For some reason the Vulkan spec required that these features must be
supported even though they only affect features that are optional
in Vulkan 1.2 and that we don't support, so enabling them doesn't
have any practical implications for us.
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17786>
A simple implementations of breadcrumbs tracking of GPU progress
intended to be the last resort when debugging unrecoverable hangs.
For best results use Vulkan traces to have a predictable place of hang.
Requires compilation with TU_BREADCRUMBS_ENABLED=1. See tu_cs_breadcrumbs.c
for details on how to use this feature.
Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15452>
It moves the explicit clamping of incoming Z from vertex stages
after interp, to the depth clamp function.
It adds support to the depth clamp function to restrict incoming
Z values to 0..1 range.
It fixes the depth test conversions to allow unrestricted depth
values.
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17612>
For unrestricted depth ranges, depth clear values for unorm buffers
need to be explicitly clamped. However this has to happen in the
driver when we know the depth buffer format, not at the API level.
Just add clamps to the non-f32 cases and separate it out.
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17612>
GLES always clamps for 32-bit float buffers, GL doesn't require
it but setting this per API causes virgl to fail some tests.
To fix is properly we'd need to introduce a CAP to expose
this between host/guest.
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17612>
This tells the drivers to expect depth values outside the 0..1 range.
This is used for support VK_EXT_depth_range_unrestricted, but could
also be used for GL_NV_depth_buffer_float in GL.
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17612>
Combine shade_quads_all() and shade_quads_mask() into new
shade_quads() function which takes a mask parameter. It calls either
jit_function[RAST_WHOLE] or jit_function[RAST_EDGE_TEST] depending on
the mask argument.
Signed-off-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17561>
Instead of 1, in shade_quads_all() and shade_quads_mask(). This fixes
a VMware test (dx10-is-front-face)
Signed-off-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17561>
Reduces sizeof(lp_fragment_shader_variant) from 296 to 288 bytes.
Signed-off-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17561>
Rename 'dst' param' to 'vals'. This parameter contains the values
which we're storing.
Signed-off-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17561>
The memset() call to zero-out the unused elements of outval[] used
the wrong bytecount. Just replace it with a simpler for loop.
Signed-off-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17561>
The writemask can be implemented with just one llvm vector shuffle.
Signed-off-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17561>
Skip multiplying by 1/w when it's 1.0.
Signed-off-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17561>
Comments, move var decls, fix code formatting.
Signed-off-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17561>
Keep sign by using int8_t to get the same result on non-x86 machines.
Reviewed-by: Axel Davy <davyaxel0@gmail.com>
Signed-off-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17698>
Possibly the bug breaking this is in box86, but given that there
aren't a whole lot of other reasons to be using Nine with a non-x86
Mesa, disable the feature unless we are running on x86.
Signed-off-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17698>
vdecl_index_map needs to be able to store negative values, but char is
not signed on all platforms, so change it to int8_t.
Signed-off-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17698>
Some of these are actually bugfixes, some like the drawcall information
are just for autotune so they are just performance fixes. However this
came from an audit into what state is used in CmdEndRenderPass().
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17378>
Emit it at EndRenderPass time, if the renderpass has tessellation. This
avoids all the special handling for secondaries, will work with
suspended/resumed render passes, and will handle secondaries that
contain render passes which will be allowed with dynamic rendering.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17378>
The workaround for draws that need a CP_WAIT_FOR_ME didn't work if the
barrier before the draw is in a separate command buffer from the draw.
The barrier would add a pending CP_WAIT_FOR_ME, but it would get dropped
on the floor at the end of the command buffer and the draw wouldn't have
a pending CP_WAIT_FOR_ME so it wouldn't emit one. We don't know in the
barrier if the destination is a draw with the workaround, so we have two
options:
- Emit any pending CP_WAIT_FOR_ME at the end of the command buffer (and
before secondaries) in case there is a workaround draw later. This
will emit an extra CP_WAIT_FOR_ME at the end of the command buffer in
case there is an indirect command barrier.
- Always assume at the beginning of the command buffer that there is a
pending CP_WAIT_FOR_ME. This will emit an extra CP_WAIT_FOR_ME before
the first workaround-requiring draw in the command buffer, in case
there was a barrier earlier.
The only draws requiring a workaround are currently
vkCmdDraw*IndirectCount(), which we assume are rarer than indirect
command barriers, so we implement the second option. This entails
treating it as a cache invalidate.
This fixes some upcoming dynamic rendering CTS tests that do
vkCmdDrawIndirectCount() in a secondary but put the barrier for it in
the primary.
Fixes: 37939e9c54 ("turnip: Fix the lack of WFM before indirect draws")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17378>
The only thing it's used for is to get the image view, and we can't rely
on it existing anyway. With dynamic rendering, we only have the format
of the attachments and sample count, so moving forward we can't rely on
anything other than that.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17378>
No driver doesn't use this option.
Signed-off-by: Georg Lehmann <dadschoorse@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17757>
This pass is problematic because it truncates sources when there are type
mismatches. With the right nir_fold_16bit_tex_image options we can ensure
that there is never a mismatch.
Other changes: bias is affected by A16 https://reviews.llvm.org/D111754
And enable G16 as we recently confirmed that it's fully independent from A16.
Signed-off-by: Georg Lehmann <dadschoorse@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17757>
The new kernel mode driver will provide the relevant information directly to
userspace, so this code is only required for pvrsrvkm.
Signed-off-by: Sarah Walker <sarah.walker@imgtec.com>
Reviewed-by: Karmjit Mahil <Karmjit.Mahil@imgtec.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17595>
Note: The format __DRI_IMAGE_FORMAT_ABGR16161616 was already
in the dri2_format_table, but had been hidden from outside view,
for lack of an official DRM fourcc. Since its fourcc has now been
assigned, it is safe to reveal the format.
Signed-off-by: Manuel Stoeckl <code@mstoeckl.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14580>
The DRM fourcc for this format is DRM_FORMAT_ABGR16161616 = 'AB48', not
__DRI_IMAGE_FOURCC_RGBA16161616 = 'RA48'. This should have no outward
effect for clients, since the format does not get revealed by
dri2_query_dma_buf_formats, and is otherwise only used within the
library.
Signed-off-by: Manuel Stoeckl <code@mstoeckl.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14580>
Iago Toral Quiroga
Eric Engestrom
Danylo Piliaiev
Marek Olšák
Dave Airlie
Adam Jackson
Brian Paul
David Heidelberg
Icecream95
Connor Abbott
Georg Lehmann
Chia-I Wu
Sarah Walker
Jordan Justen
Timur Kristóf
Manuel Stoeckl