fdo-mirrors/mesa
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/mesa/mesa.git synced 2026-04-25 04:50:38 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions 3
197158 commits 66 branches 931 tags 1,024 MiB
Commit graph

197158 commits

This branch
This branch
All branches
Author SHA1 Message Date
Alyssa Rosenzweig
b3ef0f5aa8 asahi: don't leak drm version 
valgrind.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:07 -04:00
Alyssa Rosenzweig
9ce092c982 asahi: don't leak linked shaders 
Oof!

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:07 -04:00
Alyssa Rosenzweig
ae7c9995ff asahi: don't leak binaries 
ouch. valgrind.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:07 -04:00
Alyssa Rosenzweig
cb7348eac0 asahi: don't leak blit shaders 
valgrind

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:07 -04:00
Alyssa Rosenzweig
6a27a3838c asahi: assert guard previously-subtle code 
would've caught the bug in the previous patch.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:07 -04:00
Alyssa Rosenzweig
09fde905a0 asahi: fix extremely subtle UAF 
we can get into weird situations and the clever logic isn't worth it. do
unclever logic instead and fix subtle CTS flakes. GL was a mistake.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:07 -04:00
Alyssa Rosenzweig
72a05db7ea asahi: fix indirect buffer UAF 
discovered by disabling BO cache, was working by chance before!

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:07 -04:00
Alyssa Rosenzweig
08b2e10dee asahi: add the BO even for internal dispatches 
whoops.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:07 -04:00
Alyssa Rosenzweig
fc6cf2020f asahi: use indirect ptr helper more 
ergonomic

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:07 -04:00
Alyssa Rosenzweig
28b7ad6e27 asahi: pad blit key 
avoid uninit read.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:07 -04:00
Alyssa Rosenzweig
8937277143 asahi: drop unused 
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:07 -04:00
Alyssa Rosenzweig
b40fd95eee asahi/clc: strip nir 
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:07 -04:00
Alyssa Rosenzweig
e86a35dad2 libagx: always tessellate clockwise 
easy enough to flip later in the pipeline instead and reduce significantly the
tessellator variants.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:07 -04:00
Alyssa Rosenzweig
87e6324459 libagx: make points mode dynamic 
it's a cold enough path.

16 to 10 tessellator variants.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:07 -04:00
Alyssa Rosenzweig
a7843643c6 libagx: drop generated VDM tess path (for now?) 
this was definitely the coolest thing I did in my career. but it has a lot of
drawbacks:

* complexity across the whole stack.
* perf #s even for synthetic tess workloads are... lackluster. couple % on
  terraintess. and games tend not to be tess heavy (and if they are we're
  screwed and this won't save us.) ironically, sascha willem's non-terrain tess
  is sped up a ton by the prefix sum path, so.. lol?
* more brittle for (M3?) porting.
* makes it harder to make the tessellator common code.
* doesn't play nice with the indirect path.
* pile of extra tessellator variants.
* harder to test, coverage is already not great here.

so... drop it, for now. the code isn't gone, and the idea may come back in a
future iteration, perhaps based on mesh shaders. but in its current form I don't think it's worth keeping right now.

my main resevation is actually about heap usage from doubling the index buffer
size. hopefully this is tolerable in practice.

this gets us from 24 to 16 tessellator variants.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:07 -04:00
Alyssa Rosenzweig
02e29bdea4 libagx: don't rely on loop unroll in txs 
silly, should reduce memory footprint.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:07 -04:00
Alyssa Rosenzweig
4607e0bf31 libagx: fix missing statics 
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:07 -04:00
Alyssa Rosenzweig
ddf2f2f5b1 hk: make tess partitioning dynamic 
no discernible difference in perf in terraintessellation. reduces tessellator
variants from 72 to 24.

before:
SHADER-DB:  - MESA_SHADER_COMPUTE shader: 2966 inst, 2310 alu, 2310 fscib, 1216 ic, 23148 bytes, 239 regs, 180 uniforms, 0 scratch, 384 threads, 17 loops, 0:0 spills:fills

after:
SHADER-DB:  - MESA_SHADER_COMPUTE shader: 3011 inst, 2343 alu, 2343 fscib, 1264 ic, 23508 bytes, 235 regs, 188 uniforms, 0 scratch, 384 threads, 17 loops, 0:0 spills:fills

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:07 -04:00
Alyssa Rosenzweig
d3d22039e1 hk: allow tess modes in either stage 
this makes shader objects more flexible.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:07 -04:00
Alyssa Rosenzweig
9bbe93d158 hk: fix alpha-to-coverage with sample shading 
fixes sascha willem's deferredmultisampling.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:07 -04:00
Alyssa Rosenzweig
d0b3b4c309 agx: move binary_size into info 
this simplifies serialization, and will simplify future work.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:07 -04:00
Alyssa Rosenzweig
6c5be08269 agx: pack agx_cf_binding 
dramatically reduces shader info size.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:07 -04:00
Alyssa Rosenzweig
5d628e3892 agx: fix uniform packing with local_load 
fixes with constant promotion:

dEQP-VK.compute.shader_object_spirv.workgroup_memory_explicit_layout.alias.i32_to_u16_array_scalar_func_read_write_barrier

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:07 -04:00
Alyssa Rosenzweig
773dd89fde asahi: fix no16 flag 
regressed when shuffling code for hk.

Fixes: 3cb8c1de81 ("asahi: get debug in common")
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:07 -04:00
Asahi Lina
02169e76dd agx: Fix queue destroy op for virtgpu 
Signed-off-by: Asahi Lina <lina@asahilina.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:07 -04:00
Alyssa Rosenzweig
12ab3abaac hk: bump max push size 
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:07 -04:00
Alyssa Rosenzweig
6dad37c812 hk: use push size macro 
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:07 -04:00
Alyssa Rosenzweig
fe84aaa8a7 hk: switch to 64-bit queries 
for timestamps

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:07 -04:00
Alyssa Rosenzweig
b1b125dbe6 hk: drop store_op_dontcare w/a 
updated CTS fixes this.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:07 -04:00
Alyssa Rosenzweig
3ee036b2b5 hk: wire up indirect tess 
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:07 -04:00
Alyssa Rosenzweig
a4e3ca1fc5 hk: add mechanism to test indirects 
while we wait for cts.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:07 -04:00
Alyssa Rosenzweig
b8ae31948d hk: plumb indirect_local 
for tess.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:07 -04:00
Alyssa Rosenzweig
e68aed3a13 libagx: extend indirect tess to handle indexed 
this was missed, vk cts hits it

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:07 -04:00
Alyssa Rosenzweig
5e464521fb asahi: presumably fix indirect indexed tess 
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:07 -04:00
Mary Guillemard
053d4c5666 hk: Fill deviceUUID 
Before this the device UUID was never filled.

Fixes: 5bc8284816 ("hk: add Vulkan driver for Apple GPUs")
Signed-off-by: Mary Guillemard <mary@mary.zone>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:07 -04:00
Sergio Lopez
167744dce9 hk: allow overriding sysmem with an env var 
When running in a VM, the amount of RAM seen through /proc/meminfo
doesn't necessarily reflect the real amount of memory available in the
system to be used as VRAM. For instance, on a system with 8 GB of
physical RAM, the user might want to run a VM with 2 GB of RAM and still
be able to use 4 GB as VRAM.

To achieve this, allow the reported sysmem to be overridden by the
value (in bytes) of the HK_SYSMEM environment variable.

Signed-off-by: Sergio Lopez <slp@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:07 -04:00
Sergio Lopez
2c18b0c6aa hk: limit the number of free BOs in each cmd pool 
Some clients like to have long living command pools, each one with
multiple command buffers, and trigger bursts of BO allocations on them.
This leads to the an ever increasing number of BOs allocated on each
command pool, easily reaching over a couple thousand BOs per pool.

The visible effect is an ever increasing amount of VRAM consumption,
which has been observed to easily get over 700MB for a workload
consuming around 4GB of heap.

Here we limit the number of free BOs in each pool to 32. The rest are
dereferenced when the command buffer is reset. In practice, with
workloads triggering bursts of BO allocations on command buffers, those
BOs aren't really released, keeping them in the BO cache and getting
reused shortly after. This has the nice size effect of doing bookkeeping
on the BO cache, trimming it up when needed.

With this change, the tested workloads present a significant reduction
of VRAM, combined with an stabilization of BO allocations.

Signed-off-by: Sergio Lopez <slp@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:07 -04:00
Alyssa Rosenzweig
2c513e8689 agx: consistent ffma name 
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:07 -04:00
Alyssa Rosenzweig
7e9d468891 agx: encoding_32 -> encoding 
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:07 -04:00
Alyssa Rosenzweig
22a8a0fe6d agx: drop encoding_16 
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:07 -04:00
Alyssa Rosenzweig
ce02690902 agx: special case mov_imm 
so we can rm a worse special case

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:07 -04:00
Alyssa Rosenzweig
11f797d93a agx: split up f16 ops 
they're fundamentally different instructions, model that appropriately.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:07 -04:00
Alyssa Rosenzweig
72b533f5c7 asahi,agx: clang-format 
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:07 -04:00
Alyssa Rosenzweig
7ca27ebd46 asahi: rm tessellator branch 
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:06 -04:00
Alyssa Rosenzweig
ea18348d94 asahi: defer fixed->float to tess 
this should parallelize better.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:06 -04:00
Alyssa Rosenzweig
3f6ff1b11e asahi: don't do extra runtime validation for exact 
silly.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:06 -04:00
Alyssa Rosenzweig
cc8d50baf5 asahi: make wrapping less silly 
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:06 -04:00
Alyssa Rosenzweig
c8a1c82175 asahi: get rid of default from the XML 
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:06 -04:00
Alyssa Rosenzweig
1a2938704d asahi: drop implied defaults 
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:06 -04:00
Alyssa Rosenzweig
f3612689d9 asahi: automatically set default to encode to zero 
even with mods.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31908>
 2024-10-30 10:14:06 -04:00