|
db83279a
|
2021-10-27T10:45:23
|
|
Metal: Use 'fast' normalize to match DEQP expectations
Normalize' has a fast math option that more conformantly
deals with INF in normalize functions.
Use metal::fast::normalize instead of metal::normalize
This also matches SPIRV-Cross behavior for normalization
Bug: angleproject:6629
Change-Id: I61392ecd6f61b612d21f2832e07a135e26b53ad6
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3248145
Commit-Queue: Kyle Piddington <kpiddington@apple.com>
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Gregg Tavares <gman@chromium.org>
|
|
55bd8d8f
|
2021-10-26T10:38:28
|
|
Translator: Fix precision bug in HLSL generation
Bug: chromium:1263487
Change-Id: I60bc3dc93867e4269756b6f54eb39cf29ad77d5f
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3244887
Reviewed-by: Tim Van Patten <timvp@google.com>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
066fb91b
|
2021-10-18T15:31:27
|
|
Vulkan: SPIR-V Gen: Handle scalar(const) produced by index clamp
Normally scalar(const) is folded into a constant. The index clamp
transformation may produce such a code where the index looks dynamic at
first (for example `false ? uniform_value : constant`), but becomes
constant after folding.
This change makes SPIR-V generation robust in that case. A potential
future change could avoid the clamp entirely by making FoldExpressions
adjust the op of the EOpIndexIndirect node whose index is being replaced
with a constant with EOpIndexDirect (and apply the clamp on the
argument).
Bug: chromium:1260651
Change-Id: I552b7527d821d1cb52e0e53212cc481285674861
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3226311
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Tim Van Patten <timvp@google.com>
|
|
|
ef237faf
|
2021-10-18T16:22:40
|
|
Translator: Validate precision for function args
In fragment shaders, float does not have a default precision. Any
declaration of this type must therefore specify the precision if a
default one is not provided.
This was not validated for function arguments.
Bug: chromium:1255089
Change-Id: I0d17e226ec88610692ec7dd18793cf4d471f12e7
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3226314
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Tim Van Patten <timvp@google.com>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
7f87a326
|
2021-10-18T13:06:29
|
|
Vulkan: SPIR-V Gen: Fix crash in array of struct constant
Bug: chromium:1260690
Change-Id: I51fe85a2ebc23c3fcaa3c961c4ebf84688bbed31
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3226309
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Tim Van Patten <timvp@google.com>
|
|
|
57b73c07
|
2021-10-14T15:51:18
|
|
Vulkan: SPIR-V Gen: TODO clean up
This change removes a number of stale TODOs.
Bug: angleproject:4889
Change-Id: If21d55717e61383ffdd58a7f88da5961988e035f
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3223643
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Tim Van Patten <timvp@google.com>
|
|
|
b3d77208
|
2021-10-14T15:38:51
|
|
Vulkan: SPIR-V Gen: Make unfolded constants SPIR-V constants
If a constant is not folded by the translator, for example because it's
an array, this change makes sure that it becomes a constant in SPIR-V
anyway. This is particularly important to support cases where a
constant is required, for example as argument to textureGatherOffsets().
Bug: angleproject:4889
Bug: angleproject:5362
Change-Id: Ic43e4be13a8917bb100ea64902ba90048cfbb9ee
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3223642
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Tim Van Patten <timvp@google.com>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
f3d5dac3
|
2021-08-23T17:25:15
|
|
Vulkan: SPIR-V Gen: Drop dependency to glslang
The SPIR-V gen path is now made default. Compilation through glslang is
still supported for debugging, and is enabled on the GLSL* end2end tests
for smoke testing. On release builds, glslang is not supported.
To test with glslang, add the following gn arg (only necessary if dcheck
is disabled):
angle_enable_spirv_gen_through_glslang = true
Then enable the generateSPIRVThroughGlslang feature. This can be done
by setting an environment variable:
ANGLE_FEATURE_OVERRIDES_ENABLED=generateSPIRVThroughGlslang ./angle_deqp_gles2_tests
Binary size saving:
- 1.3MB on Linux (SPIR-V gen itself: 240KB)
- 730KB on Android (SPIR-V gen itself: 140KB)
Perf tests:
- LinkProgramBenchmark.Run/vulkan_compile_single_thread
* Through glslang:
truncated mean: 1287033.36
* Direct SPIR-V Gen:
truncated mean: 244495.91 (~80% reduction)
- LinkProgramBenchmark.Run/vulkan_compile_multi_thread
* Through glslang:
truncated mean: 4565894.83
* Direct SPIR-V Gen:
truncated mean: 1158164.10 (~75% reduction)
Bug: angleproject:4889
Bug: angleproject:6210
Change-Id: I486342702977c8114e90073b97183aba115a8b2d
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3115140
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Tim Van Patten <timvp@google.com>
|
|
01341f94
|
2021-10-11T19:26:07
|
|
D3D11: implement EXT_clip_control
This implements EXT_clip_control for the D3D11 renderer, so that I can
use a reversed-Z depth buffer with ANGLE.
Tested with
angle_deqp_gles2_tests.exe --deqp-egl-display-type=angle-d3d11 --deqp-case=dEQP-GLES2.functional.clip_control.*
and
angle_end2end_tests.exe --gtest_filter=*D3D11*
Bug: angleproject:6554
Change-Id: I1d11cd04a6654c28530b11104470f0cad0009abe
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3218659
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
|
|
|
bcc89703
|
2021-10-07T11:26:42
|
|
Translator: Remove disableValidateFunctionCall from transformation
This validation is disabled in three transformations and re-enabled
afterwards. Two of these benefit from the simplicity it brings, and
this change removes it from the other transformation.
Bug: angleproject:6506
Change-Id: I946b547ed0e2ed39062914a7771e9635ff5e406d
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3212289
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Tim Van Patten <timvp@google.com>
|
|
4349703b
|
2021-10-06T11:00:53
|
|
Compiler: Remove unnecessary TODO
Remove the TODO related to removing the global pool allocator, since
that work won't be chased any further.
For more details in this area, see:
http://anglebug.com/1286
Bug: angleproject:6497
Change-Id: Ie9952e1c0e481c8d553cfdcf4a1d8ded8b64a95b
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3208510
Reviewed-by: Tim Van Patten <timvp@google.com>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Tim Van Patten <timvp@google.com>
|
|
9ba5dcf6
|
2021-10-07T06:56:14
|
|
Vulkan: Fix inconsistency in half render area width/height
The half render area rotation matrix should be removed.
DrawableArea is constant over rotations.
The frag coord rotation matrix is still needed because
the frag coord generated by GPU is still rotated.
Test:
DISPLAY=:0 out/Debug/angle_deqp_gles3_rotate90_tests
Using test config with:
linux
nvidia
debug
vulkan
prerotation
prerotation270
Test: Cuttlefish SwANGLE
./cts-tradefed run cts -m CtsDeqpTestCases -t
dEQP-GLES3.functional.shaders.builtin_variable.fragcoord_xyz
Bug: angleproject:6504
Change-Id: I37a6308fdb7586200b19a6138affc78c2c329a96
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3212427
Commit-Queue: Lingfeng Yang <lfy@google.com>
Reviewed-by: Tim Van Patten <timvp@google.com>
Reviewed-by: Charlie Lao <cclao@google.com>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
|
daaaba0a
|
2021-10-04T22:28:56
|
|
Translator: Fix uniform/struct separation for anonymous arrays
When a uniform array of anonymous structs was separated into a struct
declaration and uniform array, the uniform symbol was being replaced,
but not the intermediate nodes that index it. This caused an AST
validation error.
Bug: angleproject:4740
Change-Id: Ib9d0115bd01c39d1a007f644e46b5e35c12c9304
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3204590
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Tim Van Patten <timvp@google.com>
|
|
|
dde409be
|
2021-10-04T22:05:47
|
|
Translator: Produce error on func(void, int)
If void is specified, another parameter cannot follow. This change
ensures that this produces an error.
Bug: angleproject:6338
Change-Id: I18602ea4b52c96e498f4f2dc627d4ba2210d6fa7
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3203795
Reviewed-by: Tim Van Patten <timvp@google.com>
Reviewed-by: Jonah Ryan-Davis <jonahr@google.com>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
54d4bfe5
|
2021-09-28T17:27:57
|
|
Update ANGLE Metal to Webkit at Sept 29 2021
This commit merges changes from Webkit into ANGLE upstream.
The following commits were used:
Current:
https://git.webkit.org/?p=WebKit.git;a=commit;h=e01d0bda8f4b7dc2fd834b92802d15d8c15735f
Previous:
https://git.webkit.org/?p=WebKit.git;a=commit;h=492f078198748e8ff248eea0bb979cf79e5f5adfj
The following commits were merged in from the Webkit Repository:
(Hashes from git://git.webkit.org/WebKit-https.git)
03ea44c78ce5665d4ec9add271260121cbc7bc6c
Problems with drawElements in some conditions
https://bugs.webkit.org/show_bug.cgi?id=230107
c8dc8e0c4d1109d39a62eb197b45e95132380290
ANGLE Metal: single-component swizzles do not compile
https://bugs.webkit.org/show_bug.cgi?id=230472
7285dbaaf5af15877d6c332b30ef7a4d67225460
webgl-compressed-texture-s3tc-srgb.html fails on Intel+AMD Metal
https://bugs.webkit.org/show_bug.cgi?id=229941
4c72f92967ecd2a095666fef431384c4f5f60fb4
fragcolor-fragdata-invariant.html fails
https://bugs.webkit.org/show_bug.cgi?id=223317
cd943145467f54e5928793c0dd3dfa2313c007dd
ANGLE Metal index buffer restart range cache could be maintained..
https://bugs.webkit.org/show_bug.cgi?id=227451
f075ff77e592eabd54dd659a8e13617cc5faedc8
ANGLE Metal infinities and NaNs generated with incorrect syntax
https://bugs.webkit.org/show_bug.cgi?id=229439
5862073269122f4b2d43d96d3922757557755e86
[Metal ANGLE] Fix over-autorelease of
rx::DisplayMtl::getMetalDeviceMatchingAttribute()...
<https://webkit.org/b/229128>
85f797ad31db048cb82cbafd428ef77f0b839312 ANGLE Cocoa compiles....
https://bugs.webkit.org/show_bug.cgi?id=228987
a67918ba279ad4842b6ae84a79c3f1c0cdc35ace Avoid infinite recursion...
https://bugs.webkit.org/show_bug.cgi?id=228978
d341f67de0033adcf1ec6373ace6a54b06c4a031
Cherry-pick ANGLE: Revise WebGL's shaderSource validation
https://bugs.webkit.org/show_bug.cgi?id=228951
1e2714d981e97de8234ba055570dfdf56e8b6944
3.5 MB system-wide footprint impact due to thread-locals...
https://bugs.webkit.org/show_bug.cgi?id=228240
d32e5cca34081997d32504b0b56c18b9703ff3be
Build Default Metal library offline
https://bugs.webkit.org/show_bug.cgi?id=227333
33702279faccfd4c8d1c8a6d549925f9ca9a4e8f
WebGL2 demo doesn't work due to failing compilation....
https://bugs.webkit.org/show_bug.cgi?id=226865
0a075885d242db38c4e435a6597173dc3b082173
rAF driven WebGL submits excessive amount of GPU work...
https://bugs.webkit.org/show_bug.cgi?id=227059
f38a92b3e7c17efda269caa7066e7ffe2f828e72
WebGL shader link error in iOS 15 beta: "Internal error..."
https://bugs.webkit.org/show_bug.cgi?id=227723
98d48f011d561531470d97f26a022767b5452fb7
REGRESSION (r279466): [Big Sur] webgl/1.0.3/conformance &...
https://bugs.webkit.org/show_bug.cgi?id=227596
Bug: angleproject:6471
Change-Id: I07166d0dc4b5c3579d98353485b3245b81c7b882
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3194322
Reviewed-by: Jonah Ryan-Davis <jonahr@google.com>
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Commit-Queue: Jonah Ryan-Davis <jonahr@google.com>
Commit-Queue: Kyle Piddington <kpiddington@apple.com>
|
|
|
d3764f08
|
2021-09-30T13:46:15
|
|
Fix dEQP.GLES3/functional_fragment_out_array* on Direct to Metal
Fragment arrays with locations assigned were not being
correctly transformed.
Fix ModifyStruct to handle the changing location when
locations are set
Bug: angleproject:6477
Change-Id: Icbacd50e265abdbaff3be230b2ec201eaa868a14
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3198807
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
32d6e806
|
2021-09-24T23:50:54
|
|
Vulkan: Fix rotation of interpolateAtOffset
interpolateAtOffset's offset was multiplied by flipXY*rotationMatrix,
which is a vector. This was unable to actually apply rotation to the
offset. The multiplier is changed to a matrix.
Can be tested by running interpolate_at_offset*default_framebuffer tests
in deqp gles3.1 together with pre-rotation:
./angle_deqp_gles31_tests
--deqp-case=dEQP.GLES31/functional_shaders_multisample_interpolation_interpolate_at_offset_no_qualifiers_default_framebuffer
--emulated-pre-rotation=90
Bug: angleproject:5843
Change-Id: I9727371375e999f07463b69153df3fd8b28c2a71
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3183108
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Ian Elliott <ianelliott@google.com>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
da3db87e
|
2021-07-06T14:00:58
|
|
Upstream latest changes to Metal backend from Apple to 7/1/2021
This CL merges in the ANGLE changes between these two WebKit commits:
https://git.webkit.org/?p=WebKit.git;a=commit;h=8648b353ab1d7730438c2e08319e1a4d64982c31
https://git.webkit.org/?p=WebKit.git;a=commit;h=166e4924a52971d6a32ad48247a439b16c00e062
Include provoking vertex buffer out of bounds fix
from https://bugs.webkit.org/show_bug.cgi?id=230107
Fix bad merge of resetting of dirty bits, breaking
DepthStencilFormatsTest.DepthTextureRender test and perhaps others.
Disable GL_APPLE_clip_distance when the direct-to-Metal compiler is
active. It can not yet handle the gl_ClipDistance array.
Disable use of rectangular textures for IOSurfaces. Metal can bind
IOSurfaces to 2D textures, and this was passing all tests in the
SPIR-V Metal backend. Introducing rectangular textures breaks the
SPIR-V Metal backend, and the tests currently fail on the
direct-to-Metal backend.
Fix several bugs with ProvokingVertex, which was causing
both the SpirV and Direct backends to incorrectly draw
indices.
(https://bugs.webkit.org/show_bug.cgi?id=230107)
Skip the following tests on the Metal backend which is still failing
RobustResourceInitTestES3.BlitDepthStencilAfterClearBuffer
GLSLTest_ES3.GLVertexIDIntegerTextureDrawArrays/ES3_Metal
With these changes, angle_end2end_tests again runs to completion.
Bug: angleproject:6395
Change-Id: I3cc58f531426a95fc8f177a4ad87f56c1855a546
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3167010
Reviewed-by: Jonah Ryan-Davis <jonahr@google.com>
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Commit-Queue: Kyle Piddington <kpiddington@apple.com>
|
|
|
1a0a0764
|
2021-09-27T12:56:57
|
|
Translator: Fix outputting anonymous struct types
Anonymous struct types have an "empty" symbol name, so instead of
outputting their name, it's specified that they are anonymous.
Bug: chromium:1253208
Change-Id: Ic70767ba32bda6ceb214dfc3d0b4928173665d81
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3187810
Reviewed-by: Tim Van Patten <timvp@google.com>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
bc8fe7dd
|
2021-09-24T22:50:32
|
|
Translator: Fix removing inactive output vs SH_INIT_OUTPUT_VARIABLES
The code that SH_INIT_OUTPUT_VARIABLES adds for inactive outputs is also
removed with this change.
Bug: chromium:1253016
Change-Id: I3caeb628487354de943f7b53f7462f58baa3eca8
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3183107
Reviewed-by: Tim Van Patten <timvp@google.com>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
d71dc819
|
2021-09-22T16:01:06
|
|
Fixes for RowMajor test failures
row_major matrices get translated to column major matrices
and the expressions rewritten. To do this temporary variables
that contain the column and row index expressions are moved
to separte statements before the matrix expression itself.
When this happens it breaks side effects and short circuits
in the expression.
The solution is to call SeparateCompoundExpressions before
calling RewriteRowMajorMatrices.
Bug: angleproject:6386
Change-Id: I4d819ade959e2a875495f62733e11d86560d3fab
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3177337
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Gregg Tavares <gman@chromium.org>
|
|
|
6106da49
|
2021-09-22T16:18:21
|
|
Indent Generated GLSL
Before:
in vec4 _ua_position;
void main(){
(gl_Position = vec4(0.0, 0.0, 0.0, 0.0));
(gl_Position = _ua_position);
}
layout(shared) uniform _uG{
float _ufoobar;
} _ug;
struct _uS {
int _ufoo;
float _ubar;
};
layout(shared) uniform _uT{
_uS _us;
float _um;
} _ut;
float _ufunc(in _uS _ustructVar){
return _ustructVar._ubar;
}
out vec4 _uoutColor;
void main(){
float _uv = 0.0;
for (int _ui = 0; (_ui < _ut._us._ufoo); (++_ui))
{
int _uk = 0;
while ((_uk < _ut._us._ufoo))
{
(_uv += _ufunc(_ut._us));
if ((_uv > 100.0))
{
break;
}
else
{
(_uv += 5.0);
}
switch (_uk) {
case (0):
(_uv += 3.0);
break;
case (1):
(_uv += 1.0);
break;
default:
(_uv -= 2.0);
break;
}
}
}
(_uoutColor = (vec4(_uv) + vec4(_ug._ufoobar)));
}
After:
in vec4 _ua_position;
void main(){
(gl_Position = vec4(0.0, 0.0, 0.0, 0.0));
(gl_Position = _ua_position);
}
layout(shared) uniform _uG{
float _ufoobar;
} _ug;
struct _uS {
int _ufoo;
float _ubar;
};
layout(shared) uniform _uT{
_uS _us;
float _um;
} _ut;
float _ufunc(in _uS _ustructVar){
return _ustructVar._ubar;
}
out vec4 _uoutColor;
void main(){
float _uv = 0.0;
for (int _ui = 0; (_ui < _ut._us._ufoo); (++_ui))
{
int _uk = 0;
while ((_uk < _ut._us._ufoo))
{
(_uv += _ufunc(_ut._us));
if ((_uv > 100.0))
{
break;
}
else
{
(_uv += 5.0);
}
switch (_uk) {
case (0):
(_uv += 3.0);
break;
case (1):
(_uv += 1.0);
break;
default:
(_uv -= 2.0);
break;
}
}
}
(_uoutColor = (vec4(_uv) + vec4(_ug._ufoobar)));
}
Bug: angleproject:6428
Change-Id: Ica25151e68eb0defd85a3b6241db442d8eb84743
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3177339
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Gregg Tavares <gman@chromium.org>
|
|
|
e1a9af6a
|
2021-09-20T10:42:03
|
|
Translator: Fix HLSL vs for loop's init not having initializer
The handleExcessiveLoop code expected that if a for loop has an init
expression that it must necessarily have an initializer as well. That
is not true in (fuzzer-produced) code such as following:
for (int i; ...; ...)
Bug: chromium:1238182
Change-Id: Id761e937a34a6b2feac1266e22207350a3bd616c
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3170114
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Tim Van Patten <timvp@google.com>
|
|
|
f6616c71
|
2021-09-14T17:42:22
|
|
Reimplement transform feedback on direct-to-Metal backend
Connect the Transform Feedback code generation from Webkit ANGLE
to the shader specialization code.
Bug: angleproject:6393
Change-Id: I090c44c6ee97e8e0af8c38433bfb74c2080784f9
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3161455
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Reviewed-by: Jonah Ryan-Davis <jonahr@google.com>
Reviewed-by: Gregg Tavares <gman@chromium.org>
Commit-Queue: Kenneth Russell <kbr@chromium.org>
|
|
|
46149c0e
|
2021-09-17T15:38:47
|
|
Translator: Fix pruning functions that declare a struct
When pruning unused functions, the entire declaration of the function is
removed. If the function declares a struct as part of its return value,
the declaration of that struct was lost. This change makes sure that
declaration survives the pruning. For example:
struct S { vec4 v; } unused_func() { ... }
is now replaced with:
struct S { vec4 v; };
This change also makes struct usage validation more stringent.
Bug: chromium:1248753
Change-Id: Idd9a87fd2e785135775cfea62995fd33adaf3c09
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3169691
Reviewed-by: Tim Van Patten <timvp@google.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
158ef351
|
2021-09-15T13:40:28
|
|
Fix integer overflow in BlockLayoutEncoder
BlockLayoutEncoder::mCurrentOffset's computation had the
possibility of causing integer overflows in multiple places,
so this CL adds CheckedNumeric variables in a number of
these occurrences in order to prevent integer overflows and
causing issues.
The issue in this case was an integer overflow causing the
code in ValidateTypeSizeLimitations.cpp to use an invalid
result from "layoutEncoder.getCurrentOffset()", which ended
up compiling a shader which would later cause an OOM error.
Bug: chromium:1248665
Change-Id: I688d669f21c6dc2957e43bdf91f8f8f08180a6f7
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3163356
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Commit-Queue: Alexis Hétu <sugoi@chromium.org>
|
|
|
9112b1a9
|
2021-09-13T15:39:53
|
|
Vulkan: SPIR-V Gen: Fix precision of findMsb
In GLSL, findMsb operates on a highp operand, while returning a lowp
result. In SPIR-V, RelaxedPrecision cannot be applied on the FindSMsb
instruction as that affects the operand as well. This change makes sure
RelaxedPrecision is not applied to a FindSMsb instruction.
Bug: angleproject:4889
Bug: angleproject:6132
Change-Id: I6a0defbbd53ec703c7ecbad5cd5c79d215b11c32
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3158506
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Tim Van Patten <timvp@google.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
|
9c6f6c39
|
2021-09-10T14:46:18
|
|
nullptr_t to std::nullptr_t
The Skia build requires std::nullptr_t
Bug: angleproject:6377
Change-Id: Ic41955e1dfefcf24bc39240d164ad938f523717c
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3154997
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Commit-Queue: Gregg Tavares <gman@chromium.org>
|
|
|
2795866c
|
2021-09-10T02:08:40
|
|
Fix ConvertUnsupportedConstructorsToFunctionCalls
The code was not hanlding nested expressions. For example:
mat4(vec4(mat2x2), ...)
Switched to using TIntermRebuild
Bug: angleproject:5505
Change-Id: I845d94326324ac48d7489225d42d0c6e38622492
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3152168
Commit-Queue: Kenneth Russell <kbr@chromium.org>
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
958dce52
|
2021-09-09T14:36:37
|
|
Move IntermRebuild.h/cpp to tree_util
Bug: angleproject:5505
Change-Id: I7b68057fff0a0eb0d86c1aed01599fa9dbb7db3b
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3152167
Commit-Queue: Gregg Tavares <gman@chromium.org>
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
13e9817b
|
2021-09-08T11:57:00
|
|
Translator: Fix assert in ScalarizeVecAndMatConstructorArgs
This transformation assumed that precision can be derived for constants
in every possible scenario, but that's not true. The fuzzer produced
the following code:
void main()
{
mat4 m;
mat2(0, bvec3(m));
}
In the above, the constant 0 legimitately has no precision. The assert
was in a function that attempted to make a temporary out of the
constant, which this commit changes to use the original constant as-is.
Bug: chromium:1246781
Change-Id: I6f247264e5213cfd9449fdfb1dc312d02b99f2f1
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3149191
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Tim Van Patten <timvp@google.com>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
fa9f4cf3
|
2021-09-08T12:58:59
|
|
Fix AST validation error with gl_FragCoord
Bug: angleproject:6370
Change-Id: Ia575e0329da74c84365b74b39398551b572b91a5
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3149725
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Gregg Tavares <gman@chromium.org>
|
|
|
c9acd799
|
2021-09-07T22:25:23
|
|
Only rewrite constructors for scalars/vectors/matrices
Bug: angleproject:5505
Change-Id: I50b3a9a585b2eec6e4d7bd9b3197206a973d4400
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3147071
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Gregg Tavares <gman@chromium.org>
|
|
|
8b2fd983
|
2021-09-08T11:36:17
|
|
Translator: Fix validation failure in RemoveDynamicIndexing
This transformation adds function definitions to the tree after the
fact, so function call validation is disabled until the transformation
is complete.
Bug: chromium:1246686
Change-Id: I40f77cb25cf272d5ebc42ece572371ce5161e819
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3148850
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
0bb3fcbb
|
2021-09-07T14:28:09
|
|
Translator: Fix precision of gl_PointSize on ES3+
In ESSL100, gl_PointSize is declared as mediump. In ESSL300+, it's
declared as highp.
Bug: angleproject:4889
Bug: angleproject:6132
Change-Id: Ie904a12fdaaca04e2507f1796d9d0fe34b154eab
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3145615
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Tim Van Patten <timvp@google.com>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
ae555e85
|
2021-09-05T21:43:03
|
|
Translator: Fix uninitialized TStructure::mAtGlobalScope
This flag was not set in SeparateDeclarations when creating a struct.
Bug: chromium:1246301
Change-Id: I673a7158bcc8ead0fd36a5b00696dd990950a8c0
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3139663
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
ab187c35
|
2021-09-02T14:46:29
|
|
Handle comparing arrays of scalars/vectors/matrices
Fixes:
UniformBlockWithOneLargeArrayMemberTest.MemberArrayOperations/ES3_Metal
Bug: angleproject:5505
Change-Id: Ife35db2146ac8ce8c0880cdb98df57dc75cb35cb
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3140712
Commit-Queue: Gregg Tavares <gman@chromium.org>
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
8bb3c827
|
2021-07-22T19:06:40
|
|
Fix Multithreaded eglDestroyContext()/eglTerminate()
The following EGL calls can lead to a crash in eglMakeCurrent():
Thread A: eglMakeCurrent(context A)
Thread B: eglDestroyContext(context A)
B: eglTerminate() <<--- this release context A
Thread A: eglMakeCurrent(context B)
The eglMakeCurrent(context B) call will assert when attempting to
unMakeCurrent(), since thread A doesn't know that context A was already
destroyed by thread B.
To fix this:
1.) A Context will only be released once there are no Threads that
currently have a reference to it (no longer have the Context current).
- Context::mIsCurrent is being removed, since it was inaccurate and not
thread-safe. For example, when eglTerminate() was called, the
eglTerminate()'ing-Thread would "steal" the Context that was current on
another Thread to destroy it.
2.) A Display will only be fully terminated and its resources released
once all Contexts have been destroyed and are no longer current.
Otherwise, Display::terminate() will return if any Contexts are still in
use by a Thread.
EGL 1.5 Specification
3.2 Initialization
If contexts or surfaces, created with respect to dpy are current (see
section 3.7.3) to any thread, then they are not actually destroyed
while they remain current. If other resources created with respect to
dpy are in use by any current context or surface, then they are also
not destroyed until the corresponding context or surface is no longer
current.
With this fix, the app com.netmarble.sknightsmmo can start.
This also exposed an issue with GlslangFinalize(), since glslang can
only be initialized/finalized once per process. Otherwise, the
following EGL commands will call GlslangFinalize() without ever being
able to GlslangInitialize() again, leading to crashes since
GlslangFinalize() cleans up glslang for the entire process.
dpy1 = eglGetPlatformDisplay() |
eglInitialize(dpy1) | GlslangInitialize()
dpy2 = eglGetPlatformDisplay() |
eglInitialize(dpy2) | GlslangInitialize()
eglTerminate(dpy2) | GlslangFinalize()
eglInitialize(dpy1) | isInitialized() == true
Since Display::isInitialized() == true, the rest of
Display::initialize() is skipped and GlslangInitialize() is not called.
Later, the next test that attempts to compile a program will crash due
to glslang no longer being initialized.
Finally, this exposed the following tests leaking EGLContext handles:
- EGLSurfaceTest::initializeContext()
- EGLContextSharingTest.DisplayShareGroupContextCreation
- EGLCreateContextAttribsTest.IMGContextPriorityExtension
- EGLMultiContextTest.TestContextDestroySimple
Other tests were failing to reset the context, preventing the Display
from being terminated since there were still references to Contexts
owned by the display:
eglMakeCurrent(dpy, EGL_NO_SURFACE, EGL_NO_SURFACE, EGL_NO_CONTEXT);
Bug: angleproject:6208
Bug: angleproject:6304
Bug: angleproject:6322
Test: EGLContextSharingTest.EglTerminateMultiThreaded
Test: EGLContextSharingTestNoFixture.EglDestoryContextManyTimesSameContext
Test: Load com.netmarble.sknightsmmo
Change-Id: I160922af93db6cabe0ed396be77762fa8dfc7656
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3046961
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Tim Van Patten <timvp@google.com>
|
|
|
fefbc3fb
|
2021-09-01T16:26:20
|
|
Vulkan: SPIR-V Gen: Re-fix precision of constructors
In f3aea7447eae7f5a4a5ad1425ab375b66048ba3f, the precision of
constructors was changed such that it would be derived from its
operands, but also be able to be promoted based on neighboring operands.
Based on https://gitlab.khronos.org/opengl/GLSL/-/issues/58 this is
incorrect, and indeed the precision of the constructor (when possible)
solely depends on its own operands.
Bug: angleproject:4889
Bug: angleproject:6132
Change-Id: Ic2560d17973f56d4d836e3c7112d6649197a306c
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3138534
Reviewed-by: Tim Van Patten <timvp@google.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
e02753fc
|
2021-08-31T21:09:45
|
|
Convert constructors to function calls where needed.
MSL does not do as many conversions between types
and has more strict constructors than GLSL so
convert to function calls where necessary.
Fixes:
GLSLTest_ES3.AmbiguousConstructorCall2x2/ES3_Metal
GLSLTest_ES3.AmbiguousConstructorCall2x3/ES3_Metal
GLSLTest_ES3.ConstructMatrixFromNonFloat/ES3_Metal
GLSLTest_ES3.ConstructNonFloatVectorFromMatrix/ES3_Metal
GLSLTest_ES3.ScalarConstructor/ES3_Metal
UniformBufferTest.Std140UniformBlockWithRowMajorQualifier/ES3_Metal
UniformBufferTest.Std140UniformBlockWithPerMemberRowMajorQualifier/ES3_Metal
UniformBufferTest.Std140UniformBlockWithPerMemberColumnMajorQualifier/ES3_Metal
UniformBufferTest.Std140UniformBlockWithRowMajorQualifierOnStruct/ES3_Metal
SimpleUniformTest.FloatMatrix2UniformStateQuery/ES2_Metal
SimpleUniformTest.FloatMatrix2UniformStateQuery/ES3_Metal
SimpleUniformTest.FloatMatrix3UniformStateQuery/ES2_Metal
SimpleUniformTest.FloatMatrix3UniformStateQuery/ES3_Metal
SimpleUniformTest.ArrayOfMat3UniformStateQuery/ES2_Metal
SimpleUniformTest.ArrayOfMat3UniformStateQuery/ES3_Metal
UniformTestES3.MatrixArrayUniformStateQuery/ES3_Metal
Bug: angleproject:6306
Change-Id: Iea8a9a261f94f121f482c2ea9678192ca056570c
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3134963
Commit-Queue: Gregg Tavares <gman@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Kenneth Russell <kbr@chromium.org>
|
|
|
9d0e2851
|
2021-09-01T00:42:58
|
|
Fix SeparateDeclarations vs struct specifiers
Bug: chromium:1237696
Change-Id: I3b00f3797800e814ca83226a8e4f25b2a43cc641
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3133824
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Tim Van Patten <timvp@google.com>
|
|
|
f3aea744
|
2021-08-25T22:09:58
|
|
Vulkan: SPIR-V Gen: Fix precision of constructors
According to the spec, constructors don't have a precision (and thus
their precision is derived from the operands they are used in.
Bug: angleproject:4889
Bug: angleproject:6132
Change-Id: I557386764abf5493fbe59803dc8e260312c04b48
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3119352
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Tim Van Patten <timvp@google.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
|
fbdd0516
|
2021-08-27T00:04:11
|
|
Vulkan: SPIR-V Gen: Fix precision of imageLoad
According to the spec, imageLoad should return highp. However, glslang
produces the same precision as the image (which makes more sense). This
seems to be a spec bug per the discussion in this issue:
https://gitlab.khronos.org/opengl/GLSL/-/issues/57
This change aligns ANGLE with glslang w.r.t to the precision of
imageLoad.
Bug: angleproject:4889
Bug: angleproject:6132
Change-Id: Ia0ce493a0deab21810458da6b48fdf51dfbd41a2
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3122317
Reviewed-by: Tim Van Patten <timvp@google.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
a29b07d8
|
2021-08-26T13:11:29
|
|
InitializeVariables: Init shader IO block outputs.
These variables were not handled in the pass because we had no prior
test coverage of them. Some frame capture testing uncovered this gap.
The variables must be initialized field-by-field, since there seems
to be no defined way to initialize an entire block at once.
Bug: angleproject:6326
Change-Id: Ib7aecfb76b97b4236d786b44b3dfb706c573e221
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3123228
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Tim Van Patten <timvp@google.com>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
4c56534f
|
2021-08-25T15:56:59
|
|
Include globals when defering global initializers
Fixes these tests:
GLSLTest.StructWithInitializer/ES2_Metal
GLSLTest.StructWithInitializer/ES3_Metal
GLSLTest.StructWithUniformInitializer/ES2_Metal
GLSLTest.StructWithUniformInitializer/ES3_Metal
GLSLTest_ES3.SequenceOperatorEvaluationOrderDynamicVectorIndexingInLValue/ES3_Metal
WebGL2GLSLTest.InitUninitializedLocals/ES3_Metal
Bug: angleproject:5505
Change-Id: Ib8258898c60b9e9ffbb71f0024f8189dc6cf4d5b
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3120093
Commit-Queue: Gregg Tavares <gman@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Kenneth Russell <kbr@chromium.org>
|
|
|
9c271645
|
2021-08-27T13:07:06
|
|
Vulkan: SPIR-V Gen: Fix gl_PerVertex without clip/cull support
When the clip/cull distance extension is not supported, the
gl_ClipDistances and gl_CullDistances arrays were given a size of 0,
which made them translate as OpRuntimeArray.
This change makes sure that these arrays always have a non-zero size.
Bug: angleproject:4889
Change-Id: I8d2ffe3a7c1df9316e91eebc2aa53bdc2c87ccc1
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3123354
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
f9d261f5
|
2021-08-25T10:23:48
|
|
Translator: Don't promote precision from initializer
Take the following GLSL code:
precision mediump float;
uniform highp float u;
float x = u;
When `x` is declared, its precision is mediump due to
`precision mediump float`. Its initializer (`u`) however is highp.
Previously, ANGLE generated the following code to ensure the precision
of the initializer is applied to the variable:
highp float x = u;
Compiling the same code with glslang and looking at the SPIR-V (both
SPIR-V for OpenGL and SPIR-V for Vulkan), such a promotion is not done.
The generated SPIR-V is equivalent to:
mediump float x = u;
This change aligns ANGLE's output with glslang's by outputting the
precision of the variable itself instead of the precision of the
assignment (which is highp due to promotion from the two operands).
Bug: angleproject:4889
Bug: angleproject:6132
Change-Id: I6f338a865c0a7710329fc13bc171fd245dd30b31
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3118965
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Tim Van Patten <timvp@google.com>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
3408ea3e
|
2021-08-24T11:41:42
|
|
Translator: Fix precision of gl_FragCoord on ES3+
In ESSL100, gl_FragCoord is declared as mediump. In ESSL300+, it's
declared as highp.
Bug: angleproject:4889
Bug: angleproject:6132
Change-Id: I42583d07fb745421f195be95fa404c617227b02e
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3116307
Reviewed-by: Tim Van Patten <timvp@google.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
800e82c6
|
2021-08-23T11:05:23
|
|
Translator: Validate precisions
When declaring a variable, a struct field, function parameter etc,
there's a precision necessarily applied to the entity being declared.
AST Validation is added to enforce this. Intermediate nodes derive
their precision from these entities automatically.
Consistency of intermediate nodes is not validated. This is because AST
transformations replace a node with a transformed one, and that may not
have the same precision. Take the following code:
mediump float x = ...;
mediump float y = ...;
... x + y ...
and assume is transformed as such:
highp float driver_uniform;
... (x * driver_uniform) + y ...
The addition was originally done in mediump, but would seemingly need to
be done in highp after transformation. There are a number of options
here:
- Make sure that when nodes are replaced, the precision is unaffected.
This can be intrusive, requiring temp variables.
- Bubble up the new precision
- Accept the discrepancy
ANGLE opts for the last option, which actually respects the original
shader's intended precision for operations, even if some transformation
needs to temporarily evaluate an expression at a higher precision.
Bug: angleproject:4889
Bug: angleproject:6132
Change-Id: Ibcde3a230de159157783b1c6d5ef1cd63ceb4d8f
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3114027
Reviewed-by: Tim Van Patten <timvp@google.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
a08b667d
|
2021-08-24T12:13:23
|
|
Translator: Stabilize autogenerated builtin function parameters
When pruning the list of builtin function parameters, sorted() is used
to stabilize the result. This makes reviews of changes to the
autogenerated output easier.
Bug: angleproject:4889
Change-Id: I97afc31de2a62b34a0a8183bb5a47eab9e62d9fc
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3116306
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
411137bf
|
2021-08-24T11:46:51
|
|
Emit INFINITY for infinite constants
GLSL had `v = 1.0e40` which was output as
`v = inff` in MSL. Changed to output
`v = INFINITY`.
Fixes Tests:
GLSLTest_ES3.LiteralInfinityOutput/ES3_Metal
GLSLTest_ES3.LiteralNegativeInfinityOutput/ES3_Metal
Bug: angleproject:5505
Change-Id: Ib651583e608351784dcd49d602a23b5ecb8475cb
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3116548
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Gregg Tavares <gman@chromium.org>
|
|
|
210773db
|
2021-08-05T10:41:59
|
|
Translator: Be more explicit about precisions
GLSL ES requires that every symbol (variable, block member, function
parameter and return value) is appropriately qualified with a precision,
either individually or through the global precision specifier.
Some tree transformations however produced symbols with EbpUndefined
precision. In text GLSL output, these would produce unqualified symbols
which was often incorrect.
In this change, the transformations are made to produce explicit / more
consistent precisions. The validation (that caught these issues) is not
included in this change as there are still a few corner cases left to
address.
Bug: angleproject:4889
Bug: angleproject:6132
Change-Id: Icca8a0a5476f8646226e7243aa8f501f44acc164
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3075127
Reviewed-by: Tim Van Patten <timvp@google.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
b196eec0
|
2021-08-20T12:32:03
|
|
translator: Fix InitializeVariables with gl_ClipDistance.
The extension allows for the shader to redeclare this variable with
a different size. This would sometimes put the tree in an inconsistent
state when used with the InitializeVariables feature. Fix this by using
the user declaration of gl_ClipDistance when available.
Found when working on trace validation, where using the init output
variables option in the compiler would then cause tree validation to
fail for one specific test.
Test: ClipDistanceTest.ThreeClipDistancesRedeclared
Bug: angleproject:5133
Change-Id: I42c9bfbe313826ac5d391068d9fc18eb5793375f
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3110752
Reviewed-by: Cody Northrop <cnorthrop@google.com>
Reviewed-by: Tim Van Patten <timvp@google.com>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
|
|
|
49af677a
|
2021-08-16T07:59:18
|
|
Capture/Replay: Update serialization logic.
- Uses more common toString methods instead of custom ones.
- Uses enum to string in more places intsead of numeric enums.
- Adds missing group scopes in several places.
- Skips serializing empty image levels.
Bug: angleproject:5133
Change-Id: I2c9523981aba4ec0afb5f0623cb34154db263e6b
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3097805
Reviewed-by: Gert Wollny <gert.wollny@collabora.com>
Reviewed-by: Cody Northrop <cnorthrop@google.com>
Reviewed-by: Tim Van Patten <timvp@google.com>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
|
|
|
5ca0953d
|
2021-08-18T13:22:22
|
|
Use the correct attribute locations in Metal
Fixes these tests:
BlendIntegerTest.MRTSigned/ES3_Metal
BlendIntegerTest.MRTUnsigned/ES3_Metal
DrawBuffersTest.Gaps/ES3_Metal
DrawBuffersTest.BlendWithGaps/ES3_Metal
DrawBuffersTest.FirstAndLast/ES3_Metal
DrawBuffersTest.FirstHalfNULL/ES3_Metal
DrawBuffersWebGL2Test.TwoProgramsWithDifferentOutputsAndClear/ES3_Metal
WebGLCompatibilityTest.DrawBuffers/ES3_Metal
Bug: angleproject:5505
Change-Id: I9ea82a8c7a0c5bbc78eb69eb92552909e3f437f3
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3105439
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Commit-Queue: Kenneth Russell <kbr@chromium.org>
|
|
|
63bb0e53
|
2021-08-16T16:30:32
|
|
Add depth fix to convert from GL to Metal
Vertex shaders need to convert from GL's -1 to +1 z range
to Metal's 0 to +1 z range with:
z = (z + w) * 0.5
This code is being upstreamed from WebKit.
It fixes these end to end tests.
BlitFramebufferTest.MultisampleDepthClear/ES3_Metal
BlitFramebufferTest.MultisampleDepthClear/ES3_Metal_NoStencilOutput
BlitFramebufferTest.BlitDepthStencilPixelByPixel/ES3_Metal
BlitFramebufferTest.BlitDepthStencilPixelByPixel/ES3_Metal_NoStencilOutput
ClearTest.ClearStencilMask/ES2_Metal
ClearTest.ClearStencilMask/ES3_Metal
ClearTestES3.ClearMultipleAttachmentsIndividually/ES3_Metal
ClearTestES3.MaskedClearHeterogeneousAttachments/ES3_Metal
ClearTestES3.ScissoredClearHeterogeneousAttachments/ES3_Metal
ClearTestES3.ClearBufferivStencilMask/ES3_Metal
ClearTestES3.ClearBufferfiNoStencilAttachment/ES3_Metal
MaskedScissoredClearTest.Test/ES2_Metal__clear_s
MaskedScissoredClearTest.Test/ES2_Metal__clear_s_mask_s
MaskedScissoredClearTest.Test/ES2_Metal__scissored_clear_s_mask_s
MaskedScissoredClearTest.Test/ES2_Metal__clear_d
MaskedScissoredClearTest.Test/ES2_Metal__scissored_clear_d
MaskedScissoredClearTest.Test/ES2_Metal__clear_ds
MaskedScissoredClearTest.Test/ES2_Metal__scissored_clear_ds
MaskedScissoredClearTest.Test/ES2_Metal__clear_ds_mask_s
MaskedScissoredClearTest.Test/ES2_Metal__scissored_clear_ds_mask_s
MaskedScissoredClearTest.Test/ES2_Metal__clear_cs
MaskedScissoredClearTest.Test/ES2_Metal__clear_cs_mask_s
MaskedScissoredClearTest.Test/ES2_Metal__scissored_clear_cs_mask_s
MaskedScissoredClearTest.Test/ES2_Metal__clear_cd
MaskedScissoredClearTest.Test/ES2_Metal__scissored_clear_cd
MaskedScissoredClearTest.Test/ES2_Metal__clear_cds
MaskedScissoredClearTest.Test/ES2_Metal__scissored_clear_cds
MaskedScissoredClearTest.Test/ES2_Metal__clear_cds_mask_s
MaskedScissoredClearTest.Test/ES2_Metal__scissored_clear_cds_mask_s
MaskedScissoredClearTest.Test/ES2_Metal__clear_cs_mask_cs
MaskedScissoredClearTest.Test/ES2_Metal__scissored_clear_cs_mask_cs
MaskedScissoredClearTest.Test/ES2_Metal__clear_cd_mask_c
MaskedScissoredClearTest.Test/ES2_Metal__scissored_clear_cd_mask_c
MaskedScissoredClearTest.Test/ES2_Metal__clear_cds_mask_c
MaskedScissoredClearTest.Test/ES2_Metal__scissored_clear_cds_mask_c
MaskedScissoredClearTest.Test/ES2_Metal__clear_cds_mask_cs
MaskedScissoredClearTest.Test/ES2_Metal__scissored_clear_cds_mask_cs
MaskedScissoredClearTest.Test/ES3_Metal__clear_s
MaskedScissoredClearTest.Test/ES3_Metal__clear_s_mask_s
MaskedScissoredClearTest.Test/ES3_Metal__scissored_clear_s_mask_s
MaskedScissoredClearTest.Test/ES3_Metal__clear_d
MaskedScissoredClearTest.Test/ES3_Metal__scissored_clear_d
MaskedScissoredClearTest.Test/ES3_Metal__clear_ds
MaskedScissoredClearTest.Test/ES3_Metal__scissored_clear_ds
MaskedScissoredClearTest.Test/ES3_Metal__clear_ds_mask_s
MaskedScissoredClearTest.Test/ES3_Metal__scissored_clear_ds_mask_s
MaskedScissoredClearTest.Test/ES3_Metal__clear_cs
MaskedScissoredClearTest.Test/ES3_Metal__clear_cs_mask_s
MaskedScissoredClearTest.Test/ES3_Metal__scissored_clear_cs_mask_s
MaskedScissoredClearTest.Test/ES3_Metal__clear_cd
MaskedScissoredClearTest.Test/ES3_Metal__scissored_clear_cd
MaskedScissoredClearTest.Test/ES3_Metal__clear_cds
MaskedScissoredClearTest.Test/ES3_Metal__scissored_clear_cds
MaskedScissoredClearTest.Test/ES3_Metal__clear_cds_mask_s
MaskedScissoredClearTest.Test/ES3_Metal__scissored_clear_cds_mask_s
MaskedScissoredClearTest.Test/ES3_Metal__clear_cs_mask_cs
MaskedScissoredClearTest.Test/ES3_Metal__scissored_clear_cs_mask_cs
MaskedScissoredClearTest.Test/ES3_Metal__clear_cd_mask_c
MaskedScissoredClearTest.Test/ES3_Metal__scissored_clear_cd_mask_c
MaskedScissoredClearTest.Test/ES3_Metal__clear_cds_mask_c
MaskedScissoredClearTest.Test/ES3_Metal__scissored_clear_cds_mask_c
MaskedScissoredClearTest.Test/ES3_Metal__clear_cds_mask_cs
MaskedScissoredClearTest.Test/ES3_Metal__scissored_clear_cds_mask_cs
DepthStencilFormatsTest.DepthTextureRender/ES2_Metal
DepthStencilFormatsTest.DepthBuffer16/ES2_Metal
DepthStencilFormatsTest.DepthBuffer24/ES2_Metal
DepthStencilFormatsTest.VerifyDepth32UploadData/ES2_Metal
DepthStencilFormatsTest.VerifyDepth16UploadData/ES2_Metal
TinyDepthStencilWorkaroundTest.DepthTexturesStick/ES3_Metal
DiscardFramebufferEXTTest.ClearDepthThenDrawWithDepthTestThenDiscard/ES2_Metal
DiscardFramebufferEXTTest.ClearDepthThenDrawWithDepthTestThenDiscard/ES3_Metal
Bug: angleproject:5505
Change-Id: I9b52bb46569678636afc6a243bfa8779f101b6af
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3097161
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Kenneth Russell <kbr@chromium.org>
|
|
|
d20afde4
|
2021-08-05T23:06:44
|
|
Vulkan: Retain precision of uniform variables
When gathering in a uniform block, the precision of default uniforms was
dropped.
Bug: angleproject:4889
Bug: angleproject:6132
Change-Id: Ie6a8e2e7cef7cd3808cee08d20f886fc4e762cd5
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3076124
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
|
1654537d
|
2021-08-06T17:49:52
|
|
Fix default and driver uniforms' binding indices.
These abstractions were introduced in the SymbolTable during
upstreaming of the direct-to-Metal compiler and were accidentally
swapped, leading to widespread test failures.
Fix the comment reference in ContextMtl's DriverUniforms struct. Add
missing emulatedInstance field, and add a TODO about its missing
support in direct-to-Metal translator. Define it in SPIR-V Metal
translator; the driver uniforms for the two must match.
Bug: angleproject:6259
Change-Id: I0330ab4efca5b64a749b9cca90564fdbc84d7a80
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3078998
Commit-Queue: Kenneth Russell <kbr@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
d191bc32
|
2021-08-06T12:59:39
|
|
Reintroduce TType::isRank0.
This concept was used in the direct-to-Metal compiler, and its
replacement with isScalar wasn't completely correct. Fixes several
assertion failures in tests.
Bug: angleproject:4889
Bug: angleproject:5505
Change-Id: I9b8800ed50591560c8e53faa81eca581c4f07a02
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3078474
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Kenneth Russell <kbr@chromium.org>
|
|
|
63248e7f
|
2021-08-05T23:49:23
|
|
Translator: Fix dead-code-elimination corner case
The DCE code had a corner case bug where a switch case containing
multiple DCE'ed `break`s followed by another case would cause some of
the DCE'ed statements in the first case to be doubly pruned, failing on
an assertion.
This was due to the fact that visitBlock() was asking traversal to
continue if a new case was visited while pruning nodes, but the
traversal does not take into account that some statements need to be
skipped and so would also visit the pruned statements. If the pruned
statements contain a branch instruction, they get re-pruned.
The visitBlock() function that does the pruning is reworked so that it
more clearly traverses the statements.
Bug: chromium:1237200
Change-Id: Ib078c2ea73ade756c7d7ef5a5c489fa53c39f352
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3077659
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Tim Van Patten <timvp@google.com>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
0cd99779
|
2021-08-05T10:35:10
|
|
Translator: Propagate precision to children nodes
Similarly to glslang, when the precision of a node is determined,
propagate that precision to any of its children that doesn't already
have a precision. Ultimately these should only include
TIntermConstantUnion nodes.
Bug: angleproject:4889
Bug: angleproject:6132
Change-Id: I121231d04c7cf92fc3f07716019ffe88eca48b88
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3075126
Reviewed-by: Tim Van Patten <timvp@google.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
2e8a2d3c
|
2021-08-04T11:04:44
|
|
Translator: Helpers to derive precision
Bug: angleproject:4889
Bug: angleproject:6132
Change-Id: I75e2084c76139ccb24b266f262d5e5597d8aa4ca
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3071599
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Tim Van Patten <timvp@google.com>
|
|
|
061188a7
|
2021-08-04T10:07:47
|
|
Translator: General clean up
General clean up done as part of other changes, split to simplify
review.
Bug: angleproject:4889
Bug: angleproject:6132
Change-Id: Iade9954d187a759be9edd0e3754be007f4133c56
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3071598
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Tim Van Patten <timvp@google.com>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
6fe91881
|
2021-08-02T11:13:13
|
|
Translator: Simplify RecordConstantPrecision
Just before ESSL output, this transformation made some constants
specified through temp variables such that their precision can be
specified. This was done in cases where the precision actually affected
the expression being evaluated.
This transformation is simplified by declaring constants more
conservatively as variables, and not stringently attempting to identify
whether that's necessary. This is still not done in trivial cases.
This change is in preparation for another change that overhauls
precision propagation through the nodes. This change removes all
references to TIntermTyped::getPrecision() outside the precision
promotion code itself.
Bug: angleproject:4889
Change-Id: I135b2b3241aa2f76219b209241c3844f1d09ccd2
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3067948
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Tim Van Patten <timvp@google.com>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
798c5200
|
2021-07-31T22:17:48
|
|
Translator: Make precise-ness a property of AST nodes
Instead of creating a "set of precise nodes", the nodes themselves are
marked precise. This change is preparation for another change that
overhauls precision propagation through the nodes, and which would also
be storing the promoted precision in the nodes.
Bug: angleproject:4889
Change-Id: Ieec8864be5cd2e95ed6fb4d90b367946d9f33fa1
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3065569
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Tim Van Patten <timvp@google.com>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
1ce78397
|
2021-07-30T16:51:45
|
|
Remove support for WEBGL_debug_shader_precision
This extension was rejected, and the implementation was hacky. This
clean up is part of an ongoing work to improve precision handling.
Bug: angleproject:6059
Change-Id: If08581ec6f19cf1698ffa3dd6d248dc5e68a1d31
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3064303
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
09d5047b
|
2021-08-03T01:31:03
|
|
GL: Remove the rewriteVectorScalarArithmetic workaround
This workaround was implemented for a bug in Nvidia driver 387 which is
end-of-life. More recent driver branches (390 and 4XX) are fixed, and
so this workaround is no longer required. The implementation of the
workaround itself could introduce bugs.
Bug: chromium:772651
Bug: chromium:1201084
Change-Id: I3db179eb90d9124235bdad2daacc712302906d8e
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3067952
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
Reviewed-by: Kenneth Russell <kbr@chromium.org>
|
|
217acac2
|
2021-07-21T12:42:56
|
|
Vulkan: Add support for EXT_primitive_bounding_box
1. Added support for EXT_primitive_bounding_box extension
2. Renamed shader variable gl_BoundingBoxEXT[] to ANGLEBoundingBox[]
Bug: angleproject:3576
Test: dEQP-GLES31.functional.primitive_bounding_box.*
Change-Id: I15fa9af50c6fd8e86d225670ddd8eb39f6e65d35
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3053618
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
f9942640
|
2021-08-03T11:45:31
|
|
Translator: Fix fuzzer crashing on assertion
An assertion was triggered by the fuzzer when generating an invalid
shader that redeclared a built-in. The assertion itself was unnecessary
and the parser would appropriately generate an error if that assertion
would have fired.
Bug: angleproject:4889
Bug: chromium:1235229
Change-Id: I294b2e329491c75d4e87622b602570252555869f
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3063910
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
|
a65b4056
|
2021-07-30T23:00:57
|
|
Reland "Translator: Clean up type cloning"
This is a reland of f016c4352f5203c10511df078b1ed5359afc1b35
Original change's description:
> Translator: Clean up type cloning
>
> TType has a constructor that aids cloning + helpers to convert between
> types. A number of places where a type is constructed from the
> information gathered from another type is changed to clone the type and
> then use one of these helpers.
>
> This clean up is part of an ongoing work to improve precision handling.
> This change removes many references to TType::getPrecision().
>
> Bug: angleproject:4889
> Change-Id: Ib85659ab5363b56ad298f8648fca856edc1ebf8b
> Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3063944
> Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
> Reviewed-by: Jamie Madill <jmadill@chromium.org>
> Reviewed-by: Tim Van Patten <timvp@google.com>
Bug: angleproject:4889
Change-Id: I35772f178eb4f6cf2b64bfeb37a4a144acdb4daf
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3067802
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
c5c8a25a
|
2021-08-03T14:40:33
|
|
Revert "Translator: Clean up type cloning"
This reverts commit f016c4352f5203c10511df078b1ed5359afc1b35.
Reason for revert: Suspect for crbug.com/1236060
Original change's description:
> Translator: Clean up type cloning
>
> TType has a constructor that aids cloning + helpers to convert between
> types. A number of places where a type is constructed from the
> information gathered from another type is changed to clone the type and
> then use one of these helpers.
>
> This clean up is part of an ongoing work to improve precision handling.
> This change removes many references to TType::getPrecision().
>
> Bug: angleproject:4889
> Change-Id: Ib85659ab5363b56ad298f8648fca856edc1ebf8b
> Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3063944
> Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
> Reviewed-by: Jamie Madill <jmadill@chromium.org>
> Reviewed-by: Tim Van Patten <timvp@google.com>
Bug: angleproject:4889
Change-Id: I9b5ecacc410b41be382232292ceae2f7ea7906c7
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3064393
Bot-Commit: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com>
Commit-Queue: Austin Eng <enga@chromium.org>
|
|
|
f016c435
|
2021-07-30T23:00:57
|
|
Translator: Clean up type cloning
TType has a constructor that aids cloning + helpers to convert between
types. A number of places where a type is constructed from the
information gathered from another type is changed to clone the type and
then use one of these helpers.
This clean up is part of an ongoing work to improve precision handling.
This change removes many references to TType::getPrecision().
Bug: angleproject:4889
Change-Id: Ib85659ab5363b56ad298f8648fca856edc1ebf8b
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3063944
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Tim Van Patten <timvp@google.com>
|
|
|
aca9f2e2
|
2021-07-29T17:49:26
|
|
Vulkan: SPIR-V Gen: Fix constructors with uniform parameters
When the constructor contains parameters from interface blocks, they may
have SPIR-V type specializations that was not previously accounted for.
The code is made more robust by retrieving the column/component types
when generating constructors through TTypes instead of modifying
SpirvTypes directly.
Bug was caught by the WebGL CTS.
Bug: angleproject:4889
Change-Id: I6c9df17a97fac8622dea55ed67c7bae1ce16d1bd
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3061640
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Tim Van Patten <timvp@google.com>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
d4d6c23c
|
2021-07-31T11:41:29
|
|
Fix a recently-introduced case of -Wunreachable-code-aggressive.
Also enable this and -Wunused-but-set-variable to avoid regressing them.
Bug: chromium:1066980
Change-Id: I1fd7a57c2bf14513d657b69b70e35ad727c75ffd
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3065355
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
ae8ef039
|
2021-07-30T00:43:14
|
|
Translator: Correctly respect precision of gl_FragColor/Data
These builtins are declared with a mediump precision. However, during
translation to Vulkan GLSL, the precision was not output and as such
glslang treated them as highp.
Discovered through comparison with SPIR-V output from ANGLE which
included more RelaxedPrecision operations.
Bug: angleproject:4889
Change-Id: I7974478e87ad20115345d679f8f035492be1f349
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3062204
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Tim Van Patten <timvp@google.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
|
8fb5c2f2
|
2021-07-29T23:41:08
|
|
Vulkan: SPIR-V Gen: Fix OpImage* on multisampled images
Lod should not be implicitly specified on multisampled images per spec.
This change fixes a crash on Intel/Mesa with the
TextureSampleShadingTest.Basic/ES3_1_Vulkan test.
Old spirv-val catches this bug, but ToT has regressed this:
https://github.com/KhronosGroup/SPIRV-Tools/issues/4424
Bug: angleproject:4889
Change-Id: Ic2a661ce5264beb1f47f4e52706fe6556176628d
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3062203
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Tim Van Patten <timvp@google.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
|
ebe943e2
|
2021-07-29T11:19:33
|
|
Vulkan: SPIR-V Gen: Fix gl_Clip/CullDistance
These built-ins can be redeclared in the shader. The translator took
these redeclarations (and gl_LastFragData) as UserDefined symbols.
There were a number of hacks such as in name generation and
CollectVariables, to special-case these redeclarations.
This change instead makes sure that these variables continue to be
considered built-ins with the appropriate qualifiers. A number of fixes
are made accordingly:
- Hacks are removed.
- In fragment shaders, ANGLEClipDistance was initialized with
gl_ClipDistance for further use by the shader. The code generation
however mistakenly produced `gl_ClipDistance[0] =
ANGLEClipDistance[0];`, which caused compilation failures by glslang,
but passed the tests accidentally because they expected link failures
(see next item).
- CollectVariables is fixed to correctly collect gl_Clip/CullDistance
in fragment shaders; previously they were collected as output
varyings, and therefore the aforementioned link error was not produced
in the tests after the compilation error was fixed.
Additional fixes:
- The transformation of gl_Clip/CullDistance was always ever done on one
of them due to misplaced breaks in the loop that detected their
presence.
- The transformation of gl_CullDistance was skipped when it was not
redeclared.
Validation is added to make sure these built-ins always have the correct
qualifier even when redeclared. SPIR-V gen support for these variables
have been fixed as well.
Bug: angleproject:4889
Change-Id: Ic8bf2c508357035cb937ef41a28ae22ffc533ebe
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3059921
Reviewed-by: Tim Van Patten <timvp@google.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
a83c9108
|
2021-07-29T17:13:41
|
|
Vulkan: SPIR-V Gen: Fix nested ternary operators
The OpPhi instruction used to implement ternaries referenced to the
blocks pregenerated for the true and false cases. However, if there
were other code nested inside that could generate blocks (other
ternary expressions or short-circuits), that would be wrong.
Instead, the last block id is taken every time the true or false case is
entirely visited (similar to how it's done in the short-circuit
implementation).
Bug was caught by the WebGL CTS.
Bug: angleproject:4899
Change-Id: I71be86989dfe9596a13940ce657a7e67849e86bf
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3061639
Reviewed-by: Tim Van Patten <timvp@google.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
2b23aae4
|
2021-07-29T16:39:28
|
|
Vulkan: SPIR-V Gen: Support vec(..., mat)
Matrices specified in vector constructor arguments were not correctly
handled w.r.t to casting (caught by the WebGL CTS).
Bug: angleproject:4889
Change-Id: I70253f049a651fc5f439b56d1e4d8f785813b605
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3061638
Reviewed-by: Tim Van Patten <timvp@google.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
aec5e65c
|
2021-07-28T00:36:12
|
|
Get direct-to-Metal backend to run angle_end2end_tests.
Cherry-pick nameless struct fix from Apple in
https://bugs.webkit.org/show_bug.cgi?id=227482 .
Fix SeparateCompoundStructDeclarations pass to stop generating
multiple declarations; thanks syoussefi@ for advice.
Incorporate additional passes from TranslatorVulkan
(MonomorphizeUnsupportedFunctionsInVulkanGLSL,
RewriteArrayOfArrayOfOpaqueUniforms,
SeparateStructFromUniformDeclarations) needed by RewriteStructSamplers
pass in TranslatorMetalDirect. Fixes many assertion failures in GLSL
tests. Moved these passes out of tree_ops/vulkan. Thanks again to
syoussefi@ for advice and help.
Disable a validation check related to the RewritePipelines pass. Skip
two tests that were failing for other reasons.
With these changes, angle_end2end_tests runs to completion when the
direct-to-Metal backend is turned on. There are still ~1300 failures
of the ~4000 tests which will be investigated next.
Bug: angleproject:5505
Change-Id: Ibca77822543e8e8e8d2a8c862e92cdf74bfa3545
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3058524
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Kenneth Russell <kbr@chromium.org>
|
|
|
96ccf3f4
|
2021-07-29T12:38:21
|
|
Translator: Consolidate EvqFragDepth and EvqFragDepthEXT
They were the same thing. Fixes SPIR-V generation which only handled
EvqFragDepth.
Bug: angleproject:4889
Change-Id: Id2c35f4c5a26e43ad16a420cc0306756a1add082
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3059922
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Tim Van Patten <timvp@google.com>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
3036e090
|
2021-07-29T15:55:39
|
|
Vulkan: Direct SPIR-V Gen: Support the precise keyword
The precise keyword is different in GLSL in that it defines what
arithmetic operations _that have led to the value being assigned to a
variable_ should be done precisely (i.e. not "contracted").
A tree traverser is implemented that detects precise access chains and
applies precise-ness to the right hand side of assignment expressions to
said access chains. This is only done if the shader uses the precise
keyword in the first place. The algorithm for this is inspired by the
implementation in glslang.
This change additionally:
- Fixes parser to allow precise on function parameters
- Fixes GLSL code generation to output precise on struct members and
function parameters.
Bug: angleproject:4889
Change-Id: Ie3808c3c8c08da308e88af20f5f90379d9d14d47
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3056369
Reviewed-by: Tim Van Patten <timvp@google.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
7c6da493
|
2021-07-26T21:24:25
|
|
Translator: Make sure built-in variables are consistent
Some transformations left the tree in an inconsistent state, for example
if it used gl_SampleMask from ES3.1+OES_sample_variables and the
transformation added a usage of gl_SampleMask from ES3.2. The offending
transformations are fixed and AST validation is improved to catch such
errors.
Bug: angleproject:4889
Change-Id: I9d9ea5bb43a9408dd4c6dc3e89ec20d60dfeff73
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3054613
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Tim Van Patten <timvp@google.com>
|
|
|
c8c9e15b
|
2021-07-26T22:51:49
|
|
Translator: Fix sizing of tessellation shader arrays
If layout(vertices=N) is specified after implicitly sized array
declarations, the correct size was not being specified on them.
Bug: angleproject:4889
Change-Id: I86ccf45a7f264b493a137b87316eff744079ed6c
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3054614
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Tim Van Patten <timvp@google.com>
|
|
|
5bda5f91
|
2021-07-26T20:37:52
|
|
Fix some instances of -Wunreachable-code-aggressive.
Bug: chromium:1066980
Change-Id: I1e769b78c51d939067e8855beab32ec7c006c00e
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3055794
Reviewed-by: Nico Weber <thakis@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Nico Weber <thakis@chromium.org>
|
|
|
b8d6f8aa
|
2021-07-22T22:31:29
|
|
Translator: Facilitate integration in FireFox
Instead of relying on is_apple to build workarounds applied on apple
hardware, a new build argument is added for this purpose. This allows
FireFox to more easily include these files in their build system (which
builds every workaround on every platform).
The workarounds are reorganized too and moved from tree_ops/gl/mac to
tree_ops/apple. RewriteRowMajorMatrices is moved there too as it's now
used by both gl and metal. This workaround is large, and removing it
from non-apple builds improves binary size (~40KB on Linux).
Bug: angleproject:6188
Change-Id: I8e1dab44d0235bb6b8341d53721bdd0d157d46c6
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3047385
Reviewed-by: Jeff Gilbert <jgilbert@mozilla.com>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
dbe986d7
|
2021-07-26T01:43:44
|
|
Vulkan: SPIR-V Gen: Fixes to tessellation shader support
- TCS barrier() built-in is correctly looked up
- Patch decorations are fixed
Bug: angleproject:4889
Change-Id: I1befcaca85a965c9d46b1b41ec74597de44bf2b2
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3052687
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Tim Van Patten <timvp@google.com>
|
|
|
acfd63a4
|
2021-07-26T01:02:11
|
|
Translator: Remove GLSL rules from the ESSL-only symbol table
Shaves ~36KB off of ANGLE's binary size on Android.
Bug: chromium:1084580
Change-Id: I4711c6bd28437a43b7dc63be4cfcdaed7ae11ba2
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3052686
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
63fa7080
|
2021-07-25T23:19:02
|
|
Vulkan: SPIR-V Gen: Fix bugs with ES3.1 texture* builtins
- Add missing ImageCubeArray capability for image cube arrays
- Remove extra Lod image operand for sampler/image buffers
Bug: angleproject:4889
Change-Id: I96545092528b7dcecf902561e194b17636217ef4
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3052683
Reviewed-by: Tim Van Patten <timvp@google.com>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
715bab1f
|
2021-07-25T23:35:17
|
|
Vulkan: SPIR-V Gen: Fixes to geometry shader support
- max_vertices=0 is invalid in SPIR-V and is fixed (similarly to
anglebug.com/5411)
- gl_PrimitiveID used in fragment shaders requires the Geometry
capability.
Bug: angleproject:4889
Change-Id: If8c6a9d455d0582d61c3c52b5f773bcb5ca64b53
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3052684
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Tim Van Patten <timvp@google.com>
|
|
|
ff7eafb5
|
2021-07-25T23:05:26
|
|
Vulkan: SPIR-V Gen: Fix .length() vs ssbo arrays
Bug: angleproject:4889
Change-Id: Ib490a46fabf058064fc1b18d2c084a4bc5f9277d
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3052682
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Tim Van Patten <timvp@google.com>
|
|
|
b19fbbda
|
2021-07-25T21:57:12
|
|
Vulkan: SPIR-V Gen: Support multiview
Bug: angleproject:4889
Change-Id: Ibe66f53357473dbb4435af58775b524d83ed250b
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3052675
Reviewed-by: Tim Van Patten <timvp@google.com>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
939ea5f4
|
2021-07-23T15:53:17
|
|
Vulkan: SPIR-V Gen: Fixes to tessellation builtins
On TCS, TES and FS, gl_PrimitiveID is actually an input. Additionally,
`patch` variables (including gl_TessLevel* built-ins) need to be
decorated with Patch.
Bug: angleproject:4889
Change-Id: I326ec4b0c011fe8eaafbad4f8bcb0f537005f96f
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3048320
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Tim Van Patten <timvp@google.com>
|
|
|
9416e84e
|
2021-07-23T14:53:43
|
|
Vulkan: SPIR-V Gen: Add missing DepthReplacing execution mode
Shaders writing to gl_FragDepth need this execution mode specified.
Bug: angleproject:4889
Change-Id: I315c90f9161f4d3d51b1c856f1364fb26a4bd404
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3048317
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Tim Van Patten <timvp@google.com>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
de8dc2c9
|
2021-07-23T15:45:28
|
|
Vulkan: SPIR-V Gen: Support OES_sample_variables
Built-ins from this extension are now decorated.
Bug: angleproject:4889
Change-Id: I12f80d67e3595f94fc64bdc8cb3031d36b98c279
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3048319
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Tim Van Patten <timvp@google.com>
|
|
|
8bf2e5ec
|
2021-07-23T16:54:16
|
|
Vulkan: SPIR-V Gen: Fix switch with default at the end
A minor bug was making a switch with default at the end to jump to the
merge block instead of the default block.
Bug: angleproject:4889
Change-Id: Ied434fab949b10d45a0db1242c1b8535a5f4f773
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3048321
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Tim Van Patten <timvp@google.com>
|
|
|
82d7d888
|
2021-07-22T21:46:21
|
|
Vulkan: SPIR-V Gen: Fix bugs with texture* builtins
With this change, deqp ES3 tests pass.
Bug: angleproject:4889
Change-Id: Ica158d4e9012c7bb3e458da62b4a7f92aee5c6de
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3047380
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Tim Van Patten <timvp@google.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
|
31bc9226
|
2021-07-23T15:16:27
|
|
Vulkan: SPIR-V Gen: Fix unpack* built-ins
The parameter to these built-ins was getting extended to a vector.
Bug: angleproject:4889
Change-Id: Idddce008c9b4f6bf9205b1e20f6e89ef657ea3a1
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3048318
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Tim Van Patten <timvp@google.com>
|
|
|
6b5cec15
|
2021-07-22T17:01:41
|
|
Vulkan: SPIR-V Gen: Fixes to scalar->vector promotion
With the ternary operator, the condition may be a bool-in-interface
block that needs to be cast. This cast is now part of
createConstructorVectorFromScalar so it can't be missed.
Bug: angleproject:4889
Change-Id: I0e24cd4127301d33a3ac677ccaf560c4468e0799
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3047379
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Tim Van Patten <timvp@google.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
|
d0c03ff4
|
2021-07-22T14:55:51
|
|
Vulkan: SPIR-V Gen: Fix lvalues passed to in/inout parameters
In GLSL, these values are semantically copied when passed to a function
as an in or inout parameter. For example in:
bool f(inout vec4 a, inout vec4 b)
{
a = vec4(0);
return all(equal(a, b));
}
var = vec4(1);
bool result = f(var, var);
result is expected to be false. In SPIR-V, every parameter is
semantically passed by "reference".
glslang conservatively uses temporaries to pass to functions. An
optimization in ANGLE didn't create temporaries for unindexed lvalues,
which did not take into account the above fact. This optimization is
limited to out parameters now.
Bug: angleproject:4889
Change-Id: Ie1b4b1cecba847ba63d5810d01d0856823b89ddc
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3046103
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Tim Van Patten <timvp@google.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|