Kenneth Graunke [Wed, 23 Mar 2016 18:56:39 +0000 (11:56 -0700)]
i965: Always use BRW_TEXCOORDMODE_CUBE when seamless filtering.
When using seamless cube map mode and NEAREST filtering, we explicitly
overrode the wrap modes to CLAMP_TO_EDGE. This was to implement the
following spec text:
"If NEAREST filtering is done within a miplevel, always apply apply
wrap mode CLAMP_TO_EDGE."
However, textureGather() ignores the sampler's filtering mode, and
instead returns the four pixels that would be blended by LINEAR
filtering. This implies that we should do proper seamless filtering,
and include pixels from adjacent cube faces.
It turns out that we can simply delete the NEAREST -> CLAMP_TO_EDGE
overrides. Normal cube map sampling works by first selecting the
face, and then nearest filtering fetches the closest texel. If the
nearest texel was on a different face, then that face would have been
chosen. So it should always be within the face anyway, which
effectively performs CLAMP_TO_EDGE.
Fixes 86 dEQP-GLES31.texture.gather.basic.cube.* tests.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Suggested-by: Ian Romanick <idr@freedesktop.org>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Kenneth Graunke [Fri, 25 Mar 2016 22:33:35 +0000 (15:33 -0700)]
i965: Fix brw_render_cache_set_check_flush's PIPE_CONTROLs.
Our driver uses the brw_render_cache mechanism to track buffers we've
rendered to and are about to sample from.
Previously, we did a single PIPE_CONTROL with the following bits set:
- Render Target Flush
- Depth Cache Flush
- Texture Cache Invalidate
- VF Cache Invalidate
- Instruction Cache Invalidate
- CS Stall
This combined both "top of pipe" invalidations and "bottom of pipe"
flushes, which isn't how the hardware is intended to be programmed.
The "top of pipe" invalidations may happen right away, without any
guarantees that rendering using those caches has completed. That
rendering may continue altering the caches. The "bottom of pipe"
flushes do wait for the rendering to complete. The CS stall also
prevents further work from happening until data is flushed out.
What we wanted to do was wait for rendering complete, flush the new
data out of the render and depth caches, wait, then invalidate any
stale data in read-only caches. We can accomplish this by doing the
"bottom of pipe" flushes with a CS stall, then the "top of pipe"
flushes as a second PIPE_CONTROL. The flushes will wait until the
rendering is complete, and the CS stall will prevent the second
PIPE_CONTROL with the invalidations from executing until the first
is done.
Fixes dEQP-GLES3.functional.texture.specification.teximage2d_pbo
subtests on Braswell and Skylake. These tests hit the meta PBO
texture upload path, which binds the PBO as a texture and samples
from it, while rendering to the destination texture. The tests
then sample from the texture.
For now, we leave Gen4-5 alone. It probably needs work too, but
apparently it hasn't even been setting the (G45+) TC invalidation
bit at all...
v2: Add Sandybridge post-sync non-zero workaround, for safety.
Cc: mesa-stable@lists.freedesktop.org
Suggested-by: Francisco Jerez <currojerez@riseup.net>
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
Kenneth Graunke [Thu, 24 Mar 2016 23:21:35 +0000 (16:21 -0700)]
i965: Whack UAV bit when FS discards and there are no color writes.
dEQP-GLES31.functional.fbo.no_attachments.* draws a quad with no
framebuffer attachments, using a shader that discards based on
gl_FragCoord. It uses occlusion queries to inspect whether pixels
are rendered or not.
Unfortunately, the hardware is not dispatching any pixel shaders,
so discards never happen, and the full quad of pixels increments
PS_DEPTH_COUNT, making the occlusion query results bogus.
To understand why, we have to delve into the WM_INT internal
signalling mechanism's formulas.
The "WM_INT::Pixel Shader Kill Pixel" signal is defined as:
3DSTATE_WM::ForceKillPixel == ON ||
(3DSTATE_WM::ForceKillPixel != Off &&
!WM_INT::WM_HZ_OP &&
3DSTATE_WM::EDSC_Mode != PREPS &&
(WM_INT::Depth Write Enable || WM_INT::Stencil Write Enable) &&
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
(3DSTATE_PS_EXTRA::PixelShaderKillsPixels ||
3DSTATE_PS_EXTRA:: oMask Present to RenderTarget ||
3DSTATE_PS_BLEND::AlphaToCoverageEnable ||
3DSTATE_PS_BLEND::AlphaTestEnable ||
3DSTATE_WM_CHROMAKEY::ChromaKeyKillEnable))
Because there is no depth or stencil buffer, writes to those buffers
are disabled. So the highlighted condition is false, making the whole
"Kill Pixel" condition false. This then feeds into the following
"WM_INT::ThreadDispatchEnable" condition:
3DSTATE_WM::ForceThreadDispatch != OFF &&
!WM_INT::WM_HZ_OP &&
3DSTATE_PS_EXTRA::PixelShaderValid &&
(3DSTATE_PS_EXTRA::PixelShaderHasUAV ||
WM_INT::Pixel Shader Kill Pixel ||
WM_INT::RTIndependentRasterizationEnable ||
(!3DSTATE_PS_EXTRA::PixelShaderDoesNotWriteRT &&
3DSTATE_PS_BLEND::HasWriteableRT) ||
(WM_INT::Pixel Shader Computed Depth Mode != PSCDEPTH_OFF &&
(WM_INT::Depth Test Enable || WM_INT::Depth Write Enable)) ||
(3DSTATE_PS_EXTRA::Computed Stencil && WM_INT::Stencil Test Enable) ||
(3DSTATE_WM::EDSC_Mode == 1 && (WM_INT::Depth Test Enable ||
WM_INT::Depth Write Enable ||
WM_INT::Stencil Test Enable)))
Given that there's no depth/stencil testing, no writeable render target,
and the hardware thinks kill pixel doesn't happen, all of these
conditions are false. We have to whack some bit to make PS invocations
happen. There are many options.
Curro suggested using the UAV bit. There's some precedence in doing
that - we set it for fragment shaders that do SSBO/image/atomic writes
when no color buffer writes are enabled. We can simply include discard
here too.
Fixes 64 dEQP-GLES31.functional.fbo.no_attachments.* tests.
v2: Add a comment suggested and written by Jason Ekstrand.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Rhys Kidd [Sat, 19 Mar 2016 22:37:57 +0000 (18:37 -0400)]
vc4: Remove unused include from vc4_nir_lower_txf_ms.c
Found with grep and inspection. Test compiled on RPi hw.
Assists any future effort to remove TGSI as an intermediate stage.
Signed-off-by: Rhys Kidd <rhyskidd@gmail.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
Adam Jackson [Thu, 24 Mar 2016 17:57:57 +0000 (13:57 -0400)]
glapi/glx: Treat xserver generated targets as .PHONY
Meaning, always rebuild them when asked instead of bothering to look at
timestamps (and then wondering why nothing happened when you said make).
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Signed-off-by: Adam Jackson <ajax@redhat.com>
Adam Jackson [Thu, 24 Mar 2016 17:57:57 +0000 (13:57 -0400)]
glapi/glx: Thunk non-ABI calls through GetProcAddress
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Signed-off-by: Adam Jackson <ajax@redhat.com>
Adam Jackson [Thu, 24 Mar 2016 17:57:57 +0000 (13:57 -0400)]
glapi/glx: Emit direct GL calls instead of dispatch lookup
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Signed-off-by: Adam Jackson <ajax@redhat.com>
Adam Jackson [Thu, 24 Mar 2016 17:57:57 +0000 (13:57 -0400)]
glx: Unbreak generating some of the xorg glx headers
Broken by:
commit
9ace0b542241c77ae82a0835ac8a09e2a7510eaf
Author: Dylan Baker <baker.dylan.c@gmail.com>
Date: Wed May 20 15:49:11 2015 -0700
glapi: glX_proto_size.py: use argparse instead of getopt
Which changed most, but not all, callers to use --header-tag instead of
-h.
Reviewed-by: Dylan Baker <baker.dylan.c@gmail.com>
Signed-off-by: Adam Jackson <ajax@redhat.com>
Bas Nieuwenhuizen [Mon, 28 Mar 2016 15:01:49 +0000 (17:01 +0200)]
mesa/st: Fix NULL access if no fragment shader is bound
Signed-off-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Signed-off-by: Marek Olšák <marek.olsak@amd.com>
Rob Clark [Mon, 21 Mar 2016 23:55:37 +0000 (19:55 -0400)]
freedreno/ir3: fix for load_front_face intrinsic
Seems like trying to widen in the same instruction as the add.s does a
non-sign-extending widen.
Signed-off-by: Rob Clark <robclark@freedesktop.org>
Rob Clark [Sat, 6 Feb 2016 14:09:52 +0000 (09:09 -0500)]
freedreno/ir3: fix compiler warn
Signed-off-by: Rob Clark <robclark@freedesktop.org>
Ilia Mirkin [Mon, 28 Mar 2016 04:52:00 +0000 (00:52 -0400)]
nvc0: make sure to disable fetches from previously-set VBOs when blitting
We disable the vertex attributes, but also disable the VBO fetch details
as well, just in case. Not known to fix anything.
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Ilia Mirkin [Sun, 27 Mar 2016 02:32:43 +0000 (22:32 -0400)]
nvc0: disable primitive restart and index bias during blits
Back in the dawn of time, we used to do immediate uploads for the vertex
data, and all was well. However Maxwell dropped support for immediate
vertex data, so we started feeding in a VBO (in all cases). But we
forgot to disable some things that apply in such cases, specifically
primitive restart and index bias. The latter was causing WoW and other
Blizzard games trouble as they use a pattern where they draw with a base
vertex (aka index bias), followed by texture uploads (aka blits,
internally).
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=91526
Cc: "11.1 11.2" <mesa-stable@lists.freedesktop.org>
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Tested-by: Karol Herbst <nouveau@karolherbst.de>
Ilia Mirkin [Sun, 20 Mar 2016 21:26:13 +0000 (17:26 -0400)]
nvc0/ir: fix picking of coordinates from tex instruction for textureGrad
On Fermi, there's an argument in front of the coords that combines array
and indirect handle, while on Kepler the array and the indirect handle
are separate (and in front of the coords). We were previously only
accounting for the array bit of it, if there were an indirect access it
wouldn't be counted in the formula.
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Cc: "11.1 11.2" <mesa-stable@lists.freedesktop.org>
Ilia Mirkin [Sun, 20 Mar 2016 17:11:01 +0000 (13:11 -0400)]
nv50/ir: saturate depth writes
Apparently there's no post-FS clamping logic, so we have to do this by
hand. The depth will never be outside of the 0..1 range, even on
floating point zeta buffers, so this should be safe.
Fixes dEQP-GLES3.functional.fbo.depth.*clamp.* which tests writing
invalid values on various zeta buffer formats.
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Marek Olšák [Sun, 27 Mar 2016 17:11:09 +0000 (19:11 +0200)]
gallium/util: fix up inaccurate behavior of util_framebuffer_state_equal (v2)
v2: move the nr_cbufs check above the loop
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu> (v1)
Marek Olšák [Mon, 21 Mar 2016 11:18:40 +0000 (12:18 +0100)]
st/mesa: only minify height if target != 1D array in st_finalize_texture
The st_texture_object documentation says:
"the number of 1D array layers will be in height0"
We can't minify that.
Spotted by luck. No app is known to hit this issue.
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Miklós Máté [Thu, 24 Mar 2016 00:13:02 +0000 (01:13 +0100)]
mesa: optimize out the realloc from glCopyTexImagexD()
v2: comment about the purpose of the code
v3: also compare texFormat,
add a perf debug message,
formatting fixes
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Signed-off-by: Miklós Máté <mtmkls@gmail.com>
Signed-off-by: Marek Olšák <marek.olsak@amd.com>
Miklós Máté [Thu, 24 Mar 2016 00:13:00 +0000 (01:13 +0100)]
st/mesa: fix handling the fallback texture
This fixes crash when post-processing is enabled in SW:KotOR.
v2: fix const-ness
v3: move assignment into the if() block
Signed-off-by: Miklós Máté <mtmkls@gmail.com>
Signed-off-by: Marek Olšák <marek.olsak@amd.com>
Miklós Máté [Thu, 24 Mar 2016 00:12:58 +0000 (01:12 +0100)]
st/mesa: enable GL_ATI_fragment_shader
Signed-off-by: Miklós Máté <mtmkls@gmail.com>
Signed-off-by: Marek Olšák <marek.olsak@amd.com>
Miklós Máté [Thu, 24 Mar 2016 00:12:57 +0000 (01:12 +0100)]
st/mesa: implement GL_ATI_fragment_shader
v2: fix arithmetic for special opcodes,
fix fog state, cleanup
v3: simplify handling of special opcodes,
fix rebinding with different textargets or fog equation,
lots of formatting fixes
v4: adapt to the compile early, fix later architecture,
formatting fixes
Signed-off-by: Miklós Máté <mtmkls@gmail.com>
Signed-off-by: Marek Olšák <marek.olsak@amd.com>
Miklós Máté [Thu, 24 Mar 2016 00:12:56 +0000 (01:12 +0100)]
program: add ATI_fragment_shader to shader stages list
Signed-off-by: Miklós Máté <mtmkls@gmail.com>
Signed-off-by: Marek Olšák <marek.olsak@amd.com>
Miklós Máté [Thu, 24 Mar 2016 00:12:55 +0000 (01:12 +0100)]
mesa: optionally associate a gl_program to ATI_fragment_shader
the state tracker will use it
Acked-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Signed-off-by: Miklós Máté <mtmkls@gmail.com>
Signed-off-by: Marek Olšák <marek.olsak@amd.com>
Edward O'Callaghan [Sun, 27 Mar 2016 02:05:34 +0000 (13:05 +1100)]
gallium/p_context.h: Make comment more readable
Signed-off-by: Edward O'Callaghan <eocallaghan@alterapraxis.com>
Signed-off-by: Marek Olšák <marek.olsak@amd.com>
Edward O'Callaghan [Sat, 26 Mar 2016 07:35:07 +0000 (18:35 +1100)]
mesa/st: Remove GLSLVersion clamping
While here, remove itermediate glsl_feature_level variable.
Signed-off-by: Edward O'Callaghan <eocallaghan@alterapraxis.com>
Signed-off-by: Marek Olšák <marek.olsak@amd.com>
Edward O'Callaghan [Sat, 26 Mar 2016 07:35:06 +0000 (18:35 +1100)]
radeon/r600: Fix return type in failure branch
Commit `
d4e847ea` introduced a warning about making an
integer from a pointer without a cast, fix it here.
Signed-off-by: Edward O'Callaghan <eocallaghan@alterapraxis.com>
Signed-off-by: Marek Olšák <marek.olsak@amd.com>
Edward O'Callaghan [Sat, 26 Mar 2016 07:35:05 +0000 (18:35 +1100)]
radeon/r600_query.c: Minor style fix
Signed-off-by: Edward O'Callaghan <eocallaghan@alterapraxis.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Signed-off-by: Marek Olšák <marek.olsak@amd.com>
Dave Airlie [Wed, 23 Mar 2016 23:28:49 +0000 (09:28 +1000)]
virgl: drop next shader property for now.
Signed-off-by: Dave Airlie <airlied@redhat.com>
Timothy Arceri [Thu, 24 Mar 2016 01:11:01 +0000 (12:11 +1100)]
glsl: reduce buffer block duplication
This reduces some of the craziness required for handling buffer
blocks. The problem is each shader stage holds its own information
about a block in memory, we were copying that information to a
program wide list but the per stage information remained meaning
when a binding was updated we needed to update all versions of it.
This changes the per stage blocks to instead point to a single
version of the block information in the program list.
Acked-by: Kenneth Graunke <kenneth@whitecape.org>
Brian Paul [Fri, 25 Mar 2016 20:06:39 +0000 (14:06 -0600)]
st/xa: emit sampler view declarations in shaders
Fixes recent regressions with the VMware gallium driver.
Reviewed-by: Charmaine Lee <charmainel@vmware.com>
Tested-by: Charmaine Lee <charmainel@vmware.com>
Tim Rowley [Thu, 24 Mar 2016 17:52:51 +0000 (11:52 -0600)]
swr: [rasterizer jitter] Fix MASKLOADD AVX prototype (float -> i32)
Tim Rowley [Thu, 24 Mar 2016 06:01:23 +0000 (00:01 -0600)]
swr: [rasterizer core] NUMA optimizations...
- Affinitize hot-tile memory to specific NUMA nodes.
- Only do BE work for macrotiles assoicated with the numa node
Tim Rowley [Thu, 24 Mar 2016 00:12:11 +0000 (18:12 -0600)]
swr: [rasterizer jitter] Fix logic bug for alpha-to-coverage.
Tim Rowley [Tue, 22 Mar 2016 23:28:06 +0000 (17:28 -0600)]
swr: [rasterizer core] Fix Compute workitem retirement
Tim Rowley [Tue, 22 Mar 2016 21:13:29 +0000 (15:13 -0600)]
swr: [rasterizer core] Cleanup state ring arena after last draw that references it completes
Rather than waiting for the API thread to re-use it.
Tim Rowley [Tue, 22 Mar 2016 18:41:13 +0000 (12:41 -0600)]
swr: [rasterizer jitter] add missing include for llvm jitevents
Tim Rowley [Tue, 22 Mar 2016 15:27:18 +0000 (09:27 -0600)]
swr: [rasterizer core] Reduce Arena blocksize to 128KB (from 1MB).
With global allocator this doesn't seem to affect performance at all.
Overall memory consumption drops by up to 85%.
Tim Rowley [Mon, 21 Mar 2016 23:55:46 +0000 (17:55 -0600)]
swr: [rasterizer core] One last pass at Arena optimizations
Tim Rowley [Mon, 21 Mar 2016 23:30:03 +0000 (17:30 -0600)]
swr: [rasterizer core] CachedArena optimizations
Reduce list traversal during Alloc and Free.
Add ability to have multiple lists based on alloc size (not used for now)
Tim Rowley [Mon, 21 Mar 2016 20:08:38 +0000 (14:08 -0600)]
swr: [rasterizer jitter] support llvm-svn
Tim Rowley [Mon, 21 Mar 2016 17:15:32 +0000 (11:15 -0600)]
swr: [rasterizer core] Globally cache allocated arena blocks for fast re-allocation.
Tim Rowley [Fri, 18 Mar 2016 18:11:20 +0000 (12:11 -0600)]
swr: [rasterizer] more arena work
Tim Rowley [Fri, 18 Mar 2016 17:48:47 +0000 (11:48 -0600)]
swr: [rasterizer core] Add clipping against user clip distances in the NullPS backend.
Tim Rowley [Fri, 18 Mar 2016 00:10:25 +0000 (18:10 -0600)]
swr: [rasterizer core] Arena optimizations - preparing for global allocator.
Tim Rowley [Thu, 17 Mar 2016 22:50:46 +0000 (16:50 -0600)]
swr: [rasterizer core] Reset DrawContext arena at end of draw rather than upon reclaim of DC
Keeps overall memory consumption lower.
Also, remove unused knobs.
Tim Rowley [Thu, 17 Mar 2016 22:12:17 +0000 (16:12 -0600)]
swr: [rasterizer core] Add clipping of user clip planes in clipper.
Tim Rowley [Thu, 17 Mar 2016 21:39:13 +0000 (15:39 -0600)]
swr: [rasterizer] Reduce max in-flight draws to 96 (by default)
Tim Rowley [Thu, 17 Mar 2016 18:22:43 +0000 (12:22 -0600)]
swr: [rasterizer] Fix run-time check asserts
One innocuous (uninitialized variable), and one not so innocuous
(stack corruption).
Tim Rowley [Wed, 16 Mar 2016 23:54:04 +0000 (17:54 -0600)]
swr: [rasterizer jitter] signed immediate builder
Tim Rowley [Wed, 16 Mar 2016 17:56:50 +0000 (11:56 -0600)]
swr: [rasterizer common] changes for cygwin
Tim Rowley [Mon, 14 Mar 2016 21:54:29 +0000 (15:54 -0600)]
swr: [rasterizer] code styling and update copyrights
Tim Rowley [Fri, 11 Mar 2016 01:20:07 +0000 (19:20 -0600)]
swr: [rasterizer core] Guard against enquing work to invalid hot tiles
Tim Rowley [Fri, 11 Mar 2016 01:19:30 +0000 (19:19 -0600)]
swr: [rasterizer] Stop setting viewport size to larger than hottile array
Guard against enquing work to invalid tiles
Tim Rowley [Fri, 11 Mar 2016 00:30:40 +0000 (18:30 -0600)]
swr: [rasterizer] Discard work + misc fixes
Tim Rowley [Thu, 10 Mar 2016 21:15:40 +0000 (15:15 -0600)]
swr: [rasterizer] remove use of BYTE type
Tim Rowley [Wed, 9 Mar 2016 23:18:55 +0000 (17:18 -0600)]
swr: [rasterizer core] Fix crash that can occur when switching contexts
Tim Rowley [Wed, 9 Mar 2016 22:33:33 +0000 (16:33 -0600)]
swr: [rasterizer] remove unused knob
Tim Rowley [Wed, 9 Mar 2016 22:15:37 +0000 (16:15 -0600)]
swr: [rasterizer core] subcontext rework
Tim Rowley [Wed, 9 Mar 2016 00:58:54 +0000 (18:58 -0600)]
swr: [rasterizer common] add _simd_s[rl]lv_epi32
Tim Rowley [Tue, 8 Mar 2016 17:56:06 +0000 (11:56 -0600)]
swr: [rasterizer core] Alleviate potential stack overflow for 32bit builds
Move large stack allocations in the GS and clipper into thread local storage.
Tim Rowley [Mon, 7 Mar 2016 20:45:17 +0000 (14:45 -0600)]
swr: [rasterizer] remove use of UCHAR and UINT64 types
Tim Rowley [Mon, 7 Mar 2016 16:51:56 +0000 (10:51 -0600)]
swr: [rasterizer] remove use of FLOAT type
Tim Rowley [Mon, 7 Mar 2016 07:14:13 +0000 (01:14 -0600)]
swr: [rasterizer] Fix Coverity issues reported by Mesa developers.
Tim Rowley [Sat, 5 Mar 2016 06:53:04 +0000 (00:53 -0600)]
swr: [rasterizer] add debug/perf category to knobs
Tim Rowley [Thu, 24 Mar 2016 16:07:32 +0000 (11:07 -0500)]
swr: [rasterizer core] don't assume linux is 64-bit
Tim Rowley [Thu, 24 Mar 2016 16:07:15 +0000 (11:07 -0500)]
swr: [rasterizer common] remove old unused win32 types
Tim Rowley [Fri, 4 Mar 2016 00:19:45 +0000 (18:19 -0600)]
swr: [rasterizer jitter] vpermps support
Tim Rowley [Mon, 29 Feb 2016 18:01:48 +0000 (12:01 -0600)]
swr: [rasterizer] Add rdtsc buckets support for shaders
Pass pointer to core buckets mgr back to sim layer.
Add support for RDTSC_START/RDTSC_STOP macros in the builder.
Each unique shader now has a unique bucket associated with it,
enabling more detailed reporting at the shader level. Currently
due to some llvm issue with thread local storage, 64bit runs require
single threaded mode.
Tim Rowley [Wed, 24 Feb 2016 19:34:50 +0000 (13:34 -0600)]
swr: [rasterizer core] backend reorganization
Tim Rowley [Thu, 25 Feb 2016 01:03:33 +0000 (19:03 -0600)]
swr: [rasterizer core] store blend output in temporary instead of PS output.
Fixes additive blend problem with MSAA
Tim Rowley [Tue, 23 Feb 2016 23:29:59 +0000 (17:29 -0600)]
swr: [rasterizer core] Move InitializeHotTiles and corresponding clear code out of threads.cpp.
Tim Rowley [Tue, 23 Feb 2016 19:47:24 +0000 (13:47 -0600)]
swr: [rasterizer jitter] Cleanup use of types inside of Builder.
Also, cached the simd width since we don't have to keep querying
the JitManager for it.
Tim Rowley [Mon, 22 Feb 2016 17:00:07 +0000 (11:00 -0600)]
swr: [rasterizer jitter] Fix type mismatch on select args for SCATTERPS
Tim Rowley [Sat, 20 Feb 2016 01:05:14 +0000 (19:05 -0600)]
swr: [rasterizer core] fix rasterizing multisampling with scissor enabled
We were not evaluating the scissor edge equations at sample positions.
Tim Rowley [Fri, 19 Feb 2016 23:55:23 +0000 (17:55 -0600)]
swr: [rasterizer core] RingBuffer class for DC/DS
Use head/tail ring buffer indices for thread synchronization.
1. SwrWaitForIdle loops until ring is empty. (head == tail)
2. GetDrawContext waits until ring is not full. (head - tail) == Ring Size
3. Draw enqueues by incrementing head.
4. Last worker thread to move past a DC dequeues by incrementing tail.
Todo: To reduce contention we can cache the tail in the API thread. For
example, if you know you have 64 free entries in the ring then you don't
need to keep checking the tail until you used those 64 entries.
Tim Rowley [Fri, 19 Feb 2016 01:00:30 +0000 (19:00 -0600)]
swr: [rasterizer] switch assert uses to SWR_ASSERT
Tim Rowley [Wed, 17 Feb 2016 23:55:59 +0000 (17:55 -0600)]
swr: [rasterizer core] Split all RECT_LIST draws into 1 RECT per draw
Needed until proper RECT_LIST PrimAssembly code is written.
Tim Rowley [Tue, 16 Feb 2016 23:32:34 +0000 (17:32 -0600)]
swr: [rasterizer] Add string knob type
Sonny Jiang [Wed, 4 Nov 2015 16:01:33 +0000 (11:01 -0500)]
radeonsi: add Polaris PCI IDs
Signed-off-by: Sonny Jiang <sonny.jiang@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com> (Polaris10)
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> (Polaris11)
Sonny Jiang [Tue, 15 Dec 2015 20:33:40 +0000 (15:33 -0500)]
radeon/vce: disable two pipe mode for Polaris11
Signed-off-by: Sonny Jiang <sonny.jiang@amd.com>
Reviewed-by: Leo Liu <leo.liu@amd.com>
Sonny Jiang [Tue, 15 Dec 2015 20:16:29 +0000 (15:16 -0500)]
radeon/vce: add Polaris11 VCE firmware support
Signed-off-by: Sonny Jiang <sonny.jiang@amd.com>
Sonny Jiang [Wed, 4 Nov 2015 21:13:07 +0000 (16:13 -0500)]
radeonsi: add support for Polaris (v2)
v2: Polaris chips should be defined after Stoney
Signed-off-by: Sonny Jiang <sonny.jiang@amd.com> (v1)
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> (v1)
Signed-off-by: Leo Liu <leo.liu@amd.com> (v2 diff)
Reviewed-by: Alex Deucher <alexander.deucher@amd.com> (v2 diff)
Sonny Jiang [Tue, 3 Nov 2015 16:46:38 +0000 (11:46 -0500)]
winsys/amdgpu: addrlib - add Polaris support (v2)
v2: fix indentation as noted by Michel
Signed-off-by: Sonny Jiang <sonny.jiang@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Kenneth Graunke [Thu, 24 Mar 2016 05:35:41 +0000 (22:35 -0700)]
mesa: Check glReadBuffer enums against the ES3 table.
From the ES 3.2 spec, section 16.1.1 (Selecting Buffers for Reading):
"An INVALID_ENUM error is generated if src is not BACK or one of
the values from table 15.5."
Table 15.5 contains NONE and COLOR_ATTACHMENTi.
Mesa properly returned INVALID_ENUM for unknown enums, but it decided
what was known by using read_buffer_enum_to_index, which handles all
enums in every API. So enums that were valid in GL were making it
past the "valid enum" check. Such targets would then be classified
as unsupported, and we'd raise INVALID_OPERATION, but that's technically
the wrong error code.
Fixes dEQP-GLES31's
functional.debug.negative_coverage.get_error.buffer.read_buffer
v2: Only call read_buffer_enuM_to_index when required (Eduardo).
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eduardo Lima Mitev <elima@igalia.com>
Jason Ekstrand [Sun, 14 Feb 2016 01:31:05 +0000 (17:31 -0800)]
nir: Add a pass to inline functions
This commit adds a new NIR pass that lowers all function calls away by
inlining the functions.
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Jason Ekstrand [Sat, 26 Dec 2015 18:48:14 +0000 (10:48 -0800)]
nir/builder: Add helpers for easily inserting copy_var intrinsics
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Jason Ekstrand [Sun, 14 Feb 2016 01:08:57 +0000 (17:08 -0800)]
nir: Add return lowering pass
This commit adds a NIR pass for lowering away returns in functions. If the
return is in a loop, it is lowered to a break. If it is not in a loop,
it's lowered away by moving/deleting code as needed.
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Jason Ekstrand [Mon, 28 Dec 2015 06:50:14 +0000 (22:50 -0800)]
nir: Add a cursor helper for getting a cursor after any phi nodes
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Jason Ekstrand [Sun, 14 Feb 2016 01:14:27 +0000 (17:14 -0800)]
nir/builder: Add a helper for inserting jump instructions
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Jason Ekstrand [Thu, 24 Dec 2015 02:10:08 +0000 (18:10 -0800)]
nir/cf: Make extracting or re-inserting nothing a no-op
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Jason Ekstrand [Sat, 26 Dec 2015 18:32:10 +0000 (10:32 -0800)]
nir: Add a function for comparing cursors
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Jason Ekstrand [Fri, 18 Dec 2015 19:27:00 +0000 (11:27 -0800)]
nir/cf: Handle relinking top-level blocks
This can happen if a function ends in a return instruction and you remove
the return.
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Jason Ekstrand [Sat, 13 Feb 2016 05:52:46 +0000 (21:52 -0800)]
nir: Add a pass to repair SSA form
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Jason Ekstrand [Sat, 13 Feb 2016 05:48:26 +0000 (21:48 -0800)]
nir/vars_to_ssa: Use the new nir_phi_builder helper
The efficiency should be approximately the same. We do a little more work
per phi node because we have to sort the predecessors. However, we no
longer have to walk the blocks a second time to pop things off the stack.
The bigger advantage, however, is that we can now re-use the phi placement
and per-block SSA value tracking in other passes.
As a side-benifit, the phi builder actually handles unreachable blocks
correctly. The original vars_to_ssa code, because of the way it iterated
the blocks and added phi sources, didn't add sources corresponding to
predecessors of unreachable blocks. The new strategy employed by the phi
builder creates a phi source for each predecessor and should correctly
handle unreachable blocks by setting those sources to SSA undefs.
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Jason Ekstrand [Tue, 29 Dec 2015 23:25:43 +0000 (15:25 -0800)]
nir/dominance: Handle unreachable blocks
Previously, nir_dominance.c didn't properly handle unreachable blocks.
This can happen if, for instance, you have something like this:
loop {
if (...) {
break;
} else {
break;
}
}
In this case, the block right after the if statement will be unreachable.
This commit makes two changes to handle this. First, it removes an assert
and allows block->imm_dom to be null if the block is unreachable. Second,
it properly skips unreachable blocks in calc_dom_frontier_cb.
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Jason Ekstrand [Sat, 13 Feb 2016 05:41:42 +0000 (21:41 -0800)]
nir: Add a phi node placement helper
Right now, we have phi placement code in two places and there are other
places where it would be nice to be able to do this analysis. Instead of
repeating it all over the place, this commit adds a helper for placing all
of the needed phi nodes for a value.
v2: Add better documentation
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Jason Ekstrand [Sun, 17 Jan 2016 00:42:06 +0000 (16:42 -0800)]
util/bitset: Allow iterating over const bitsets
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Rob Clark [Thu, 24 Mar 2016 19:44:35 +0000 (15:44 -0400)]
ttn: remove stray global from header
Signed-off-by: Rob Clark <robclark@freedesktop.org>
Samuel Pitoiset [Wed, 23 Mar 2016 22:29:20 +0000 (23:29 +0100)]
nv50/ir: silence unhandled TGSI_PROPERTY_NEXT_SHADER info
radeonsi uses this property to make the best decision about which
shader to compile, but this is not currently used by our codegen.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Kenneth Graunke [Thu, 24 Mar 2016 06:46:12 +0000 (23:46 -0700)]
mesa: Handle negative length in glPushDebugGroup().
The KHR_debug spec doesn't actually say we should handle this, but that
is most likely an oversight - it says to check against strlen and
generate errors if length is negative. It appears they just forgot to
explicitly spell out that we should then proceed to actually handle it.
Fixes crashes from uncaught std::string exceptions in many
dEQP-GLES31.functional.debug.error_filters.* tests.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eduardo Lima Mitev <elima@igalia.com>