Erik Faye-Lund [Tue, 9 Jun 2020 19:54:23 +0000 (21:54 +0200)]
zink: only report device-local memory as video-memory
While the definition of "video memory" isn't super clear, I think it's
pretty reasonable to assume host-memory isn't meant to be included. So
let's only count dedicated memory here.
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/3107
Reviewed-by: Witold Baryluk <witold.baryluk@gmail.com>
Tested-by: Witold Baryluk <witold.baryluk@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5409>
Samuel Pitoiset [Tue, 9 Jun 2020 06:36:17 +0000 (08:36 +0200)]
ac/nir: fix integer comparisons with pointers
If we get a comparison between a pointer and an integer, LLVM
complains if the operands aren't of the same type.
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/3085
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5397>
Pierre-Eric Pelloux-Prayer [Tue, 9 Jun 2020 10:24:41 +0000 (12:24 +0200)]
radeonsi/ngg: try GS multi-cycling mode if default mode failed
If gsprim_lds_size is larger than target_lds_size then gfx10_ngg_calculate_subgroup_info
will fail.
This commit adds a logic to try the multi-cycling in this case because it's
using less memory.
This fix glsl-1.50-gs-max-output when using NGG.
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5401>
Pierre-Eric Pelloux-Prayer [Tue, 9 Jun 2020 10:23:04 +0000 (12:23 +0200)]
radeonsi: add return value to gfx10_ngg_calculate_subgroup_info
gfx10_ngg_calculate_subgroup_info uses assert to detect invalid configuration,
but if asserts are disabled it will continue its execution.
This commits adds a boolean return value to let the caller know that something
went wrong and that the results mustn't be used.
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/3103
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5401>
Andrii Simiklit [Wed, 3 Jun 2020 15:59:02 +0000 (18:59 +0300)]
glsl: fix crash on glsl macro redefinition
In case shader contains two equal macro defines, first one with trailing spaces
and the second one without.
`#define A 1 `
`#define A 1`
The parser crashes
Fixes: 0346ad37741b11d640c1c4970b275c1f0c7f9e75 ("glsl: ignore trailing whitespace when define redefined")
Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Signed-off-by: Andrii Simiklit <andrii.simiklit@globallogic.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5312>
Jason Ekstrand [Tue, 9 Jun 2020 01:37:51 +0000 (20:37 -0500)]
anv/allocator: Compare to start_offset in state_pool_free_no_vg
In
d11e4738a86ec, we started using a start_offset to allow us to
allocate pools where the base address isn't at the start of the pool.
This is useful for binding table pools which want to be relative to
surface state base address (more or less), among other things. However,
we had a bug where, if you have a negative offset, everything returned
to the pool would end up being returned to the "back" of the pool. This
isn't what we want for binding tables in the softpin world. This was
causing us to never actually re-use any binding table blocks. How this
passed CTS, I have no idea.
Closes: #3100
Fixes: d11e4738a86ec "anv/allocator: Add a start_offset to anv_state_pool"
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5395>
Alyssa Rosenzweig [Tue, 9 Jun 2020 20:04:37 +0000 (16:04 -0400)]
panfrost: Ensure we have ro before using it
Even through the resouce requested has a BIND_SCANOUT or related tag,
this does not mean that we have a render-only driver.
This can trivially happen as one requests such resource from GBM, while
using the panfrost fd (and hence panfrost_dri.so)
Forward port of !3000
Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Signed-off-by: Robert Foss <robert.foss@collabora.com>
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Closes: #2664
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5410>
Samuel Pitoiset [Thu, 4 Jun 2020 09:33:28 +0000 (11:33 +0200)]
radv/aco: enable shaderInt8 and VK_KHR_shader_float16_int8 on GFX6-GFX7
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5226>
Samuel Pitoiset [Thu, 7 May 2020 08:54:12 +0000 (10:54 +0200)]
radv/aco: enable shaderInt16 on GFX6-GFX7
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5226>
Samuel Pitoiset [Tue, 5 May 2020 07:07:33 +0000 (09:07 +0200)]
radv/aco: enable 8-bit/16-bit storage on GFX6-GFX7
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5226>
Daniel Schürmann [Mon, 25 May 2020 09:51:27 +0000 (11:51 +0200)]
aco: remove unnecessary split- and create_vector instructions for subdword loads
This helps GFX6/7 by removing unnecessary shuffle code.
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5226>
Samuel Pitoiset [Mon, 25 May 2020 16:33:18 +0000 (18:33 +0200)]
aco: fix alignment of vectors with 4 elements
I think this case was just missing.
This fixes a bunch of 16-bit storage related CTS failures like
dEQP-VK.ssbo.phys.layout.single_basic_type.std430.u16vec4.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5226>
Samuel Pitoiset [Thu, 7 May 2020 08:55:28 +0000 (10:55 +0200)]
aco: implement 8-bit/16-bit conversions on GFX6-GFX7
Use v_bfe to implement small bitsize conversions because the
compiler probably optimizes this better.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5226>
Daniel Schürmann [Mon, 11 May 2020 15:42:37 +0000 (16:42 +0100)]
aco: optimize packing of 16bit subdword registers on GFX6/7
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5226>
Daniel Schürmann [Fri, 5 Jun 2020 20:21:02 +0000 (21:21 +0100)]
aco: skip partial copies on first iteration when lowering to hw
Helps some Detroit : Become Human shaders.
Totals from affected shaders: (VEGA)
Code Size:
47693912 ->
47670212 (-0.05 %) bytes
Instructions:
9183788 ->
9177863 (-0.06 %)
Copies: 910052 -> 904127 (-0.65 %)
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5226>
Daniel Schürmann [Thu, 7 May 2020 17:15:59 +0000 (18:15 +0100)]
aco: coalesce copies more aggressively when lowering to hw
Helps some Detroit : Become Human shaders.
Totals from affected shaders: (VEGA)
Code Size:
9880420 ->
9879088 (-0.01 %) bytes
Instructions:
1918553 ->
1918220 (-0.02 %)
Copies: 177783 -> 177450 (-0.19 %)
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5226>
Daniel Schürmann [Wed, 27 May 2020 17:31:33 +0000 (18:31 +0100)]
aco: add and use scratch SGPR to lower subdword p_create_vector on GFX6/7
This is needed to lower some corner cases correctly,
in case the same operand occurs multiple times:
e.g. v0 = p_create_vector(v0[0:8], v0[0:8], v0[0:8], v0[0:8])
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5226>
Daniel Schürmann [Wed, 27 May 2020 10:08:31 +0000 (11:08 +0100)]
aco: adjust GFX6 subdword lowering workarounds for 8bit
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5226>
Daniel Schürmann [Sat, 16 May 2020 16:30:21 +0000 (17:30 +0100)]
aco: Workarounds subdword lowering on GFX6/7
As there are no SDWA instructions, we need to take care not to overwrite
the upper bits of other copy_operation's operands.
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5226>
Daniel Schürmann [Wed, 6 May 2020 10:58:02 +0000 (11:58 +0100)]
aco: use full-register instructions to implement subdword packing on GFX6/7
On GFX6/7, there are no SDWA instructions.
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5226>
Daniel Schürmann [Fri, 5 Jun 2020 20:05:31 +0000 (21:05 +0100)]
aco: simplify statistics collection for copies
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5226>
Daniel Schürmann [Mon, 25 May 2020 20:31:38 +0000 (21:31 +0100)]
aco: fix register assignment for p_create_vector on GFX6/7
In case, some operand was already placed in the definition space,
it could happen that it wasn't considered for live-range splits.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5226>
Mike Blumenkrantz [Mon, 8 Jun 2020 13:55:19 +0000 (09:55 -0400)]
zink: emit interpolation decorations for ntv outputs
this matches up with nir internal states pre/post ntv
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5384>
Mike Blumenkrantz [Wed, 3 Jun 2020 13:38:57 +0000 (09:38 -0400)]
zink: track program usages for each shader
when shaders are created and destroyed in large numbers, the same pointers
get reused for different shaders, which can lead to bad lookups in the
program_cache hash table.
now each shader tracks its program usage to automatically remove itself from
that program in order to avoid hash collisions
fixes mesa/mesa#3053
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5315>
Erik Faye-Lund [Tue, 9 Jun 2020 19:08:05 +0000 (21:08 +0200)]
zink: assert that image-view format isn't undefined
Reviewed-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5407>
Erik Faye-Lund [Tue, 9 Jun 2020 19:07:48 +0000 (21:07 +0200)]
zink: emulate B8G8R8X8_SRGB with B8G8R8A8_SRGB
Reviewed-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5407>
Eric Anholt [Tue, 2 Jun 2020 21:21:30 +0000 (14:21 -0700)]
turnip: Expose robustBufferAccess.
It is a required device feature, and all enabled tests in
dEQP-VK.robustness.* pass.
Reviewed-by: Jonathan Marek <jonathan@marek.ca>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5266>
Eric Anholt [Tue, 2 Jun 2020 23:14:36 +0000 (16:14 -0700)]
ci: Use rsync for initial nfsroot population on cheza.
rm -rf and then copying over all the contents again is a waste of time
when we'll almost always be using the same rootfs. Saves about 30s of job
time.
Closes: #3065
Reviewed-by: Christian Gmeiner <christian.gmeiner@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5266>
Eric Anholt [Mon, 25 May 2020 19:57:25 +0000 (12:57 -0700)]
ci: Enable pre-merge fractional vulkan CTS runs on the turnip driver.
Test 1/50th of the CTS on a630 pre-merge, since we've got hardware that
can do it and infrastructure that should handle instability with a
less-mature driver.
Reviewed-by: Christian Gmeiner <christian.gmeiner@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5266>
Eric Anholt [Thu, 28 May 2020 19:50:51 +0000 (12:50 -0700)]
ci: Build the full VK CTS for baremetal testing.
I'm going to enable the VK CTS on cheza, so swap the deqp we have in the
container. build-deqp-vk already included GLES deqp binaries and data,
and is a newer branch than the last opengl-es-cts tag.
This brings a few things back over from build-deqp-gl for testlog
extraction, and copyes out the GLES mustpass lists.
Reviewed-by: Christian Gmeiner <christian.gmeiner@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5266>
Eric Anholt [Thu, 28 May 2020 19:37:29 +0000 (12:37 -0700)]
ci: Disable shader cache on vulkan CI runs.
I found it to be flaky in freedreno CI, and tracked down the issue to
parallel-deqp-runner needing to manage the shader cache
(https://gitlab.freedesktop.org/mesa/parallel-deqp-runner/-/merge_requests/13).
Until we fix that in the runner, disable it. This should matter less now
that we prebuild the SPIRV, though.
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Christian Gmeiner <christian.gmeiner@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5266>
Eric Anholt [Thu, 28 May 2020 19:33:37 +0000 (12:33 -0700)]
ci: Bump up to the current version of the VK CTS.
For enabling VK CTS on freedreno, I've heard there were important
stability fixes in the CTS recently.
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5266>
Eric Anholt [Tue, 9 Jun 2020 18:03:12 +0000 (11:03 -0700)]
turnip: Fix crashes in compute with no descriptors to load.
Found when trying to rebase cheza VK CI on top of this change.
Fixes: 334204823eee ("tu: Fix context faults loading unused descriptor sets")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5266>
Thong Thai [Mon, 8 Jun 2020 14:29:01 +0000 (10:29 -0400)]
frontends/vdpau: Default destination rect to source rect
mpv is passing in a NULL destination_video_rect, which results in a
black screen when playing videos using VDPAU in some cases.
Signed-off-by: Thong Thai <thong.thai@amd.com>
Reviewed-by: Leo Liu <leo.liu@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5386>
Marek Olšák [Tue, 9 Jun 2020 05:49:02 +0000 (01:49 -0400)]
radeonsi: require LLVM 11 for gfx10.3
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5383>
Marek Olšák [Fri, 27 Mar 2020 02:08:18 +0000 (22:08 -0400)]
radeonsi: add support for Sienna Cichlid
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5383>
Marek Olšák [Wed, 29 Jan 2020 04:35:49 +0000 (23:35 -0500)]
ac: align num_vgprs for gfx10.3
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5383>
Marek Olšák [Thu, 9 Apr 2020 16:25:49 +0000 (12:25 -0400)]
radeonsi: don't set any XNACK options on gfx10.3
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5383>
Marek Olšák [Thu, 12 Mar 2020 01:26:54 +0000 (21:26 -0400)]
radeonsi: set BIG_PAGE fields on gfx10.3
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5383>
Marek Olšák [Fri, 20 Mar 2020 01:37:15 +0000 (21:37 -0400)]
radeonsi: move L2_CACHE_CONTROL registers into si_emit_framebuffer_state
the next commit will set more fields.
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5383>
Marek Olšák [Tue, 23 Jul 2019 04:36:05 +0000 (00:36 -0400)]
radeonsi: implement R9G9B9E5 render target and image store support on gfx10.3
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5383>
Marek Olšák [Tue, 3 Sep 2019 21:54:47 +0000 (17:54 -0400)]
radeonsi: enable larger SDMA clears and copies on gfx10.3
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5383>
Marek Olšák [Thu, 26 Sep 2019 20:59:23 +0000 (16:59 -0400)]
radeonsi: honor a user-specified pitch on gfx10.3
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5383>
Marek Olšák [Sat, 18 Apr 2020 00:44:39 +0000 (20:44 -0400)]
ac/surface: add displayable DCC code for gfx10.3
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5383>
Marek Olšák [Fri, 27 Mar 2020 02:02:13 +0000 (22:02 -0400)]
ac,radeonsi: start adding support for gfx10.3
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5383>
Marek Olšák [Fri, 27 Mar 2020 02:01:33 +0000 (22:01 -0400)]
ac,radeonsi: replace == GFX10 with >= GFX10 where it's needed
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5383>
Marek Olšák [Wed, 6 May 2020 19:18:25 +0000 (15:18 -0400)]
radeonsi: enable ARB_sparse_buffer
This seems to be working now, but it wasn't working before.
I don't know what fixed this. Tested on Raven and Navi14.
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5402>
Connor Abbott [Tue, 9 Jun 2020 12:40:58 +0000 (14:40 +0200)]
tu: Fix context faults loading unused descriptor sets
The app is allowed to never bind descriptor sets that are statically
unused by the pipeline, which would've caused a context fault since
CP_LOAD_STATE6 would try to load the descriptors that don't exist. Fix
this by not preloading descriptors from unused descriptor sets. We could
do more fine-grained accounting of which descriptors are used, but this
is enough to fix the problem.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5400>
Danylo Piliaiev [Mon, 25 May 2020 16:07:08 +0000 (19:07 +0300)]
i965: Work around incorrect usage of glDrawRangeElements in UE4
Unreal Engine 4 has a bug in usage of glDrawRangeElements,
causing it to be called with a number of vertices in place
of "end" parameter (which specifies the maximum array index
contained in indices).
Since there is unknown amount of games affected and we
could not identify that a game is built with UE4 - we are
forced to make a blanket workaround, disregarding max_index
in range calculations. Fortunately all such calls look like:
glDrawRangeElements(GL_TRIANGLES, 0, 3, 3, ...);
So we are able to narrow down this workaround.
This was uncovered after
b684030c3a656ffdbc93581856034e0982db46fd
broke a bunch of UE4 games.
Cc: 20.1 <mesa-stable@lists.freedesktop.org>
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/2917
Signed-off-by: Danylo Piliaiev <danylo.piliaiev@globallogic.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5203>
Connor Abbott [Thu, 2 Apr 2020 15:48:19 +0000 (17:48 +0200)]
tu: Rewrite flushing to use barriers
Replace the various ad-hoc flushes that we've inserted, copied from
freedreno, etc. with a unified system that uses the user-supplied
information via vkCmdPipelineBarrier() and subpass dependencies.
There are a few notable differences in behavior:
- We now move setting RB_CCU_CNTL up a little in the gmem case, but
hopefully that won't matter too much. This matches what the Vulkan blob
does.
- We properly implement delayed setting of events, completing our
implementaton of events.
- Finally, of course, we should be a lot less flush-happy. We won't emit
useless CCU/cache flushes with multiple copies, renderpasses, etc. that
don't depend on each other, and also won't flush/invalidate the cache
around renderpasses unless we actually need to.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4964>
Connor Abbott [Thu, 2 Apr 2020 14:55:00 +0000 (16:55 +0200)]
tu: Remove useless event_write helpers
tu6_emit_cache_flush() was wrongly named, and with the removal of the
last parameter both are useless.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4964>
Connor Abbott [Tue, 9 Jun 2020 13:30:43 +0000 (15:30 +0200)]
tu: Don't actually track seqno's for events
We just dropped the last user which actually cared about the seqno.
This never worked anyway, since the seqno was never reset between
multiple executions of the same command buffer. Turn the part of the
control buffer which used to track the seqno into a dummy dword, and
figure out automatically whether we need to include it. We will
implement seqnos again eventually, with timline semaphores, but that
will likely be totally different.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4964>
Connor Abbott [Tue, 9 Jun 2020 13:25:40 +0000 (15:25 +0200)]
tu: Remove useless post-binning flushes
The Vulkan blob doesn't do this, and based on my understanding of how
the blob works this is unnecessary. CACHE_FLUSH is already serialized
against all 3d commands so you don't need to wait for rendering commands
to finish before issuing it, and the subsequent wfi + WAIT_FOR_ME will
cause the CP to wait for the CACHE_FLUSH to finish, so there's also no
need to wait for it to complete. The CACHE_INVALIDATE also seems
unnecessary, and also isn't done by the blob.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4964>
Icecream95 [Sat, 4 Apr 2020 23:49:07 +0000 (11:49 +1200)]
panfrost: Mark PIPE_BUFFER BOs as not renderable
Without this, memory usage explodes by 16x due to height alignment.
Closes: #2715
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4451>
Pierre-Eric Pelloux-Prayer [Fri, 5 Jun 2020 12:40:01 +0000 (14:40 +0200)]
winsys/radeon: do not cast bo->va as void*
Using a util_hash_table_create_ptr_keys to store bo->va address doesn't
work on 32 bits.
This commit makes radeon_drm_winsys::bo_vas a hash_table_u64 instead.
Tested by Miklós Máté.
CC: 20.1 <mesa-stable@lists.freedesktop.org>
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/3056
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5380>
Christian Gmeiner [Sun, 7 Jun 2020 21:19:50 +0000 (23:19 +0200)]
ci: use separate docker images for baremetal builds
Using arm_test-base as a separate base layer as well for storage &
network bandwidth efficiency.
Signed-off-by: Christian Gmeiner <christian.gmeiner@gmail.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5381>
Christian Gmeiner [Sun, 7 Jun 2020 20:03:34 +0000 (22:03 +0200)]
ci: add arm_test-base docker image
Similar to x86_build-base.
Signed-off-by: Christian Gmeiner <christian.gmeiner@gmail.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5381>
Samuel Pitoiset [Fri, 5 Jun 2020 13:58:36 +0000 (15:58 +0200)]
radv/llvm: expose VK_EXT_shader_demote_to_helper_invocation with LLVM 9+
It should already work with the LLVM backend.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5361>
Marek Olšák [Mon, 8 Jun 2020 12:46:10 +0000 (08:46 -0400)]
glthread: sync in glFlush for multiple contexts
See the code comment.
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5382>
Marek Olšák [Mon, 1 Jun 2020 19:56:12 +0000 (15:56 -0400)]
gallium/u_vbuf: add a faster path for uploading non-interleaved attribs
+1% higher FPS in torcs.
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5304>
Marek Olšák [Mon, 1 Jun 2020 19:49:36 +0000 (15:49 -0400)]
gallium/u_vbuf: get rid of some pointer dereferences
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5304>
Ben Skeggs [Sat, 6 Jun 2020 06:58:00 +0000 (16:58 +1000)]
nir: use bitfield_insert instead of bfi in nir_lower_double_ops
NVIDIA hardware doesn't have an equivilant to bfi, but we do already have
a lowering for bitfield_insert->bfi.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5373>
Jonathan Marek [Sat, 6 Jun 2020 17:06:34 +0000 (13:06 -0400)]
turnip: fix VFD_CONTROL for binning pass
Fixes some cases with TU_DEBUG=forcebin, specifically the failures in:
dEQP-VK.glsl.*_vertex
Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5370>
Jonathan Marek [Sat, 6 Jun 2020 15:38:18 +0000 (11:38 -0400)]
turnip: use common emit_xs_cntl to fill a6xx_sp_xs_ctrl_reg0
Note this changes the value of SP_GS_CTRL_REG0, by using FOUR_QUADS and
setting MERGEDREGS. ir3 expects MERGEDREGS, and using FOUR_QUADS instead
of TWO_QUADS doesn't seem to hurt.
Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5370>
Jonathan Marek [Sat, 6 Jun 2020 15:26:11 +0000 (11:26 -0400)]
turnip: fix HW binning with geometry shader
Fixes failures with TU_DEBUG=forcebin and geometry shaders, for example:
dEQP-VK.binding_model.*geometry*
dEQP-VK.transform_feedback.simple.query*
Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5370>
Jonathan Marek [Sat, 6 Jun 2020 15:21:58 +0000 (11:21 -0400)]
turnip: correctly emit non-binning vs in transform feedback case
The offset given to tu6_emit_shader_object was wrong, binning_vs_offset
should only be used when using the binning pass vs.
Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5370>
Jonathan Marek [Sat, 6 Jun 2020 15:16:16 +0000 (11:16 -0400)]
freedreno/a6xx: use nonbinning VS when GS is used
The current "ds = state->bs" seems broken, and the "vs = state->bs" is
unnecessary (already set above). Since it was added as part of a GS-related
patch, I think this is what was intended.
Note: tesselation disables GMEM rendering so we shouldn't have to worry
about hs/ds + binning interaction.
Fixes: 0eebedb6190fdab8956769 ("freedreno/a6xx: Emit program state for GS")
Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5370>
Jonathan Marek [Sun, 7 Jun 2020 02:08:41 +0000 (22:08 -0400)]
turnip: clamp sampler minLod/maxLod
Otherwise A6XX_TEX_SAMP_1_{MIN,MAX}_LOD silently overflows.
This fixes these tests:
dEQP-VK.texture.explicit_lod.2d.derivatives.*
Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5375>
Jonathan Marek [Sun, 7 Jun 2020 02:07:09 +0000 (22:07 -0400)]
turnip: update some properties based on blob driver
subTexelPrecisionBits/mipmapPrecisionBits change fixes some failures in:
dEQP-VK.texture.explicit_lod.*
Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5375>
Jonathan Marek [Sun, 7 Jun 2020 17:06:42 +0000 (13:06 -0400)]
turnip: move HLSQ_UPDATE_CNTL write to before xs config writes
This matches the blob and gallium driver more closely, and fixes a
rendering issue observed on a650.
Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5376>
Caio Marcelo de Oliveira Filho [Sat, 6 Jun 2020 06:05:42 +0000 (23:05 -0700)]
nir: Fix logic that ends combine barrier sequence
The combination must stop when we see a scoped barrier that have
execution scope, i.e. it has control barrier behavior. The code was
mistakenly looking at the wrong scope.
Fixes: 345b5847b42 ("nir: Replace the scoped_memory barrier by a scoped_barrier")
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5365>
Caio Marcelo de Oliveira Filho [Fri, 5 Jun 2020 23:04:13 +0000 (16:04 -0700)]
intel/fs: Add Fall-through comment
Just to clarify the missing break is intentional.
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5365>
Caio Marcelo de Oliveira Filho [Fri, 5 Jun 2020 22:58:32 +0000 (15:58 -0700)]
spirv: Memory semantics is optional for OpControlBarrier
Fixes: 3ed2123d77d ("spirv: Use scoped barriers for SpvOpControlBarrier")
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5365>
Caio Marcelo de Oliveira Filho [Fri, 5 Jun 2020 22:26:47 +0000 (15:26 -0700)]
nir: Fix printing execution scope of a scoped barrier
Fixes: 345b5847b42 ("nir: Replace the scoped_memory barrier by a scoped_barrier")
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5365>
Christian Gmeiner [Sat, 6 Jun 2020 10:28:13 +0000 (12:28 +0200)]
etnaviv: drop translate_blend(..)
PIPE_BLEND_* matches 1:1 the hardware defines.
Signed-off-by: Christian Gmeiner <christian.gmeiner@gmail.com>
Reviewed-by: Lucas Stach <l.stach@pengutronix.de>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4187>
Danylo Piliaiev [Fri, 29 May 2020 10:51:32 +0000 (13:51 +0300)]
glsl: inline functions with unsupported return type before converting to nir
glsl_to_nir doesn't expect non-vector/scalar return types in functions.
Fixes: 7e60d5a501f311fa5e7bca5335adc6013dc1aeb9
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/3058
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/3060
Signed-off-by: Danylo Piliaiev <danylo.piliaiev@globallogic.com>
Tested-by: Witold Baryluk <witold.baryluk@gmail.com>
Reviewed-by: Witold Baryluk <witold.baryluk@gmail.com>
Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5333>
Rhys Perry [Thu, 4 Jun 2020 13:36:00 +0000 (14:36 +0100)]
aco: use v_xor3_b32
fossil-db (Navi):
Totals from 334 (0.26% of 128321) affected shaders:
CodeSize:
3345532 ->
3345484 (-0.00%); split: -0.00%, +0.00%
Instrs: 624662 -> 622778 (-0.30%); split: -0.30%, +0.00%
Mostly affects some parallel-rdp shaders
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5357>
Rhys Perry [Thu, 4 Jun 2020 14:03:11 +0000 (15:03 +0100)]
ac/gpu_info, radv: set max_wave64_per_simd to 20 on GFX10
Fixes RADV max_waves reporting for GFX10
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5356>
Samuel Pitoiset [Wed, 3 Jun 2020 12:35:02 +0000 (14:35 +0200)]
nir/lower_explicit_io: fix NON_UNIFORM access for UBO loads
Make sure to propagate the NON_UNIFORM access for UBO loads, so
that non-uniform loads are correctly lowered.
Cc: <mesa-stable@lists.freedesktop.org>
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5311>
Arcady Goldmints-Orlov [Wed, 15 Apr 2020 21:55:53 +0000 (16:55 -0500)]
nir/spirv/glsl450: increase asin(x) precision
asin(x) is now implemented using a piecewise approximation, which
improves the precision for |x| < 0.5
Previously, we were using a polynomial approximation for both the
asin() and acos() functions. Unfortunately, for asin(), this polynomial
does not have enough precision to satisfy the Vulkan CTS requiremenents,
which define the asin() precision based on the precision of
atan2(x, sqrt(1.0 - x*x)). The piecewise approximation gives the needed
precision in the problematic range.
v2: Skip the piecewise approximation for acos
Closes: #1843
Acked-by: Francisco Jerez <currojerez@riseup.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3809>
Samuel Pitoiset [Thu, 4 Jun 2020 07:28:16 +0000 (09:28 +0200)]
ac/nir: adjust an assertion for D16 on GFX6-GFX7
16-bit types can be used with MUBUF on GFX6-GFX7.
Fixes: c3e0ba52a0a ("ac/nir: support 16-bit data in buffer_load_format opcodes")
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5325>
Peter Seiderer [Sun, 22 Mar 2020 12:03:00 +0000 (13:03 +0100)]
v3d_bufmgr: fix time_t printf
Fixes:
error: format ‘%ld’ expects argument of type ‘long int’, but argument 3 has type ‘time_t’ {aka ‘long long int’}
Signed-off-by: Peter Seiderer <ps.report@gmx.net>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4279>
Peter Seiderer [Sun, 22 Mar 2020 10:48:31 +0000 (11:48 +0100)]
pan_bo.h: add time.h include for time_t
Fixes:
../src/gallium/drivers/panfrost/pan_bo.h:93:9: error: unknown type name ‘time_t’
Signed-off-by: Peter Seiderer <ps.report@gmx.net>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4279>
Peter Seiderer [Sun, 22 Mar 2020 10:42:35 +0000 (11:42 +0100)]
vc4_bufmgr: fix time_t printf
Fixes:
error: format ‘%ld’ expects argument of type ‘long int’, but argument 3 has type ‘time_t’ {aka ‘long long int’}
Signed-off-by: Peter Seiderer <ps.report@gmx.net>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4279>
Timothy Arceri [Fri, 5 Jun 2020 03:57:40 +0000 (13:57 +1000)]
glsl: fix potential slow compile times for GLSLOptimizeConservatively
See code comment for full description of the change.
Fixes: 0a5018c1a483 ("mesa: add gl_constants::GLSLOptimizeConservatively")
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/3034
Tested-by: Witold Baryluk <witold.baryluk@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5346>
Charmaine Lee [Fri, 5 Jun 2020 00:46:33 +0000 (17:46 -0700)]
llvmpipe: do not enable tessellation shader without llvm coroutines support
Tessellation shader in llvmpipe depends on llvm coroutines support. So do not
advertise tessellation shader support in llvmpipe if GALLIVM_HAVE_CORO is FALSE.
This fixes assertion in LLVMTokenTypeInContext() running tessellation shader
tests with llvm version < 6.
Fixes: eb522717 "llvmpipe: add support for tessellation shaders"
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Neha Bhende <bhenden@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5366>
Marcin Ślusarz [Fri, 5 Jun 2020 09:52:42 +0000 (11:52 +0200)]
intel/compiler: fix Android build
Signed-off-by: Marcin Ślusarz <marcin.slusarz@intel.com>
Fixes: 689acc739899 ("intel/compiler: Extract control barriers from scoped barriers")
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/3087
Reviewed-by: Sagar Ghuge <sagar.ghuge@intel.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5354>
Vinson Lee [Fri, 5 Jun 2020 07:36:55 +0000 (00:36 -0700)]
intel/genxml: Migrate from deprecated xml.etree.ElementTree getchildren.
xml.etree.ElementTree getchildren was deprecated since Python 2.7 and
will be removed in Python 3.9.
Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Reviewed-by: Eric Engestrom <eric@engestrom.ch>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5348>
Mauro Rossi [Fri, 5 Jun 2020 20:33:48 +0000 (22:33 +0200)]
android: svga: fix build for GL4.1 support
Fixes the following building errors:
external/mesa/src/gallium/drivers/svga/svga_context.c:184: error: undefined reference to 'svga_init_ts_functions'
external/mesa/src/gallium/drivers/svga/svga_context.c:100: error: undefined reference to 'svga_cleanup_tcs_state'
out/target/product/x86_64/obj_x86/STATIC_LIBRARIES/libmesa_pipe_svga_intermediates/libmesa_pipe_svga.a(svga_state.o):svga_state.c:hw_draw_state_sm5: error: undefined reference to 'svga_hw_tes'
out/target/product/x86_64/obj_x86/STATIC_LIBRARIES/libmesa_pipe_svga_intermediates/libmesa_pipe_svga.a(svga_state.o):svga_state.c:hw_draw_state_sm5: error: undefined reference to 'svga_hw_tcs'
Fixes: ccb4ea5a "svga: Add GL4.1(compatibility profile) support in svga driver"
Signed-off-by: Mauro Rossi <issor.oruam@gmail.com>
Reviewed-by: Charmaine Lee <charmainel@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5364>
Mauro Rossi [Fri, 5 Jun 2020 20:37:38 +0000 (22:37 +0200)]
android: util: fix build for GL4.1 support
Fixes the following building errors:
external/mesa/src/gallium/drivers/svga/svga_state_tgsi_transform.c:154: error: undefined reference to 'tgsi_write_vpos'
external/mesa/src/gallium/drivers/svga/svga_state_tgsi_transform.c:201: error: undefined reference to 'tgsi_remove_dynamic_indexing'
Fixes: 48a7456f ("util: Add util functionality for GL4.1 support")
Signed-off-by: Mauro Rossi <issor.oruam@gmail.com>
Reviewed-by: Charmaine Lee <charmainel@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5364>
Vinson Lee [Fri, 5 Jun 2020 07:46:10 +0000 (00:46 -0700)]
Switch from cElementTree to ElementTree.
The xml.etree.cElementTree module will be removed in Python 3.9. Since
Python 3.3 the xml.etree.cElementTree module has been deprecated, the
xml.etree.ElementTree module uses a fast implementation whenever
available.
Builds using Python 2.7 can still work but with the slower
implementation.
Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Acked-by: Eric Engestrom <eric@engestrom.ch>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5349>
Icecream95 [Fri, 5 Jun 2020 04:52:23 +0000 (16:52 +1200)]
Revert "panfrost: Keep cached BOs mmap'd"
This reverts commit
794c239a990e08b5a42d42607e9c5e5b0921390c.
A kernel bug causes cached BOs to not be unmapped correctly,
triggering "bad page cache" kernel messages and causing short hangs.
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5355>
Icecream95 [Fri, 5 Jun 2020 08:17:27 +0000 (20:17 +1200)]
pan/midgard: Use a signed value for checking inline constants
Inline constants are sign extended, so we should use a int16_t instead
of an unsigned type.
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5351>
Eric Anholt [Fri, 29 May 2020 23:49:43 +0000 (16:49 -0700)]
freedreno/ir3: Handle cases where we decide not to lower UBO 0 loads.
We advertize 4096 vec4s of GL uniform storage, but the HW can only store
512 vec4s in the const buffer.
Closes: #3049
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5273>
Eric Anholt [Fri, 29 May 2020 23:35:43 +0000 (16:35 -0700)]
freedreno/ir3: Drop the max_const on a6xx to 512.
The GLES blob on the p3a limits constlen to 512 between VS and FS across
a6xx gpu ids (615, 630, 640, and 650). Experimentally, exceeding that
limit in any one stage results in rendering corruption or GPU hangs
(though my most detailed testing had a loop limit in a uniform, so that
may the cause of the hang). Clamp the limit we use inside of a shader so
we don't exceed it within a stage.
This commit doesn't resovle limiting inter-stage. Experimentally, I've
found that I can push up to a total of ~768 vec4s between VS and FS on
a630, with or without uniform updates between each draw. We'll need to do
some shader key-based limiting of constlen at draw time to respect that
limit, but that's left for future work, and this commit is enough for the
google earth case that initiated this work.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5273>
Eric Anholt [Fri, 29 May 2020 23:31:43 +0000 (16:31 -0700)]
freedreno/ir3: Account for driver params in UBO max const upload.
The const state setup needs to be able to push its driver params, so
account for them in the analyze_ubo_ranges.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5273>
Eric Anholt [Fri, 29 May 2020 17:21:02 +0000 (10:21 -0700)]
freedreno/ir3: Stop shifting UBO 1 down to be UBO 0.
It turns out the GL uniforms file is larger than the hardware constant
file, so we need to limit how many UBOs we lower to constbuf loads. To do
actual UBO loads, we'll need to be able to upload UBO 0's pointer or
descriptor.
No difference on nohw 1 UBO update drawoverhead case (n=35).
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5273>
Eric Anholt [Mon, 1 Jun 2020 18:40:36 +0000 (11:40 -0700)]
freedreno/ir3: Drop unnecessary alignment of pushed UBO size.
The analysis pass gives us vec4-aligned size, and all of our other
constbuf allocations here are in vec4 units, so we can just divide by 16.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5273>
Eric Anholt [Mon, 1 Jun 2020 18:32:04 +0000 (11:32 -0700)]
freedreno/ir3: Stop pushing immediates once we've filled the constbuf.
If we filled the constbuf up with UBOs, we may need to avoid generating
more immediate push constants.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5273>
Eric Anholt [Mon, 1 Jun 2020 18:20:51 +0000 (11:20 -0700)]
freedreno/ir3: Refactor ir3_cp's lower_immed().
There was duplicated handling in the callers that we can just move inside.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5273>
projects / mesa.git / log