If we reorder state atoms to emit the render condition after cache flushes,
it will automatically give us the behavior we want in si_draw.
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24732>
This allows us to remove a little bit of code from si_draw, and enable
removing more code in the future.
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24732>
This merges both loops in si_draw by tracking which pm4 states are dirty
using the state atom mechanism used for other states. pm4 states now have
to set their own emit function.
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24732>
si_pm4_state will use si_atom, and both loops in si_emit_all_states will
be merged. This is a preparation for that because si_pm4_emit needs to know
the state index.
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24732>
We have only 1 variant per gfx version except gfx10+, which have 2.
The motivation is to remove instructions from si_draw_vbo.
Code size before this commit:
si_draw_vbo<GFX11, no tess, no GS, has NGG, has pairs>: 8616 bytes
si_draw_rectangle: 272 bytes
Code size after this commit:
si_draw_vbo<GFX11, no tess, no GS, has NGG, has pairs>: 8534 bytes
si_draw_rectangle<GFX11, has NGG, has pairs>: 2295 bytes
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24732>
instead of this custom code doing the same thing. This tracks changes to LS,
ES, and VS user SGPRs separately, so that we can skip more redundant register
changes when enabling/disabling GS and tess. The perf impact should be neutral.
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24732>
Currently SPIR-V does pull all the NIR constants it creates into the
first block, but we plan to change that behavior to let those constants
be defined as they are used.
This pass was written to provide a fallback to the old behavior, it will
be used for radv to avoid regressions when performing the SPIR-V change.
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5282>
To acknowledge for disable freedreno or collabora farm, split definitions into:
- google-* (a306, a530, a630)
- collabora-* (a618, a660)
This let us control when jobs will run. This rules gets also used in zink.
Acked-by: Guilherme Gallo <guilherme.gallo@collabora.com>
Signed-off-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24665>
At least these two can be easily used in bare-metal or Labgrid setups.
Currently I already have MR for implementing these for Labgrid.
Reviewed-by: Guilherme Gallo <guilherme.gallo@collabora.com>
Signed-off-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24665>
This is required to get rid of unneeded memory operations, like direct
scratch stores/loads to the same location.
Fixes: 66c6061491 ("rusticl/kernel: get rid of initial function_temp type lowering")
Signed-off-by: Karol Herbst <git@karolherbst.de>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24734>
VkPipelineColorBlendStateCreateInfo::attachmentCount cannot be used to
generate the CWE mask since it cannot be read if enough dynamic state is in use
instead just pass the max mask and let drivers figure it out
cc: mesa-stable
Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24673>
The blob generates following values for e.g. this call.
glBlendColor(0.002000f, 0.018000f, 0.030000f, 1.0)
0xff010508, /* [01424] PE.ALPHA_BLEND_COLOR := B=0x8,G=0x5,R=0x1,A=0xff */
etnaviv's etna_cfloat_to_uint8(..) creates different values.
0.002000: 0x0
0.018000: 0x4
0.030000: 0x7
The same applies for the alpha reference value.
Lets drop this hand-rolled conversion helper to get the same values as
blob.
Signed-off-by: Christian Gmeiner <cgmeiner@igalia.com>
Reviewed-by: Lucas Stach <l.stach@pengutronix.de>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24727>
When Marge rebases, it creates two pipelines, one in the author's account
due to the rebase and another one in the target account due to the merge
request event. Depending on the order they appear, Marge erroneously
check the author's pipeline, and since it doesn't have the rights to
start this pipeline, Marge fails to merge because it timed out (since the
pipeline never got run).
Fix this by disabling the author's pipeline (source of type "push") when
a merge request is open.
We only disable when the pipeline is triggered by marge to not affect
running ci_run_n_monitor.py script
Signed-off-by: Helen Koike <helen.koike@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24730>
The write helper is just pulling code we already have out into a helper
and flipping the order of the loop and the switch. The read helper will
be useful in the next commit where we add small constant support. This
keeps the two helpers right next to each other in the file where they're
easy to compare and we can ensure that they stay in sync.
Reviewed-by: Konstantin Seurer <konstantin.seurer@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9000>
