AlexIndustrial/mesa
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
196,452 Commits 1 Branch 0 Tags
0714fbff8ff0192f501a45ea1695787c0ec24dee
Commit Graph

196452 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
Dylan Baker 0714fbff8f anv: assert that we don't read off the end color_att array 
Coverity notices that we've insured that index index is < MAX_RTS in one
case, but that it must be greater in one case. Since `color_att_count`
is a uint32_t, it can easily exceed MAX_RTS (8), and would thus create
an out-of-bounds read situation. While the type system would allow this,
the actually implementation shouldn't, so an assert should make Coverity
happy and help us check our assumption.

CID: 1620440
Fixes: d2f7b6d5a7 ("anv: implement VK_KHR_dynamic_rendering_local_read")
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31640>
 2024-10-16 19:09:25 +00:00
Samuel Pitoiset 17dc91709d radv: use radv_normalize_blend_factor() more 
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31688>
 2024-10-16 18:24:22 +00:00
Samuel Pitoiset 38e12dd53c radv: optimize breaking batch when CB_TARGET_MASK change 
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31688>
 2024-10-16 18:24:22 +00:00
Deborah Brouwer f406595ec7 ci/deqp-runner: implement max-fails 
If we get more than 40 unexpected fails/crashes in a merge pipeline,
then stop early to preserve pipeline resources.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31429>
 2024-10-16 17:32:53 +00:00
Deborah Brouwer fd9e0a41dd ci/deqp-runner: uprev from 0.20.0 to 0.20.2 
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31429>
 2024-10-16 17:32:53 +00:00
Deborah Brouwer aa53f73be1 ci/deqp-runner: update instructions for uprevving 
Since deqp-runner is already being built in the debian base image for test
jobs, it’s not necessary to build it subsequently in GL and VK debian
test images. Update the image tag instructions for uprevving deqp-runner.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31429>
 2024-10-16 17:32:53 +00:00
Deborah Brouwer 6a564d84e2 ci/deqp-runner: print version info 
Since deqp-runner is built in several different images, print the
deqp-runner version before running it just to double check that we are
using the expected version.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31429>
 2024-10-16 17:32:53 +00:00
Deborah Brouwer 5acdef547e ci/deqp-runner: fix option to build from a commit 
Since the 'git clone --branch' option only accepts branch names or tags as
arguments, it’s not currently possible to build deqp-runner directly from
a git commit hash.

Revise the deqp-runner build script so that it can build from a commit
even if it doesn’t have a tag or version number.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31429>
 2024-10-16 17:32:53 +00:00
Deborah Brouwer 3e901f7c37 ci: remove obsolete build-hang-detection script 
The build-hang-detection script was part of the old parallel-deqp-runner
repository which is now superseded by deqp-runner.

Remove the script because it’s no longer used.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31429>
 2024-10-16 17:32:53 +00:00
Chia-I Wu 7f4f20c811 panvk: advertise VK_KHR_external_{fence,semaphore}_fd 
Remove panvk_GetPhysicalDeviceExternal{Fence,Semaphore}Properties and
let vk_common_GetPhysicalDeviceExternal{Fence,Semaphore}Properties
handle them.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31647>
 2024-10-16 16:48:09 +00:00
Samuel Pitoiset fc52c6358c radv: stop recomputing the viewport xform for guarband/viewport 
It's already computed when viewports are bound.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31680>
 2024-10-16 16:10:12 +00:00
Chia-I Wu da80d7dc86 panvk: advertise VK_EXT_external_memory_dma_buf 
It is already supported.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31649>
 2024-10-16 15:19:25 +00:00
Chia-I Wu a78b7d3c52 panvk: advertise VK_KHR_external_memory_fd 
It is already supported.

Also remove an outdated TODO because pan_kmod_bo_import supports
reference counting.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31649>
 2024-10-16 15:19:25 +00:00
Chia-I Wu 21f6ec9e8e panvk: remove panvk_GetPhysicalDeviceImageFormatProperties 
Let vk_common_GetPhysicalDeviceImageFormatProperties handle it.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31649>
 2024-10-16 15:19:25 +00:00
Chia-I Wu 95db0250e6 panvk: fill in VkExternalImageFormatProperties 
Opaque fds are internally dma-bufs.  We also support both export and
import.  But for dma-bufs, we additonally require the image tiling
allows vkGetImageSubresourceLayout for export and
VkImageDrmFormatModifierExplicitCreateInfoEXT for import.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31649>
 2024-10-16 15:19:25 +00:00
Chia-I Wu 75f833ca82 panvk: fill in VkExternalBufferProperties 
Opaque fds are internally dma-bufs.  We also support both export and
import.

Fixes: 17b81d1fdc ("panvk: expose KHR_external_memory and KHR_external_memory_capabilities")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31649>
 2024-10-16 15:19:25 +00:00
Valentine Burley 50a9d1b987 vulkan/format: Update vk_format_from_pipe_format with additional formats 
This has been neglected.

Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Signed-off-by: Valentine Burley <valentine.burley@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30821>
 2024-10-16 14:30:16 +00:00
Valentine Burley aa01a8c018 vulkan/format: Properly translate 10-bit and 12-bit formats 
Some hardware have native 10-bit and 12-bit formats. In order to be able to support these, we need
to translate these VK_FORMATs to new padded pipe formats instead of regular 16-bit formats.

Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Signed-off-by: Valentine Burley <valentine.burley@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30821>
 2024-10-16 14:30:16 +00:00
Valentine Burley af338d2d2e tu: Fix incorrect bpcs value for padded formats 
This doesn't make a difference for now as we don't expose any padded formats yet,
but will be needed for VK_FORMAT_G10X6_B10X6R10X6_2PLANE_420_UNORM_3PACK16.

Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Signed-off-by: Valentine Burley <valentine.burley@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30821>
 2024-10-16 14:30:15 +00:00
Valentine Burley c8a8543af7 vulkan: Fix incorrect bpcs value for padded formats 
Skip padding channels and only consider valid color channels.
Add and use a common helper for this.

Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Signed-off-by: Valentine Burley <valentine.burley@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30821>
 2024-10-16 14:30:15 +00:00
Valentine Burley 9eb315ff98 hk: Remap 10 and 12 bit formats to 16 bit formats 
Preserves the previous behavior while handling the new formats.

Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Acked-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Signed-off-by: Valentine Burley <valentine.burley@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30821>
 2024-10-16 14:30:15 +00:00
Valentine Burley a33538cb9f nvk: Remap 10 and 12 bit formats to 16 bit formats 
Preserves the previous behavior while handling the new formats.

Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Signed-off-by: Valentine Burley <valentine.burley@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30821>
 2024-10-16 14:30:15 +00:00
Valentine Burley ab298b9c3a lavapipe: Remap 10 and 12 bit formats to 16 bit formats 
Preserves the previous behavior while handling the new formats.

Reviewed-by: Konstantin Seurer <konstantin.seurer@gmail.com>
Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Signed-off-by: Valentine Burley <valentine.burley@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30821>
 2024-10-16 14:30:15 +00:00
Konstantin Seurer 0098f8ef35 radv: Remap 10 and 12 bit formats to 16 bit formats 
Preserves the previous behavior while handling the new formats.

Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30821>
 2024-10-16 14:30:15 +00:00
Valentine Burley 81ebd6ea8d util/format: Add new 12-bit P012 RGB/planar formats 
Introduce three 12-bit formats: X4R12_UNORM, X4R12X4G12_UNORM, and X4G12_X4B12X4R12_420_UNORM.
These formats allocate 12 bits for each color channel with 4 bits of padding to align with Vulkan's P012
and related planar formats.

Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Signed-off-by: Valentine Burley <valentine.burley@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30821>
 2024-10-16 14:30:15 +00:00
Valentine Burley 1134ad8799 util/format: Add new 10-bit P010 RGB/planar formats 
Introduce three 10-bit formats: X6R10_UNORM, X6R10X6G10_UNORM, and X6G10_X6B10X6R10_420_UNORM.
These formats allocate 10 bits for each color channel with 6 bits of padding to align with Vulkan's P010
and related planar formats.

Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Signed-off-by: Valentine Burley <valentine.burley@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30821>
 2024-10-16 14:30:15 +00:00
Boris Brezillon ad13268c72 panvk: Fix the ZStencil descriptor initialization in the preload logic 
We're not expected to write stencil/depth if the preload draw is reloading
color attachments.

Fixes: 0bc3502ca3 ("panvk: Implement a custom FB preload logic")
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com>
Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31678>
 2024-10-16 13:20:56 +00:00
Boris Brezillon 2806ff022d panvk: Set .is_blit=true when compiling preload shaders 
Preload shaders are blit shaders.

Fixes: 0bc3502ca3 ("panvk: Implement a custom FB preload logic")
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com>
Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31678>
 2024-10-16 13:20:56 +00:00
Boris Brezillon 9f099ca6e0 pan/util: Discard depth/stencil writes when early fragment test is forced 
Writing depth/stencil when update/kill is set to force-early seems to
trip out Valhall GPUs. Since depth/stencil writes are supposed to be
ignored in that case anyway, drop them from shader.

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com>
Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31678>
 2024-10-16 13:20:56 +00:00
Christian Gmeiner 8dc8fa88ed etnaviv: nir: Switch to derivative intrinsics 
Signed-off-by: Christian Gmeiner <cgmeiner@igalia.com>
Acked-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Reviewed-by: Lucas Stach <l.stach@pengutronix.de>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31034>
 2024-10-16 12:32:07 +00:00
Christian Gmeiner 5a15b36a64 etnaviv: nir: Enforce stricter swizzle for virtual scalar x register 
This change enforces stricter swizzle behavior for the virtual scalar x
register, addressing a regression encountered in piglit's
spec@glsl-1.10@execution@derivatives@glsl-derivs-abs-sign test.
The regression occurred after switching to derivative intrinsics.

CC: mesa-stable
Signed-off-by: Christian Gmeiner <cgmeiner@igalia.com>
Acked-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Reviewed-by: Lucas Stach <l.stach@pengutronix.de>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31034>
 2024-10-16 12:32:07 +00:00
Danylo Piliaiev 7b09fc98fb nir/opt_16b_tex_image: Sign extension should matter for texel buffer txf 
Texel buffer could be arbitrary large, so the assumption being made in
the following comment is wrong:

 "Zero-extension (u16) and sign-extension (i16) have
  the same behavior here - txf returns 0 if bit 15 is set
  because it's out of bounds and the higher bits don't matter."

Sign extension should matter for GLSL_SAMPLER_DIM_BUF.

This fixes the case of doing texelFetch with u16 offset:

  uniform itextureBuffer s1;
  uint16_t offset = some_ssbo.offset;
  value = texelFetch(s1, offset).x;

If the offset is higher than s16 optimization incorrectly
left it as 16b.

In spirv the above glsl is translated into:

  %22 = OpLoad %ushort %21
  %23 = OpUConvert %uint %22
  %24 = OpBitcast %int %23
  %26 = OpImageFetch %v4int %16 %24

Cc: mesa-stable

Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Reviewed-by: Georg Lehmann <dadschoorse@gmail.com>
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Reviewed-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31664>
 2024-10-16 10:10:00 +00:00
Martin Krastev 5f7f33f323 svga/ci: disable vmware farm 
Disable farm for DUT maintenance.

Signed-off-by: Martin Krastev <martin.krastev@broadcom.com>
 2024-10-16 11:38:48 +03:00
Benjamin Lee 49e9cd5211 panvk: don't dirty preloaded tiles for writeback 
Tiles that are only touched by the preload shader don't need to be
written back, because their contents will be the same. The comment
suggests that this was the original intent, and that the value of
clean_fragment_write was just swapped by accident.

Signed-off-by: Benjamin Lee <benjamin.lee@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31652>
 2024-10-16 07:00:29 +00:00
Timothy Arceri aa7c59e02c nir/glsl: set deref cast mode for blocks during function inlining 
More cast fixes this time for UBO and SSBO. Which were missing testing
previously.

Fixes: d681cf96fb ("nir/glsl: set deref cast mode during function inlining")
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/11587

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31668>
 2024-10-16 06:25:57 +00:00
Leder, Brendan Steve bd6f3e30f2 amd/vpelib: Fix output_ctx gamma curve cache bug 
[Why]
Reprogramming gamma curve unnecessary and wastes power/time.

[How]
Bool array checked instead of bool inside said array, add index.

[Testing]
Tested with corresponding test cases.

Reviewed-by: Roy Chan <Roy.Chan@amd.com>
Reviewed-by: Krunoslav Kovac <Krunoslav.Kovac@amd.com>
Acked-by: Chih-Wei Chien <Chih-Wei.Chien@amd.com>
Signed-off-by: Brendan Leder <breleder@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31605>
 2024-10-16 02:30:17 +00:00
Jude Shih 259b44d97b amd/vpelib: Restructure CDC FE/BE 
[Why]
CDC has two separate functionalities : backend and frontend.
split them into two for better resource management.

Reviewed-by: Roy Chan <Roy.Chan@amd.com>
Acked-by: Chih-Wei Chien <Chih-Wei.Chien@amd.com>
Signed-off-by: Jude Shih <shenshih@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31605>
 2024-10-16 02:30:17 +00:00
Roy Chan 0d7f51724b amd/vpelib: Add documentation 
Reviewed-by: Navid Assadian <navid.assadian@amd.com>
Acked-by: Chih-Wei Chien <Chih-Wei.Chien@amd.com>
Signed-off-by: Roy Chan <roy.chan@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31605>
 2024-10-16 02:30:17 +00:00
Brendan 5826ef5e44 amd/vpelib: Visual confirm on input 
[Why]
The gap pointer should advance if it is used.
Not a functional bug fix but a nice-to-have change.

[How]
Advancing the pointer once the memory is used

Reviewed-by: Jesse Agate <Jesse.Agate@amd.com>
Acked-by: Chih-Wei Chien <Chih-Wei.Chien@amd.com>
Signed-off-by: Brendan <brendanSteve.leder@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31605>
 2024-10-16 02:30:17 +00:00
Assadian, Navid a0d2f074cd amd/vpelib: Add mirror & rotation capability check function for vpe1.1 
Reviewed-by: Roy Chan <Roy.Chan@amd.com>
Acked-by: Chih-Wei Chien <Chih-Wei.Chien@amd.com>
Signed-off-by: Navid Assadian <navid.assadian@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31605>
 2024-10-16 02:30:16 +00:00
Brendan af01d7a181 amd/vpelib: Update configs to be per-pipe specific 
[Why]
config settings should be per pipe in concept

[How]
update the framework to store configs per pipe

Reviewed-by: Roy Chan <Roy.Chan@amd.com>
Acked-by: Chih-Wei Chien <Chih-Wei.Chien@amd.com>
Signed-off-by: Brendan <brendanSteve.leder@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31605>
 2024-10-16 02:30:16 +00:00
Assadian, Navid 6a68af7d21 amd/vpelib: Add better rotation and mirror capability check 
Add a general rotation and mirror support check support.

Reviewed-by: Roy Chan <Roy.Chan@amd.com>
Acked-by: Chih-Wei Chien <Chih-Wei.Chien@amd.com>
Signed-off-by: Navid Assadian <navid.assadian@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31605>
 2024-10-16 02:30:16 +00:00
Evan 3ef8e6a6ae amd/vpelib: Luma AND Color Keyer Full Support 
[New]
- Added new vpe_stream params for color keying
- Added new struct in dpp to capture keying params
- Added new capability for color keying
- Added keying support in vpe1.0

[Updated]
- Updated capability check
- Updated Luma and Color Keying functions to better implement the entire feature
- Updated resource to map stream params -> dpp keyer param

Reviewed-by: Roy Chan <Roy.Chan@amd.com>
Acked-by: Chih-Wei Chien <Chih-Wei.Chien@amd.com>
Signed-off-by: Evan <evan.damphousse@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31605>
 2024-10-16 02:30:16 +00:00
Assadian, Navid 338760d9b5 amd/vpelib: Document public API structures 
Doxygen style inline comments are added to the public API functions and
structures.

Reviewed-by: Roy Chan <Roy.Chan@amd.com>
Acked-by: Chih-Wei Chien <Chih-Wei.Chien@amd.com>
Signed-off-by: Navid Assadian <navid.assadian@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31605>
 2024-10-16 02:30:16 +00:00
Brendan ded1a2b3f0 amd/vpelib: Multiple instance support in caching framework 
Generalize the caching to work with multiple instances of objects.
Change some static functions to public functions to maximize function
re-use possibilities.

Reviewed-by: Roy Chan <Roy.Chan@amd.com>
Acked-by: Chih-Wei Chien <Chih-Wei.Chien@amd.com>
Signed-off-by: Brendan <brendanSteve.leder@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31605>
 2024-10-16 02:30:16 +00:00
Evan 7a293a812a amd/vpelib: Color Keyer Implementation 
[New] - New color keyer function. Take input from VPE_STREAM and FORMAT, program - Added color keyer call in Front end - Hard code Alpha Keyer Range to be entire range - Added vpe_is_yuv() to determine if input format is any YUV format. Combines existing functions - Set Per-pixel Alpha in vpelib when stream.enable_luma_key=1

[Updated]
- Updated the macros to include the luma_key field
- Updated vpe10 and resource calls to have mathching function calls
- removed unused data structure for keying mode since hard code keyer mode to 0x3 in real world
- Updated Luma Key enum to properly represent reg spec

[Removed]
- Removed unused alpha keyer struct. Opted to take in directly from stream_ctx

Reviewed-by: Brendan Steve Leder <brendanSteve.leder@amd.com>
Acked-by: Chih-Wei Chien <Chih-Wei.Chien@amd.com>
Signed-off-by: Evan <evan.damphousse@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31605>
 2024-10-16 02:30:16 +00:00
Marek Olšák 2272db2ac6 radeonsi: set the valid buffer range for bindless image buffers 
This was missing.

Cc: mesa-stable

Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31651>
 2024-10-16 00:27:56 +00:00
Marek Olšák 88f057eb53 gallium/u_threaded: fix crash in tc_create_image_handle due to resource == NULL 
Fixes: 3df9d8ed80 - gallium/u_threaded: implement pipelined partial buffer uploads using CPU storage
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/12020

Reviewed-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31651>
 2024-10-16 00:27:56 +00:00
Paulo Zanoni 4d60f905b3 anv/trtt: extract anv_trtt_first_bind_init_queue() 
The code that initializes each queue got big enough that the
repetitive error handling is getting ugly and it could benefit from
being on its own function.

v2: Rebase, try to improve the comments.

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30953>
 2024-10-15 23:05:31 +00:00
Paulo Zanoni 8280a6fee9 anv/trtt: convert anv_trtt_bind arrays to util_dynarray 
Since the L2 bug fix we've been overestimating l3l2_binds by a lot in
most of the cases: almost every single call to anv_sparse_bind_trtt
ends up using either 0 or 1 elements for l3l2_binds, with occasionally
something using 512 or more. By switching to util_dynarray we can
guarantee the best of every case:
  - l1_binds will remain a stack array for the vast majority of the
    calls
      - even more than before, since STACK_ARRAY was limited to 8
        elements and now we do 32
  - l1 will be properly dimensioned without the need for reallocs
  - l3l2_binds will be completely empty most of the times and only
    trigger allocations when necessary

Here's the top 10 most common results of anv_sparse_bind_trtt() for a
trace of Assassin's Creed: Valhalla. The first column is how many
times we had that case while running the trace. After this patch, all
these cases will proceed without any memory allocations.

    168 trtt_binds: num_vm_binds:04 l3l2:0000 l1:0004
    344 trtt_binds: num_vm_binds:01 l3l2:0000 l1:0004
    420 trtt_binds: num_vm_binds:01 l3l2:0000 l1:0012
    422 trtt_binds: num_vm_binds:04 l3l2:0000 l1:0008
    479 trtt_binds: num_vm_binds:01 l3l2:0000 l1:0024
    560 trtt_binds: num_vm_binds:03 l3l2:0000 l1:0003
   1005 trtt_binds: num_vm_binds:01 l3l2:0000 l1:0002
   1024 trtt_binds: num_vm_binds:02 l3l2:0000 l1:0004
   2145 trtt_binds: num_vm_binds:02 l3l2:0000 l1:0002
   3735 trtt_binds: num_vm_binds:01 l3l2:0000 l1:0001

Only 70 out of total 11340 calls to anv_sparse_bind_trtt() contained
l3l2 elements.

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30953>
 2024-10-15 23:05:31 +00:00