e16918df03
nvk: Implement vkCmdDraw[Indexed]Indirect()
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:59 +00:00
9ceec19adb
nvk: Use MME for vkCmdDraw[Indexed]()
...
This lets us move the instance loop to the MME and stop emitting state
per-instance.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:59 +00:00
02783c822c
nouveau/mme: Add helper macros for setting fields
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:59 +00:00
d6ac9dcac7
nvk: Use MME for clears
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:59 +00:00
d753d0a809
nvk: Add MME infrastructure
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:59 +00:00
f50e7cf015
nouveau/mme: Add unit tests for the Turing+ MME simulator
...
These verify that its behavior is the same as the hardware. Very useful
for reverse engineering.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:59 +00:00
e4b936d817
nouveau/mme: Make the winsys headers C++ safe
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:59 +00:00
58b9108468
nouveau/mme: Add an isaspec-based dumper
...
This actually dumps everything, unlike the pretty printer which tries to
dump something readable and may ignore bits of instructions if it thinks
they're no-ops. Useful for debugging the pretty printer.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:59 +00:00
4a5c0b686a
nouveau/mme: Add a tiny simulator for the Turing+ MME
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:59 +00:00
51e8194736
nouveau/mme: Add a builder for the Turing+ MME
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:59 +00:00
e9170ef452
nouveau/mme: Add an assembler and disassembler for the Turring+ MME
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:59 +00:00
01d0c78386
nouveau/mme: Add isaspec XML for the Turing+ MME
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:59 +00:00
af961f9aec
nouveau/mme: Add basic structures for the Turing+ MME
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:59 +00:00
a6c62732da
nouveau/parser: Add more arrays
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:59 +00:00
9db0c9fb2d
nouveau/push: Fix a void pointer arithmetic bug
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:59 +00:00
33ed4b9bc5
nil: Don't claim texture support for 2-bit SNORM
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:59 +00:00
1e3e00dad7
nvk: Increase the graphics pipeline push space
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:59 +00:00
82fc528773
nvk: Use VK_IMAGE_USAGE_*_ATTACHMENT_BIT for image clears
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:59 +00:00
46f54d53e8
nvk: Drop a TODO
...
Dynamic buffer bindings work fine
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:59 +00:00
8825d8effe
nvk: Use f for extension features
...
It's shorter and more ergonamic and the driver is still small so we can
pick a reasonable convention before things get out-of-hand.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:59 +00:00
e96c3cb965
nvk: Fix unused result warnings in pushbuf resets.
...
We know we aren't hitting the resize paths here, so errors can't be
thrown.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:59 +00:00
2773e55e86
nvk: Fix uninitialized result usage in NVK_DEBUG_ZERO_MEMORY.
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:59 +00:00
5d0aa694ee
nvk: Avoid strict aliasing warning in the pushbuffer encoding.
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:59 +00:00
5c853ed300
nvk: fix zero page refing
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:59 +00:00
7f413ef104
nouveau/parser: Put the dump helpers in C files
...
This should make the build faster since we don't have to compile those
giant functions with giant switch statements constantly.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:59 +00:00
7e1291d7be
nouveau/parser: Use more idiomatic python
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:59 +00:00
9855b80a7a
nouveau/parser: Convert to mako
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:59 +00:00
309c6cfd99
nvk: alloc a zero page and use it for vertex runouts
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:59 +00:00
c87a7b87fa
nvk: Initialize fixed draw/default state once
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:59 +00:00
daa6f9c700
nvk: Start every command buffer with a nop
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:59 +00:00
7cdb86a843
nvk: Split nvk_queue into its own file
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:58 +00:00
adb71cace7
nvk: fix mem leaks
...
device would leak in various situations.
Separate out lables were needed in the earlier
versions of this function but now we can just
use a single label
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:58 +00:00
c9f0ec90fc
nvk: delete commented code
...
nvk_CmdClearColorImage first lived in nvk_cmd_copy,
then moved to nvk_cmd_meta, and finally in nvk_cmd_clear
I seems that this was left behind in one of the moves
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:58 +00:00
c086d60466
nvk: fix implicit-fallthrough warnings with clang
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:58 +00:00
6bfecf8448
nvk: use common physical device enumeration
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:58 +00:00
3f16c30516
nvk: fix OOB read inside nvk_get_va_format
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:58 +00:00
6ba98ca4a1
nvk: Update to the new command buffer infrastructure
...
We leave nvk_command_pool because we'll want to use it for recycling
individual pushbufs. Drop reset_on_submit becuse there was never really
that much point in that micro-optimization and it isn't thread-safe
anyway if we recycle things in the pool.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:58 +00:00
fa8be32aec
nouveau/ws: push chaining
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:58 +00:00
3d5b9d8b78
nouveau/ws: rework refing push buffer bos
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:58 +00:00
c791fae67b
nouveau/ws: bound check nouveau_ws_push_append
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:58 +00:00
6f702f9939
nvk: Add some asserts for nv50 compiler image restrictions
...
These just ensure that we assert instead of doing something that the
nv50 compiler is likely to screw up.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:58 +00:00
c36ee41fdb
nvk: Zero unused descriptors
...
When a descriptor gets freed because a buffer/image veiw or sampler is
deleted, zero out the descriptor memory. This prevents the descriptor
from being used later and faulting on possibly non-existant memory.
While we're here, clean up the function a bit.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:58 +00:00
67a45d1c5f
nvk: Implement descriptor table growing
...
Now that command buffers are no longer referencing descriptor table BOs
directly and we're not handing out maps, we can support growing.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:58 +00:00
cad1834ba4
nvk: Rework nvk_descriptor_table_add/remove
...
The new api returns a VkResult for more accurate error checking. More
importantly, it also does copies the descriptor data into the table
instead of returning a map. This means the users of the descriptor
table will never see a map and we can make sure that the copy happens
under the lock in case we have to grow. Since each descriptor is only
32 bytes, this isn't going to be a perf problem.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:58 +00:00
9e549ce08b
nvk/sampler: Free descriptor table entries
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:58 +00:00
b13faff71c
nvk: Invalidate texture header and sampler caches each submit
...
It's possible descriptors have been added or removed. We could track
it, but there's really no point. Per-submit shouldn't be terrible.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:58 +00:00
6935fe5c68
nvk: Rework TLS/SLM and image/sampler table handling
...
When Dave originally added SLM, he did so with a preamble pushbuf which
sets up the SLM area prior to executing any actual command buffers.
This has the advantage that the SLM BO is never directly referenced in
any command buffer and we can swap it out even after command buffers
have been built. When I added image/sampler tables, I sort-of went
along with it but then did something different for 3D.
Thanks to buffer reference counting, we can employ a fairly simple
locking scheme. Each of the globals: images, samplers, and TSL has a
lock which gets taken whenever the table is modified. When we go to do
a submit, we take each lock and take a queue-local reference to the BO
and grab a copy of its size parameter under the lock. If anything has
changed, we update the queue-local preamble pushbuf. Because the queue
also holds a reference, we can safely proceed to submit command buffers
which reference those global BOs without the individual global locks
taken. If SLM or one of the descriptor tables resizes while we're
mid-submit, our reference to the BO will remain valid until either the
destruction of the queue or the next submit.
The one small bit of threaded cleverness (one always has to be careful
with that) is that the SLM size can be checked without taking the lock
because the SLM size only ever increases.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:58 +00:00
e8183da71f
nvk: Drop most buffer tracking
...
If it's backed by an nvk_device_memory, we don't need to track it since
we always add all of those to every submit so that bindless works.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:58 +00:00
a43e3fc80b
nvk/dispatch: Rename push to p
...
I took to using this convention in nvk_cmd_draw.c because it's shorter.
Let's use it other places too.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:58 +00:00
26fc226a7e
nvk/blit: Rename push to p
...
I took to using this convention in nvk_cmd_draw.c because it's shorter.
Let's use it other places too.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24326 >
2023-08-04 21:31:58 +00:00
Faith Ekstrand