It crashes in LLVM when the type is not scalar integer.
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35950>
LLVM removed typed pointers a long time ago. These types and bitcasts
had no effect.
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35950>
The alignment change allows LLVM to select 32-bit stores when 8-bit and
16-bit stores are properly aligned.
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35950>
These are only memory loads.
We could do the same for LDS loads, which are not truly speculatable
in merged shaders (can't be moved before the barrier), but that's fine
because LICM only moves code out of loops, which can't have barriers.
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35950>
LLVM also generates better code with this.
(-0.51% code size in 153 shaders, less SGPR spilling)
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35950>
This pulls in the newest version of the file form
open-gpu-kernel-modules version 575.64.05. This should give us updated
names that correspond to our new hardware support.
The new version also added a ton of lines where the name was "(null)".
Since these would only make our device names more confusing, I've
removed all of these lines in the new version using a script that is
also included in this commit.
Backport-to: 25.2
Acked-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/36333>
The old Nouveau GL driver has no support for GPUs after Ada. Instead,
users will always get NVK+Zink on Hopper+. Right now, if the user sets
NOUVEAU_USE_ZINK=false, the loader will return "nouveau" and EGL/GLX
will try to load that, fail, and then fall back to Zink. With this
patch, we instead print a warning message and then load Zink anyway.
Backport-to: 25.2
Reviewed-by: Mel Henning <mhenning@darkrefraction.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/36329>
When building for C23 the compiler warns about assigning boolean to
a different type.
Change the code fixing the following error:
-----------------------------------------------------------------------
../src/gallium/drivers/zink/zink_query.c: In function ‘find_or_allocate_qp’:
../src/gallium/drivers/zink/zink_query.c:213:25: error: incompatible types when assigning to type ‘_Bool *’ from type ‘_Bool’
213 | vkq_needs_reset = false;
| ^~~~~
-----------------------------------------------------------------------
Reviewed-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Fixes: 92811d9a56 ("zink: use maint9 implicit query resets when available")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/36326>
This was the only meta function (including those built into honeykrisp)
which wasn't setting view usage so we can also assert that it's always
provided.
Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/36301>
Keep the support in gen_sort_xml.py to allow it still convert
old MRs into the new format.
Acked-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Acked-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/36138>
And change the gen_sort_xml.py script to default to the new format.
Acked-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Acked-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/36138>
Change code to temporarily support both the start/end old format and the
dword/bits new format.
Acked-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Acked-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/36138>
Add a `--bits-format` argument to normalize the output to either of the
formats described below. For now, defaults to the old format.
The documentation in PRMs and BSpec describe the fields with the dword
and the bit range. Using the same convention makes easier to spot
issues.
Old format:
```
<field name="Disable SLM Read Merge Optimization" start="38" end="38" type="bool" />
<field name="Pixel Async Compute Thread Limit" start="39" end="41" type="uint" prefix="PACTL">
```
New format:
```
<field name="Disable SLM Read Merge Optimization" dword="1" bits="6:6" type="bool" />
<field name="Pixel Async Compute Thread Limit" dword="1" bits="9:7" type="uint" prefix="PACTL">
```
For Groups, we store the dword and if needed a offset_bits, in case
a group starts in a non-aligned position. Size and count for groups are
not changed.
Do this first for gen_sort_tags.py in case is convenient to have for the
stable tree to convert future patches from the new back into the old
format. Later patches will add support to the rest of the code.
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/36138>
Avoid noise when changing to the new GenXML field format -- which would
try to also sort it. Also add this file to be checked as part of tests.
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/36138>
so we can gather_info later.
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com>
Reviewed-by: Marek Olšák <maraeo@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/36265>
Correctly/optimally using nir_opt_varyings directly is pretty tricky. For GL, we
have all the right logic in the GLSL linker. for VK, we don't want to duplicate
this dance in every driver. Wrap it all up in a nir_opt_varyings_bulk helper
that operates on an entire pipeline of nir_shader's, following the GLSL linker's
logic. This is suitable for Vulkan drivers.
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com>
Reviewed-by: Marek Olšák <maraeo@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/36265>
Despite appearances, the current vk_pipeline implementation fails to link any
shaders, unless GPL is used or the link_geom_stages option is set (which no
drivers do). Notably monolithic pipelines don't get linked.
This patch attempts to fix our linking issues. Monolithic pipelines now get
linked, GPL optimized pipelines do too. GPL fast link is still not linked.
Geometry stages are now always linked because - despite the option - I think all
hardware wants this. Apps love writing random dead varyings for literally no
reason, which isn't free even on NVIDIA. This removes the option, effectively
setting it for all drivers, which in retrospect is the right decision.
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com>
Acked-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/36265>
