SPM is only implemented for GFX10+ on RADV/RadeonSI, although it's
technically possible on GFX9 but unused by RGP, so don't care.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25211>
According to PAL, these SQ counters can be sampled at WGP granularity.
Some SPM counters captured for RGP are using this GPU block.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25175>
Lists of files changed are only really relevant if we're pushing an MR.
We don't want to trigger 'user added .ci-farms-disabled/igalia' because
the user pushed a branch when this file was already present upstream.
Signed-off-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25205>
If we have diverging rulesets, we can have the base container jobs
possibly not being created, and the leaf container jobs then trying to
run, which ends up in us failing to create pipelines.
Signed-off-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25205>
For consistency, make our 'when: never' rules be the first rules to
match any job, either build or container.
Otherwise we could end up with some jobs having when: never and others
having when: manual.
Signed-off-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25205>
Make sure to take the db alignment into account when sizing the underlying
images.
Fixes a 360p sample from Lynne.
Cc: mesa-stable
Acked-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Lynne <dev@lynne.ee>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25168>
Any build which requires Turnip to link to libdrm will cause a
runtime dependency which could be a compatibility hazard on
platforms such as Android which generally don't ship with libdrm.
This hazard could be unintentionally triggered and to avoid this
from happening, a warning is now issued when building FD with msm
alongside other backends or if DRM is enabled due to building a
Gallium driver using DRM.
Signed-off-by: Mark Collins <mark@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25005>
These dependencies are required for building freedreno tools but on
platforms such as Android, these aren't included and would need to
be built from source. The easiest way to do this was using the
Meson WrapDB. Additionally, warnings were added when these deps are
missing as the decode tools will silently not be built otherwise.
Signed-off-by: Mark Collins <mark@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25005>
As `freedreno_kmds` can contain both `msm` and `kgsl`, this case
would just produce a build with just KGSL due to the precedence of
the check which is inconsistent behavior. Now `replay` is built for
all active KMDs with a suffix for the KMD (eg: replay-kgsl/replay-msm).
Signed-off-by: Mark Collins <mark@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25005>
These are both MSM-DRM specific subprojects which depend on it and
cannot be built without, this change is required to allow for KGSL
freedreno tool builds to compile successfully.
Signed-off-by: Mark Collins <mark@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25005>
fprintf was utilized with %n to count the amount of characters it
outputted but this was redundant since it was at the end of the
statement and fprintf itself returns the amount of characters it
writes out.
It should be noted that %n is a compatibility hazard on Android as
it isn't supported on Android due to security reasons.
Signed-off-by: Mark Collins <mark@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25005>
Could be used in shader assembly from rddecompiler when overriding
the cmdstream.
Usage:
print r1.x, r3.z
First arguments is three consecutive registers used for internal print
needs.
Second arg is register to print.
Output:
Shader Log Entries: 36
[0] 00000004 0.0000
[1] 00000000 0.0000
[2] 00000000 0.0000
[3] 00000000 0.0000
[4] beadc429 -0.3394
[5] beadc429 -0.3394
Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25005>
KGSL doesn't support IOVAs being set form userspace, so as a workaround
we pre-allocate a single buffer and hope that it always starts from the
same address.
Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25005>
With the previous commit we are now able to build Anv without
including i915_drm.h from common code.
This is important as avoids that i915 specific code is included in
common code by mistake.
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Sagar Ghuge <sagar.ghuge@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25044>
Imported bos already imported need special handling in i915.
That handling was moved to
anv_i915_gem_import_bo_alloc_flags_to_bo_flags() as the number of
imported bos is low.
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Sagar Ghuge <sagar.ghuge@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25044>
The bo_flags are i915 specific and should not be handled in common
code, so here adding it to backend as it is in the hot-path.
There still i915 bo_flags handling in anv_device_import_bo() that
will be handled in the next patch.
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Sagar Ghuge <sagar.ghuge@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25044>
Currently zink_vertex_elements_hw_state::b.strides is an array of 32-bit
unsigned values, while CmdBindVertexBuffers2EXT expects an array of
VkDeviceSize (64-bit unsigned). Instead of converting them for a call,
just change the type of the array itself.
Fixes: 76725452 ("gallium: move vertex stride to CSO")
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/9765
Signed-off-by: Sviatoslav Peleshko <sviatoslav.peleshko@globallogic.com>
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25208>
in a stream like:
* set fb state (A)
* flush
* set fb state (B)
* draw -> driver query
* flush
the "driver query" should return the tc info corresponding to the most
recent fb state (B). previously this would increment to C because
the flag for incrementing at the start of a batch was set
Fixes: 07017aa137 ("util/tc: implement renderpass tracking")
Acked-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25206>
