It means that threads are created on demand except for the first one.
It reduces process startup time.
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24173>
We were using a stale FAU buffer which meant we wrote to garbage addresses. I'm
guessing this was fallout from the big sysval rework but not 100% sure.
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Cc: mesa-stable
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24199>
dEQP-GLES31.functional.texture.texture_buffer.render.as_vertex_array_as_index_array_as_fragment_texture.offset_1_alignments
seems to got fixed by accident?
Signed-off-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24203>
Otherwise, Marge queue has to wait for someone manually trigger it.
Fixes: d3b66356 ("ci/microsoft: partly revert rename from container-rules to manual-rules")
Acked-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Signed-off-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24048>
_eglAddDevice() is now only used to look up eglDevices,
which means that the function name is a bit misleading.
Since this function is only used for finding the
eglDevice, it should also no longer support adding
a device to the egl devices list.
Signed-off-by: Robert Foss <rfoss@kernel.org>
Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/23742>
_eglAddDevice() is now only used to look up eglDevices,
which means that the function name is a bit misleading.
Since this function is only used for finding the
eglDevice, it should also no longer support adding
a device to the egl devices list.
Signed-off-by: Robert Foss <rfoss@kernel.org>
Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/23742>
droid_open_device() lists all of the DRM devices available,
but since EGL already already has created such a list use that
instead.
Signed-off-by: Robert Foss <rfoss@kernel.org>
Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/23742>
surfaceless_probe_device() lists all of the DRM devices available,
but since EGL already already has created such a list use that instead.
Signed-off-by: Robert Foss <rfoss@kernel.org>
Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/23742>
eglInitialize() must be called before the dri2 platforms probe
in order for the EGL DeviceList to be populate with devices.
Signed-off-by: Robert Foss <rfoss@kernel.org>
Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/23742>
In order to not have the different dri2 platforms all build
lists of available devices, use the device list provided by EGL.
Signed-off-by: Robert Foss <rfoss@kernel.org>
Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/23742>
It can't be 0 in Vulkan.
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24158>
NIR is tracking this better than us.
New CTS tests seem to add the requirement where in the presence of some
builtin's like gl_SampleID in a shader, even if unused, sample shading
is expected to be enabled.
See https://gitlab.khronos.org/Tracker/vk-gl-cts/-/issues/3712
Fixes:
dEQP-VK.draw.*.implicit_sample_shading.sample_id_static_use
dEQP-VK.draw.*.implicit_sample_shading.sample_position_static_use
Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24193>
Sadly GitLab CI always evaluate `rules:changes` as true, when running
scheduled runs. So we have to add explicit condition to ignore it there.
Every line using `rules:changes` has to have following condition
ignoring "schedule" runes.
Cc: mesa-stable
Reported-by: Emma Anholt <emma@anholt.net>
Reviewed-by: Eric Engestrom <eric@igalia.com>
Signed-off-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24051>
Lowers tess_coord to tess_coord_xy and math. Based on ir3's version.
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24159>
This intrinsic (vec2 tess_coord) is generally useful for non-r600 backends.
Promote it.
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24159>
It happened because glCallList was restoring varying_vp_inputs, which
caused every glCallList to process the state change again.
This loosely reverts commit 3a294ff01f
"mesa: move the _mesa_set_varying_vp_inputs call to where the state changes".
Fixes: 3a294ff01f - "mesa: move the _mesa_set_varying_vp_inputs call to where the state changes"
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24165>
Some automatic jobs, such as 'rustfmt' and 'clang-format', are skipped
during the graph sweep because their parents are already included in the
node set.
This commit ensures all visited jobs are in DAG and fixes iteration
modification using deepcopy.
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/9376
Signed-off-by: Guilherme Gallo <guilherme.gallo@collabora.com>
Acked-by: David Heidelberg <david.heidelberg@collabora.com>
Reviewed-by: Eric Engestrom <eric@engestrom.ch>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24176>
This was looking at the wrong sources. src0 is the condition.
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Konstantin Seurer <konstantin.seurer@gmail.com>
Fixes: 72ac3f6026 ("nir: add nir_unsigned_upper_bound and nir_addition_might_overflow")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/23990>
It seems this is still necessary if conformant_trunc_coord=true.
Fixes various dEQP-VK.texture.shadow.cube_array.* and
dEQP-VK.glsl.texture_functions.texture*samplercubearray* failures with
conformant_trunc_coord.
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Georg Lehmann <dadschoorse@gmail.com>
Cc: 23.2 <mesa-stable>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24137>
NGG lowering can ballots with a 32-bit destination even if the ballot bit
size is 64.
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Cc: mesa-stable
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24140>
For wave64-as-wave32, this would return 32.
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Cc: mesa-stable
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24140>
This was using the wrong pass.
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Fixes: 3f272fd15e ("ac/llvm: fix build with LLVM 17")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24140>
While we might want to employ this orphaning trick one day to avoid stalls
on busy resources it's certainly not as easy as the comment implies, as
this would involve changing all relocs in cached state objects referencing
the resource.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Christian Gmeiner <cgmeiner@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24164>
The level parameter is already checked in etna_transfer_map, so there
is no way that the unmap could end up violating the assert.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Christian Gmeiner <cgmeiner@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24164>
etnaviv gernerally uses the style where no braces are used if the
statement inside a conditional clause fits on a single line. Remove
some superfluous braces to be more consistent.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Christian Gmeiner <cgmeiner@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24164>
Move the valid buffer range tracking next to all the other things we
do on write mapping.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Christian Gmeiner <cgmeiner@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24164>
The comment makes it seem like we don't track resource status, which might
have been the case long ago, before the driver was even merged in upstream
Mesa, but is definitely no longer true. The kernel driver keeps track of
submitted work on BOs, the Mesa driver properly tracks unsubmitted usages
of any resource.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Christian Gmeiner <cgmeiner@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24164>
The current wording isn't really accurate. It talks about depth surfaces,
while in reality any surface used by the PE can have TS allocated. Also
with compression the TS doesn't only track "holes" aka clear tiles, but
instead carries the compression tags.
Reword to hopefully match reality a bit better.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Christian Gmeiner <cgmeiner@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24164>
Presence of the TS buffer doesn't necessarily mean that there is valid
TS information that we need to take into account. This is pretty harmless
as the blit into the staging resource will just ignore invalid TS data
and most TS enabled resources are (super-)tiled anyways, which we will
handle with a staging resource nonetheless, but better not keep this bad
example around and replace with the appropriate check.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Christian Gmeiner <cgmeiner@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24164>
The transfer only targets a specific level of a resource, so we don't
need to allocate all potential resource levels for the staging resource,
but instead shrink the allocation to a single level.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Christian Gmeiner <cgmeiner@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24164>
Allow to copy between different level in the source and destination
resource.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Christian Gmeiner <cgmeiner@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24164>
