct rst-syntax

Reviewed-by: Eric Engestrom <eric@igalia.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19461>

docs/tgsi: quote literal numbers

This makes them stand out a bit, and render using a monospace font.

Reviewed-by: Eric Engestrom <eric@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19460>

docs/tgsi: mark pseudocode as such

This is more semantically strong.

Reviewed-by: Eric Engestrom <eric@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19460>

docs/tgsi: always use uppercase for instruction names

This reads a bit cleaner.

Reviewed-by: Eric Engestrom <eric@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19460>

docs/tgsi: use src1 instead of "second arg"

This is more consistent with what we do elsewhere here.

Reviewed-by: Eric Engestrom <eric@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19460>

docs/tgsi: put emphasis on src etc

This makes them stand out a bit more.

Reviewed-by: Eric Engestrom <eric@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19460>

docs/gallium: use math-role for zero and one also

Without this, the zero and one renders with a different font than the
other equations. There's no good reason for this, so let's just use the
math-role for these too.

Reviewed-by: Eric Engestrom <eric@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19460>

docs/gallium: use table instead of list

This makes this a bit easer to read, as the equations line up properly.

Reviewed-by: Eric Engestrom <eric@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19460>

docs/gallium: use definition-lists for member-lists

For some reason, we started with headings for some of these, and then
switched to definition lists midway though.

Let's stick with the definition lists, as that looks a bit more tidy.

Reviewed-by: Eric Engestrom <eric@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19460>

ac/nir: create gs copy shader with scalar outputs

radeonsi assume IO has been scalarized, this simplify the code
and radeonsi implementation.

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19455>

rusticl: remove Option from set_global_binding

We never bind NULL resources through set_global_binding and drivers depend
on this behaviour already.

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18581>

panfrost: Don't merge workgroups with variable shared mem

If nir->info.shared_size = 0 but grid->variable_shared_mem > 0, the shader uses
shared memory but the compiler may not realize that. We need to disable
workgroup merging even in this case. The alternate approach is to statically
check for shared intrinsics in the compiler, but this is a bit easier all things
considered.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18581>

rusticl: create the constant buffer ahead of launches

v2: don't bind a const buffer if there is none

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18581>

rusticl: create the CSO ahead of launches

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18581>

rusticl: unbind compute state before deleting

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18581>

iris: invalidate sysvals if grid dimension changes

Cc: mesa-stable
Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Emma Anholt <emma@anholt.net>
Reviewed-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18581>

gallium: split up req_local_mem

This will be required if a frontend has to request additional shared mem
on top of the shader declared one, but wants to create the CSO before
knowing the total amount.

In OpenCL applications can bind additional shared mem through kernel
arguments and this happens quite late.

Note: Clover sets the req_local_mem incorrectly before so we can leave it
as broken.

v2: fix panfrost code (Alyssa)

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Alyssa Rosenzweig <alyssa@collabora.com>
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18581>

rusticl: use real constant buffer for cb0

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18581>

rusticl/nir: add load_kernel_input lowering pass

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18581>