|
d63243f5
|
2015-10-29T14:23:37
|
|
renderergl_utils: Require explicit location qualifier for OpenGLES3
BUG=525930
Change-Id: I591308078c5a392a527291419567a11241a8be17
Reviewed-on: https://chromium-review.googlesource.com/309636
Tested-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tryjob-Request: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
|
|
91dbc188
|
2015-06-17T16:19:29
|
|
Enable needed extensions in generated GLSL shaders and call them when possible.
reland: use the defined preprocessor function.
BUG=angleproject:1044
Change-Id: I6171708a2d55ec085ba2c598a1a863779842da6c
Reviewed-on: https://chromium-review.googlesource.com/311270
Tryjob-Request: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Tested-by: Geoff Lang <geofflang@chromium.org>
|
|
15200047
|
2015-11-04T16:56:31
|
|
Fix issues in comma operator parsing
Always qualify comma operator result with EvqTemporary in ESSL3, as
specified.
Also, it is possible that in the future some expressions are qualified
as EvqConst but they'd still have side effects, in which case discarding
them when they're the left operand of the comma operator would be wrong.
This would be the case if ANGLE allowed "(a = b).length()" for example.
For this reason it is better to check whether the left node has side
effects, rather than check its const qualification, and only discard it
if it doesn't.
Also, Intermediate::addComma() never returns null, so there's no need to
check the result.
BUG=angleproject:1201
TEST=WebGL conformance tests
conformance/glsl/misc/sequence-operator-returns-constant.html
conformance2/glsl3/sequence-operator-returns-non-constant.html
Change-Id: Ibfbd92baa4910b14c0dc8f8a3c3008440d191cd6
Reviewed-on: https://chromium-review.googlesource.com/311171
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tryjob-Request: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Tested-by: Olli Etuaho <oetuaho@nvidia.com>
|
|
|
0b2d2dcf
|
2015-11-04T16:35:32
|
|
Move comma operator parsing to ParseContext
This change is pure refactoring. It will help with fixing comma operator
return value qualifier bug.
BUG=angleproject:911
BUG=angleproject:1201
Change-Id: I481c337adbaf789cc959c8a1106f99ad7275e1e5
Reviewed-on: https://chromium-review.googlesource.com/311170
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Tested-by: Olli Etuaho <oetuaho@nvidia.com>
|
|
|
cdeafd20
|
2015-11-05T14:59:18
|
|
Fix missing init when parsing storage qualifier fails
When parsing a storage qualifier results in an error, the parser should
not leave the type uninitialized. Otherwise after recovering from the
error the parsed value could contain uninitialized memory, which could
cause flaky asserts in further parsing.
TEST=dEQP-GLES2.functional.shaders.functions.*
BUG=angleproject:1200
Change-Id: Ia3add76df243898e35fee97d01555b953cff8379
Reviewed-on: https://chromium-review.googlesource.com/311140
Tested-by: Olli Etuaho <oetuaho@nvidia.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Tested-by: Corentin Wallez <cwallez@chromium.org>
|
|
|
e7fcf1e3
|
2015-11-05T21:23:53
|
|
Revert "Enable needed extensions in generated GLSL shaders and call them when possible."
This reverts commit 233b334213c1a4853e81f71ba90aa223b2a68dce.
Change-Id: Id99024b736324ad030ce63c5d0baae32b6d7f54b
Reviewed-on: https://chromium-review.googlesource.com/311181
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Tested-by: Geoff Lang <geofflang@chromium.org>
|
|
|
233b3342
|
2015-06-17T16:19:29
|
|
Enable needed extensions in generated GLSL shaders and call them when possible.
BUG=angleproject:1044
Change-Id: Ife94d2258fb912974aa97484a0e30f289dd96734
Reviewed-on: https://chromium-review.googlesource.com/278324
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Geoff Lang <geofflang@chromium.org>
|
|
|
b1edc4f5
|
2015-11-02T17:20:03
|
|
Accept const array initialization in shader parsing
Array constructors are not folded, unlike all other constant expressions.
Change initializer parsing path so that it accepts constant initializers
whether they are folded or not.
Some parts need to be adapted to work with expressions that are qualified
as constant but that are not necessarily folded:
1. Identifier parsing
2. Indexing parsing
3. Field selection parsing
4. HLSL output for variable declarations
5. Determining unary operator result type
6. Determining binary operator result type
7. Determining built-in function call result type
8. Determining ternary operator result type
Corner cases that are not supported yet:
1. Using array constructors inside case labels
2. Using array constructors inside array size expressions
3. Detecting when a negative constant expression containing an array
constructor is used to index an array
In these cases being able to constant fold the expression is essential to
validating that the code is correct, so they require a more sophisticated
solution. For now we keep the old code that rejects the shader if ANGLE
hasn't been able to constant fold the case label or array size. In case of
indexing an array with a negative constant expression containing an array
constructor, ANGLE will simply treat it as a non-constant expression.
BUG=541551
BUG=angleproject:1094
TEST=dEQP-GLES3.functional.shaders.constant_expressions.* (all pass),
angle_unittests
Change-Id: I0cbc47afd1651a4dece3d68acf7ec72a01fdf047
Reviewed-on: https://chromium-review.googlesource.com/310231
Tested-by: Olli Etuaho <oetuaho@nvidia.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
|
82c29ed2
|
2015-11-03T13:06:54
|
|
Move variable identifier parsing to ParseContext
This change is purely refactoring existing functionality to make it
easier to change variable identifier parsing.
BUG=angleproject:911
BUG=541551
Change-Id: I282fbb66d40cd71fa0f4804e4a8e893744cd6bfc
Reviewed-on: https://chromium-review.googlesource.com/309724
Tested-by: Olli Etuaho <oetuaho@nvidia.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
|
|
|
5d91dda9
|
2015-06-18T15:47:46
|
|
Remove dynamic indexing of matrices and vectors in HLSL
Re-re-relanding after clang warning fix.
Re-re-landing with fix to setting qualifiers on generated nodes. The
previous version failed when a uniform was indexed, because it would
set the uniform qualifier on some of the generated nodes and that
interfered with the operation of UniformsHLSL.
Re-landing after fixing D3D9 specific issues.
HLSL doesn't support dynamic indexing of matrices and vectors, so replace
that with helper functions that unroll dynamic indexing into switch/case
and static indexing.
Both the indexed vector/matrix expression and the index may have side
effects, and these will be evaluated correctly. If necessary, index
expressions that have side effects will be written to a temporary
variable that will replace the index.
Besides dEQP tests, this change is tested by a WebGL 2 conformance test.
In the case that a dynamic index is out-of-range, the base ESSL 3.00 spec
allows undefined behavior. KHR_robust_buffer_access_behavior adds the
requirement that program termination should not occur and that
out-of-range reads must return either a value from the active program's
memory or zero, and out-of-range writes should only affect the active
program's memory or do nothing. This patch clamps out-of-range indices so
that either the first or last item of the matrix/vector is accessed.
The code is not transformed in case the it fits within the limited subset
of ESSL 1.00 given in Appendix A of the spec. If the code isn't within
the restricted subset, even ESSL 1.00 shaders may require this
workaround.
BUG=angleproject:1116
TEST=dEQP-GLES3.functional.shaders.indexing.* (all pass after change)
WebGL 2 conformance tests (glsl3/vector-dynamic-indexing.html)
Change-Id: I9f6d7c7ecda8ac4dc3c30b39e15a9a0b5381c5a8
Reviewed-on: https://chromium-review.googlesource.com/310010
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Tested-by: Geoff Lang <geofflang@chromium.org>
|
|
a5f64de7
|
2015-10-30T12:31:00
|
|
Revert "Remove dynamic indexing of matrices and vectors in HLSL"
Failing build on Clang-win:
..\..\third_party\angle\src\compiler\translator\RemoveDynamicIndexing.cpp(128,43) : error: expected '(' for function-style cast or type construction
fieldType.setPrimarySize(unsigned char(indexedType.getRows()));
~~~~~~~~ ^
BUG=angleproject:1116
This reverts commit 7535b761dd4740c8e76b888d7c58c7cbeefd2083.
Change-Id: I7b502e3dcd45e17b7ed88fec18be702614d9ac65
Reviewed-on: https://chromium-review.googlesource.com/309772
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>
|
|
|
f541f529
|
2015-10-13T12:21:01
|
|
Fix parsing integers larger than 0x7FFFFFFF
Parsing should accept all values between 0 and 0xFFFFFFFF as specified in
ESSL 3.00 section 4.1.3.
When a signed literal is parsed, it's interpreted as if it specifies the
bit pattern of a two's complement integer. For example, parsing
"0xFFFFFFFF" results in -1. Decimal literals behave the same way, so for
example parsing "3000000000" results in -1294967296.
This change affects parsing of literals in ESSL 1.00 as well. In ESSL
3.00, an out-of-range integer literal now generates a compiler error.
Unit tests are added based on examples in the ESSL 3.00 spec and one
example in GLSL 4.5 spec that ESSL should match.
BUG=541550
TEST=angle_unittests
Change-Id: I82f8ef5cfa2881019a3f80d77ff99707d61c000d
Reviewed-on: https://chromium-review.googlesource.com/305420
Reviewed-by: Zhenyao Mo <zmo@chromium.org>
Tested-by: Olli Etuaho <oetuaho@nvidia.com>
Reviewed-by: Corentin Wallez <cwallez@google.com>
|
|
|
ae69d7e1
|
2015-10-07T17:19:50
|
|
Output layout qualifiers in GLSL output
Output layout qualifiers for vertex shader attributes and fragment
shader outputs. This applies both to ESSL output and GLSL output. If
the source shader is in ESSL1, the code has no effect since there's
no layout information in the AST.
BUG=525930
TEST=angle_unittests, angle_end2end_tests
Change-Id: I48b3aa56116c15d11599b030eed4c45be2c8fc7e
Reviewed-on: https://chromium-review.googlesource.com/304550
Tryjob-Request: Jamie Madill <jmadill@chromium.org>
Tested-by: Olli Etuaho <oetuaho@nvidia.com>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@google.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
|
7535b761
|
2015-06-18T15:47:46
|
|
Remove dynamic indexing of matrices and vectors in HLSL
Re-re-landing with fix to setting qualifiers on generated nodes. The
previous version failed when a uniform was indexed, because it would
set the uniform qualifier on some of the generated nodes and that
interfered with the operation of UniformsHLSL.
Re-landing after fixing D3D9 specific issues.
HLSL doesn't support dynamic indexing of matrices and vectors, so replace
that with helper functions that unroll dynamic indexing into switch/case
and static indexing.
Both the indexed vector/matrix expression and the index may have side
effects, and these will be evaluated correctly. If necessary, index
expressions that have side effects will be written to a temporary
variable that will replace the index.
Besides dEQP tests, this change is tested by a WebGL 2 conformance test.
In the case that a dynamic index is out-of-range, the base ESSL 3.00 spec
allows undefined behavior. KHR_robust_buffer_access_behavior adds the
requirement that program termination should not occur and that
out-of-range reads must return either a value from the active program's
memory or zero, and out-of-range writes should only affect the active
program's memory or do nothing. This patch clamps out-of-range indices so
that either the first or last item of the matrix/vector is accessed.
The code is not transformed in case the it fits within the limited subset
of ESSL 1.00 given in Appendix A of the spec. If the code isn't within
the restricted subset, even ESSL 1.00 shaders may require this
workaround.
BUG=angleproject:1116
TEST=dEQP-GLES3.functional.shaders.indexing.* (all pass after change)
WebGL 2 conformance tests (glsl3/vector-dynamic-indexing.html)
Change-Id: I16119f9092360fb72798f9550a6f4d3cfffdc92f
Reviewed-on: https://chromium-review.googlesource.com/308790
Reviewed-by: Zhenyao Mo <zmo@chromium.org>
Tested-by: Zhenyao Mo <zmo@chromium.org>
|
|
|
b066669d
|
2015-10-26T10:38:18
|
|
Revert "Remove dynamic indexing of matrices and vectors in HLSL"
This reverts commit 3766a40d6fda7e7190514ab7838a3f37169d863f.
This CL was causing crashes in UniformHLSL.cpp, where an internal
uniform "base" was attempted to be declared in HLSL. Was crashing
on an external WebGL 3D canvas page (http://www.taccgl.org/?dbg=t).
BUG=546686
Original commit message:
Re-landing after fixing D3D9 specific issues.
HLSL doesn't support dynamic indexing of matrices and vectors, so replace
that with helper functions that unroll dynamic indexing into switch/case
and static indexing.
Both the indexed vector/matrix expression and the index may have side
effects, and these will be evaluated correctly. If necessary, index
expressions that have side effects will be written to a temporary
variable that will replace the index.
Besides dEQP tests, this change is tested by a WebGL 2 conformance test.
In the case that a dynamic index is out-of-range, the base ESSL 3.00 spec
allows undefined behavior. KHR_robust_buffer_access_behavior adds the
requirement that program termination should not occur and that
out-of-range reads must return either a value from the active program's
memory or zero, and out-of-range writes should only affect the active
program's memory or do nothing. This patch clamps out-of-range indices so
that either the first or last item of the matrix/vector is accessed.
The code is not transformed in case the it fits within the limited subset
of ESSL 1.00 given in Appendix A of the spec. If the code isn't within
the restricted subset, even ESSL 1.00 shaders may require this
workaround.
BUG=angleproject:1116
TEST=dEQP-GLES3.functional.shaders.indexing.* (all pass after change)
WebGL 2 conformance tests (glsl3/vector-dynamic-indexing.html)
Change-Id: I1d4b2e3888e91af7d5eebf743d12778698b6b903
Reviewed-on: https://chromium-review.googlesource.com/308770
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>
|
|
|
a6996685
|
2015-10-12T14:32:30
|
|
Automatically enable highp in fragment shaders on ESSL3
Most code using the translator already enables highp with the resources
flag when a shader spec that accepts ESSL3 is used, but for example the
shader_translator utility doesn't. This fix makes sure that highp is
always enabled when a fragment shader written in ESSL3 or newer is being
compiled. This will make shader_translator easier to use for testing
ESSL3 shaders.
BUG=541550
TEST=angle_unittests
Change-Id: Ia1911677c55f3c5d921829a8cbb808847ac8b636
Reviewed-on: https://chromium-review.googlesource.com/305190
Tryjob-Request: Olli Etuaho <oetuaho@nvidia.com>
Reviewed-by: Zhenyao Mo <zmo@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Olli Etuaho <oetuaho@nvidia.com>
|
|
|
2f6ddf31
|
2015-09-22T16:10:07
|
|
Allow double underscore in macro names
Double underscore is allowed according to GLSL ES 3.10, and based on
Khronos discussions the intent is that this should also apply to older
specs. The dEQP tests also check this, and WebGL tests that check the
opposite were recently removed. The error is changed into a warning.
BUG=angleproject:989
TEST=angle_unittests
dEQP-GLES3.functional.shaders.preprocessor.* (2 tests start passing)
dEQP-GLES2.functional.shaders.preprocessor.* (2 tests start passing)
Change-Id: I582c01b4adc8fc416354351e02b776f2cc602408
Reviewed-on: https://chromium-review.googlesource.com/300965
Tested-by: Olli Etuaho <oetuaho@nvidia.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Zhenyao Mo <zmo@chromium.org>
Tryjob-Request: Olli Etuaho <oetuaho@nvidia.com>
|
|
|
b3fbd867
|
2015-09-30T17:55:02
|
|
Fix setting const qualifier on indexing expression
Resubmitting now that separate fix for updating mangled names of types
has been merged. A unit test is added to make sure that the same
regression doesn't happen again.
Previously, constness of indexing expressions did not take the index
expression into account. Now constness correctly takes into account both
the base expression and the index expression.
Setting the type of expressions that index arrays is also simplified.
BUG=angleproject:1170
TEST=angle_unittests
Change-Id: Ie9b6aaee8185b945c03657b13d9513cc55cd2a9f
Reviewed-on: https://chromium-review.googlesource.com/303601
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Olli Etuaho <oetuaho@nvidia.com>
Reviewed-by: Zhenyao Mo <zmo@chromium.org>
|
|
|
b12531c9
|
2015-10-05T14:40:40
|
|
Invalidate mangled name of TType if the type is changed
Sometimes, the mangled name of a type is already generated when the type
is still being changed. For example, this could happen when the type of
A[i] is generated by copying the type of A (already mangled) and calling
clearArrayness() on the copy. Make sure that the mangled name gets
updated if the type is changed by clearing the mangled name when
something affecting it is changed.
BUG=angleproject:1170
BUG=538692
TEST=angle_unittests
Change-Id: I9dfacea653f56536e1573a6dbf60ff21da7cc5ab
Reviewed-on: https://chromium-review.googlesource.com/303846
Tested-by: Olli Etuaho <oetuaho@nvidia.com>
Reviewed-by: Zhenyao Mo <zmo@chromium.org>
|
|
7e921616
|
2015-10-01T14:00:04
|
|
Add texture2DRect -> texture conversion for OpenGL core profile
BUG=534114
Change-Id: If0e1557a2399436b1b160407ced2d265f42df5f0
Reviewed-on: https://chromium-review.googlesource.com/303761
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tryjob-Request: Jamie Madill <jmadill@chromium.org>
Tested-by: ccameron chromium <ccameron@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
|
|
|
5f579b1b
|
2015-08-14T17:44:43
|
|
Improve handling of internal function calls
Many parts of the shader translator that deal with function calls have
been written without internal function calls in mind. Fix some of these
so that they can handle internal function calls.
-Fix TLValueTrackingTraverser handling a shader where there are an
internal and non-internal function of the same name.
-Maintain internalness when shallow copying function calls in
SeparateExpressionReturningArrays and ArrayReturnValueToOutParameter
AST transformations.
-Output function internalness in intermOut.
BUG=angleproject:1116
TEST=angle_unittests
Change-Id: Ic65e2803062b807651f1b3952409face6aceb780
Reviewed-on: https://chromium-review.googlesource.com/303353
Tested-by: Olli Etuaho <oetuaho@nvidia.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Zhenyao Mo <zmo@chromium.org>
|
|
|
e623bd46
|
2015-10-02T19:14:15
|
|
Revert "Fix setting const qualifier on indexing expression"
Failing to compile shaders which are used in some layout_tests,
see http://crbug.com/538692 for logs and bad shaders.
We should diagnose if the shaders are faulty or if there's a
bug in this CL and take appropriate action, but first priority
is to get the tests running again.
BUG=538692
BUG=angleproject:1170
This reverts commit 16a79cd169420cdf0254964b1bac9dd9964a8119.
Change-Id: Iea14c58d87041bcf5ba645b7076ba0936dea6b9d
Reviewed-on: https://chromium-review.googlesource.com/303794
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>
|
|
|
16a79cd1
|
2015-09-30T17:55:02
|
|
Fix setting const qualifier on indexing expression
Previously, constness of indexing expressions did not take the index
expression into account. Now constness correctly takes into account both
the base expression and the index expression.
Setting the type of expressions that index arrays is also simplified.
BUG=angleproject:1170
TEST=angle_unittests
Change-Id: Ie2d122020cc252655ab0eea96886b9f85931b80a
Reviewed-on: https://chromium-review.googlesource.com/303350
Tested-by: Olli Etuaho <oetuaho@nvidia.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Zhenyao Mo <zmo@chromium.org>
|
|
|
d4b5054d
|
2015-09-28T12:19:26
|
|
compiler: Rewrite do-while loops as while loops
This works around a Mac driver shader compiler bug that makes many
do-while loops cause GPU-hangs when ran.
BUG=angleproject:891
Change-Id: I29828d6ea9e887ad0ed0c577f1deb41fb632a900
Reviewed-on: https://chromium-review.googlesource.com/302465
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Corentin Wallez <cwallez@chromium.org>
|
|
5f0246ca
|
2015-07-22T10:30:35
|
|
Implement gl_FragDepth for GLES SL 3.0
Makes it an error to access gl_FragDepthEXT in #version 300 es shader.
TODO:
Lacks the feature to make "#extension GL_EXT_frag_depth : require" an
error for #version 300 es.
Reland of: https://chromium-review.googlesource.com/#/c/287570
BUG=angleproject:1102
TEST=angle_unittest
Change-Id: I064d918d65f37539cb1e14f12173ca5591a4ea3f
Reviewed-on: https://chromium-review.googlesource.com/301711
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Geoff Lang <geofflang@chromium.org>
|
|
|
7258e30c
|
2015-09-22T10:40:24
|
|
TOutputGLSLBase::visitLoop clarify the function slightly
Since only for loops can be unrolled, handle each type of loop only
once. This improvement was going to be part of a do-while workaround for
Mac that will instead be done at the AST level.
BUG=angleproject:891
Change-Id: Ic8ecf41aa6d9abd7a01c2230e908c85152f52b62
Reviewed-on: https://chromium-review.googlesource.com/302332
Tryjob-Request: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Tested-by: Corentin Wallez <cwallez@chromium.org>
|
|
|
2d58f7a2
|
2015-09-28T10:44:55
|
|
RewriteElseBlocks.cpp: add missing override
Change-Id: Ic98de554cabffabd3cdd58b2a44a554cb2c3cd53
Reviewed-on: https://chromium-review.googlesource.com/302464
Tryjob-Request: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Corentin Wallez <cwallez@chromium.org>
|
|
75782629
|
2015-09-25T11:28:50
|
|
Fix possible compiler errors on linux
Change-Id: Ifc45446c749690eddc406f66f144304262f04664
Reviewed-on: https://chromium-review.googlesource.com/302478
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Corentin Wallez <cwallez@chromium.org>
|
|
|
247374cb
|
2015-09-09T15:07:24
|
|
Allow limited form of expressions in #line directives
Reuse ExpressionParser that's also used for parsing preprocessor
conditionals to parse line and file numbers in #line directives.
According to recent Khronos discussions, the intent of the spec is that
expressions in #line directives should be interpreted similarly to
expressions in conditional directives, so reusing ExpressionParser is a
natural way to implement this. This enables simple math operators
operating on integers. There are a few unclear corner cases, but this
approach is enough to support practical use cases and pass the dEQP
tests.
Valid line directives have one of the following forms:
#line line-expression
#line line-expression file-expression
ExpressionParser is first run to parse the line-expression. In ambiguous
cases the ExpressionParser consumes as much of the line as possible to
form line-expression. Then, if end-of-line hasn't been reached,
file-expression is parsed by running ExpressionParser again. As an
example of an ambiguous case:
#line 1 + 2
This could alternatively be interpreted to mean line-expression "1" and
file-expression "+ 2" where + is the unary + operator, but ANGLE now
interprets it to mean line-expression "1 + 2". Because of these ambiguous
cases, a bison grammar that would parse multiple expressions on the same
line couldn't be easily constructed, so this solution where
ExpressionParser is run twice was chosen instead.
The problematic corner cases are:
- ExpressionParser uses 64-bit integers internally for evaluating the
expression's value. It's possible to interpret the ESSL3 spec so that
32-bit integer wraparound behavior would be required also for #line
directive expressions.
- It's unclear whether the defined operator can be used in #line
expressions. In this patch it is disabled. Hoping for further
clarification from Khronos.
- It's unclear how short-circuiting should affect the parsing of
undefined identifiers in #line expressions. Now it's consistent with #if
expressions (undefined identifiers are OK if they're short-circuited).
dEQP expectations are updated for preprocessor tests, including ones
not affected specifically by this change.
BUG=angleproject:989
TEST=angle_unittests,
dEQP-GLES3.functional.shaders.preprocessor.* (4 start passing),
dEQP-GLES2.functional.shaders.preprocessor.* (4 start passing)
Change-Id: I55c5bf75857da5de855cc600d3603ee19399f328
Reviewed-on: https://chromium-review.googlesource.com/300964
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tryjob-Request: Olli Etuaho <oetuaho@nvidia.com>
Tested-by: Olli Etuaho <oetuaho@nvidia.com>
|
|
|
3187a38e
|
2015-09-09T12:00:12
|
|
Make preprocessor ExpressionParser only lex what it can parse
This cleans up ExpressionParser so that the lexer only consumes one extra
token in case the parser finishes. The parser will also finish with
YYACCEPT once it has parsed a complete expression. This will make the
preprocessor to generate a more informative unexpected token error
instead of a syntax error if there are extra tokens after #if.
This will also enable reusing ExpressionParser for parsing expressions in
line directives. The format for a line directive that specifies both line
and file numbers is as follows:
#if line-expression file_expression
ExpressionParser will need to be run twice for each line: first to parse
line-expression and then to parse file-expression. For that reason, it is
essential that ExpressionParser for line-expression stops before
consuming more than one token of file-expression.
BUG=angleproject:989
TEST=angle_unittests, dEQP-GLES3.functional.shaders.preprocessor.*
Change-Id: I0bb92f733c18891eeddbc61e7c5bebdf1003559a
Reviewed-on: https://chromium-review.googlesource.com/300962
Tested-by: Olli Etuaho <oetuaho@nvidia.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Zhenyao Mo <zmo@chromium.org>
Tryjob-Request: Olli Etuaho <oetuaho@nvidia.com>
|
|
|
261f5379
|
2015-09-18T10:34:31
|
|
Support parsing defined operator generated by macro expansion
dEQP tests enforce that the defined operator should be parsed even when
it is generated as a result of macro expansion, even though this is
undefined according to the C++ preprocessor spec.
Implement support for this by putting the parsing for the defined
operator inside MacroExpander. The operator gets processed right after
it is generated by macro expansion. Parsing the defined operator is
toggled with a boolean according to the context where MacroExpander
is used.
BUG=angleproject:989
TEST=angle_unittests,
dEQP-GLES3.functional.shaders.preprocessor.* - 2 tests start passing:
dEQP-GLES3.functional.shaders.preprocessor.conditional_inclusion.basic_2*
Change-Id: I780e63bd4558253657d898685d62339017564a06
Reviewed-on: https://chromium-review.googlesource.com/300970
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Olli Etuaho <oetuaho@nvidia.com>
|
|
|
62d31cb6
|
2015-09-11T13:25:51
|
|
Re^6-land "Move Uniform and UBO info to the gl::Program layer."
This data was previously stored entirely in the Impl level. Move
as much as possible to the GL level, using a read-only view in the
Impl level. Some information in D3D-specific, and should be stored
separately in the Impl.
This patch has a lot of refactoring that splits the D3D and GL info,
and moves as much validation as possible to the GL layer, where it
is shared between the back-ends.
Re-land with fix for dEQP unused uniforms. The fix involves storing
a local copy of all uniform data in the GL layer. This will also
let us validate sampler indexes during draw calls at the GL layer.
Re-re-land with a fix for multiply defined symbols on Clang.
Re-re-re-land with a fix for boolean uniforms and Uniform{1234}f.
Re^4-land with a fix for boolean uniform arrays and UBO uniforms.
Re^5-land with a fix for a test warning on Linux.
Re^6-land with a fix for transposed matrix uniform arrays.
BUG=angleproject:1123
TEST=end2end_tests, bots, dEQP GLES3.ubo and GLES2.uniform_api
Change-Id: Ie6fcde1c16eb05d67191b629338b88302a2563f5
Reviewed-on: https://chromium-review.googlesource.com/298971
Tryjob-Request: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>
|
|
|
fb53603c
|
2015-09-11T13:19:49
|
|
Revert "Re^5-land "Move Uniform and UBO info to the gl::Program layer.""
Failing dEQP-GLES3.functional.uniform_api.random.22 and 23:
There's a bug with arrays of tranpsosed matrix uniforms.
BUG=angleproject:1123
This reverts commit 78d356929bd0441d81e2dd8a63130cd6788e2fde.
Change-Id: If39b5908af39671dfe98965e6a1ba77fd18ea8fc
Reviewed-on: https://chromium-review.googlesource.com/299320
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>
|
|
|
78d35692
|
2015-09-11T08:32:51
|
|
Re^5-land "Move Uniform and UBO info to the gl::Program layer."
This data was previously stored entirely in the Impl level. Move
as much as possible to the GL level, using a read-only view in the
Impl level. Some information in D3D-specific, and should be stored
separately in the Impl.
This patch has a lot of refactoring that splits the D3D and GL info,
and moves as much validation as possible to the GL layer, where it
is shared between the back-ends.
Re-land with fix for dEQP unused uniforms. The fix involves storing
a local copy of all uniform data in the GL layer. This will also
let us validate sampler indexes during draw calls at the GL layer.
Re-re-land with a fix for multiply defined symbols on Clang.
Re-re-re-land with a fix for boolean uniforms and Uniform{1234}f.
Re^4-land with a fix for boolean uniform arrays and UBO uniforms.
Re^5-land with a fix for a test warning on Linux.
BUG=angleproject:1123
TEST=end2end_tests, bots, dEQP GLES3.ubo and GLES2.uniform_api
Change-Id: I03afcd446d9e74573c4d4c712ed7407b91d8001c
Reviewed-on: https://chromium-review.googlesource.com/299180
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>
|
|
|
a2eb02c9
|
2015-09-11T12:31:41
|
|
Revert "Re^4-land "Move Uniform and UBO info to the gl::Program layer.""
Test has a warning on Linux:
In file included from ../../third_party/angle/src/tests/test_utils/ANGLETest.h:13:
../../testing/gtest/include/gtest/gtest.h:1392:16: error: comparison of integers of different signs: 'const int' and 'const unsigned int' [-Werror,-Wsign-compare]
if (expected == actual) {
~~~~~~~~ ^ ~~~~~~
../../testing/gtest/include/gtest/gtest.h:1422:12: note: in instantiation of function template specialization 'testing::internal::CmpHelperEQ<int, unsigned int>' requested here
return CmpHelperEQ(expected_expression, actual_expression, expected,
^
../../third_party/angle/src/tests/gl_tests/UniformBufferTest.cpp:357:5: note: in instantiation of function template specialization 'testing::internal::EqHelper<false>::Compare<int, unsigned int>' requested here
EXPECT_EQ(GL_FLOAT, type);
^
BUG=angleproject:1123
This reverts commit 2ae1ee6161c24aab065190b1528f6195e80e34d0.
Change-Id: Ic3dabea54068377d1f2c5f722ba52966cf1a0491
Reviewed-on: https://chromium-review.googlesource.com/299170
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>
|
|
|
2ae1ee61
|
2015-09-10T10:04:24
|
|
Re^4-land "Move Uniform and UBO info to the gl::Program layer."
This data was previously stored entirely in the Impl level. Move
as much as possible to the GL level, using a read-only view in the
Impl level. Some information in D3D-specific, and should be stored
separately in the Impl.
This patch has a lot of refactoring that splits the D3D and GL info,
and moves as much validation as possible to the GL layer, where it
is shared between the back-ends.
Re-land with fix for dEQP unused uniforms. The fix involves storing
a local copy of all uniform data in the GL layer. This will also
let us validate sampler indexes during draw calls at the GL layer.
Re-re-land with a fix for multiply defined symbols on Clang.
Re-re-re-land with a fix for boolean uniforms and Uniform{1234}f.
Re^4-land with a fix for boolean uniform arrays and UBO uniforms.
BUG=angleproject:1123
TEST=end2end_tests, bots, dEQP GLES3.ubo and GLES2.uniform_api
Change-Id: I4c9f5ed31b81380507bef7981f97086d642801ae
Reviewed-on: https://chromium-review.googlesource.com/298451
Tryjob-Request: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>
|
|
|
86f8dd7c
|
2015-08-12T12:37:48
|
|
Implement a minimal EGL -> CGL backend
This succesfully renders HelloTriangle and some samples but fails on a
lot of tests. In particular it doesn't handle resizing the window and
doesn't have depth or stencil buffers.
BUG=angleproject:891
Change-Id: I16356471b470f764acb38e8dd3589e9c0129829d
Reviewed-on: https://chromium-review.googlesource.com/290770
Tested-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
|
|
54c311fc
|
2015-09-09T15:37:31
|
|
win clang build fix
BUG=
TEST=win clang bots build
Change-Id: I802be2e72a884e50cf13120d3cd9c76a8f8eb86f
Reviewed-on: https://chromium-review.googlesource.com/298253
Tested-by: Zhenyao Mo <zmo@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
|
|
|
61b8dd97
|
2015-09-09T19:04:04
|
|
Revert "Re-re-re-land "Move Uniform and UBO info to the gl::Program layer.""
Seems to be failing the dEQP-GLES2.functional.uniform_api tests relating to boolean arrays.
BUG=angleproject:1123
This reverts commit 892a6a4b17b35e89898be8b4605c1ee595d3ae13.
Change-Id: Ifc4ce3ca806ef88a02ac6f693334ac37ce098a88
Reviewed-on: https://chromium-review.googlesource.com/298520
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>
|
|
|
892a6a4b
|
2015-09-09T10:17:15
|
|
Re-re-re-land "Move Uniform and UBO info to the gl::Program layer."
This data was previously stored entirely in the Impl level. Move
as much as possible to the GL level, using a read-only view in the
Impl level. Some information in D3D-specific, and should be stored
separately in the Impl.
This patch has a lot of refactoring that splits the D3D and GL info,
and moves as much validation as possible to the GL layer, where it
is shared between the back-ends.
Re-land with fix for dEQP unused uniforms. The fix involves storing
a local copy of all uniform data in the GL layer. This will also
let us validate sampler indexes during draw calls at the GL layer.
Re-re-land with a fix for multiply defined symbols on Clang.
Re-re-re-land with a fix for boolean uniforms and Uniform{1234}f.
BUG=angleproject:1123
TEST=end2end_tests,Canary WebGL (D3D11/GL), dEQP-GLES2+3.functional.uniform_api
Change-Id: Ia40820a5ce2f34ec2d27648b1dc940a8955e9999
Reviewed-on: https://chromium-review.googlesource.com/298440
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>
|
|
|
e6432c85
|
2015-09-08T14:21:38
|
|
Fix preprocessor macro replacement list location
According to the dEQP tests, a macro replacement list generated by a
function-like macro invocation should get its location from the closing
parenthesis of the invocation. The tests check this by using __LINE__ in
a macro with a multi-line invocation. It's not quite clear from the spec
that the enforced behavior is expected as opposed to the replacement
list getting its location from the macro name, but a minor correction to
the preprocessor makes the dEQP tests pass.
Newlines in the preprocessor unit tests are generated according to the
source locations in the token list produced by the preprocessor, so the
expectations of a few tests also need to be updated.
BUG=angleproject:989
TEST=dEQP-GLES3.functional.shaders.preprocessor.predefined_macros.*
(2 start passing with this change),
angle_unittests
Change-Id: I4cc9da09bd0985310a05ebf6def680916a46308a
Reviewed-on: https://chromium-review.googlesource.com/297990
Tested-by: Olli Etuaho <oetuaho@nvidia.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Zhenyao Mo <zmo@chromium.org>
|
|
|
b654f659
|
2015-09-08T20:40:00
|
|
Revert "Re-re-land "Move Uniform and UBO info to the gl::Program layer.""
Failes the gles2_conform_test:
GLES2ConformTest.GL2Tests_glGetUniform_input_run
Possibly also WebGL failures, will investigate.
BUG=angleproject:1123
This reverts commit 10750cb936288d8dd09d49fadd592904c06c56f9.
Change-Id: I1ae59325e1831589019bc5a27ffc2091d3994a65
Reviewed-on: https://chromium-review.googlesource.com/298200
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>
|
|
|
10750cb9
|
2015-09-04T14:23:52
|
|
Re-re-land "Move Uniform and UBO info to the gl::Program layer."
This data was previously stored entirely in the Impl level. Move
as much as possible to the GL level, using a read-only view in the
Impl level. Some information in D3D-specific, and should be stored
separately in the Impl.
This patch has a lot of refactoring that splits the D3D and GL info,
and moves as much validation as possible to the GL layer, where it
is shared between the back-ends.
Re-land with fix for dEQP unused uniforms. The fix involves storing
a local copy of all uniform data in the GL layer. This will also
let us validate sampler indexes during draw calls at the GL layer.
Re-re-land with a fix for multiply defined symbols on Clang.
BUG=angleproject:1123
TEST=end2end_tests,Canary WebGL (D3D11/GL), dEQP-GLES2+3.functional.uniform_api
Change-Id: I7b55e4964fbe64e0106c37928c894146937a2476
Reviewed-on: https://chromium-review.googlesource.com/298110
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>
|
|
|
2119dd05
|
2015-09-08T19:30:37
|
|
Revert "Re-land "Move Uniform and UBO info to the gl::Program layer.""
Problems with multiply defined symbols in the Linux builder.
BUG=angleproject:1123
This reverts commit ff1abe63c3d8af28e9755c8d0c95b2aeca27a0d9.
Change-Id: Ifcb2ad885d0841c128ffd69afb6594e37eba17fc
Reviewed-on: https://chromium-review.googlesource.com/298100
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>
|
|
|
ff1abe63
|
2015-09-04T14:23:52
|
|
Re-land "Move Uniform and UBO info to the gl::Program layer."
This data was previously stored entirely in the Impl level. Move
as much as possible to the GL level, using a read-only view in the
Impl level. Some information in D3D-specific, and should be stored
separately in the Impl.
This patch has a lot of refactoring that splits the D3D and GL info,
and moves as much validation as possible to the GL layer, where it
is shared between the back-ends.
Re-land with fix for dEQP unused uniforms. The fix involves storing
a local copy of all uniform data in the GL layer. This will also
let us validate sampler indexes during draw calls at the GL layer.
BUG=angleproject:1123
TEST=end2end_tests,Canary WebGL (D3D11/GL), dEQP-GLES2+3.functional.uniform_api
Change-Id: I9074c057c2c67d4d3221ec11c5556a26e07c7ceb
Reviewed-on: https://chromium-review.googlesource.com/297070
Tryjob-Request: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>
|
|
|
809ec546
|
2015-08-26T14:30:57
|
|
Don't evaluate short-circuited preprocessor expressions
Resubmit with clang build issue fixed. The result of a short-circuited
operation is now either 0 or 1.
ESSL 3.00 spec section 3.4 mentions that the second operand in a logical
&& or || preprocessor operation is evaluated only if the first operand
doesn't short-circuit the expression. The non-evaluated part of a
preprocessor expression may also have undefined identifiers.
Make the expression parser follow the spec by ignoring errors that are
generated inside short-circuited expressions. This includes undefined
identifiers and divide by zero.
BUG=angleproject:347
TEST=dEQP-GLES3.functional.shaders.preprocessor.undefined_identifiers.*
angle_unittests
Change-Id: I4163f96ec46d40ac859ffb39d91b89490041e44d
Reviewed-on: https://chromium-review.googlesource.com/297252
Tested-by: Olli Etuaho <oetuaho@nvidia.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
|
9a1b49f7
|
2015-09-08T14:32:26
|
|
Revert "Don't evaluate short-circuited preprocessor expressions"
Build break on Clang:
FAILED: /b/build/goma/gomacc ../../third_party/llvm-build/Release+Asserts/bin/clang++ -MMD -MF obj/third_party/angle/src/compiler/preprocessor/preprocessor.ExpressionParser.o.d -DV8_DEPRECATION_WARNINGS -DCLD_VERSION=2 -D__ASSERT_MACROS_DEFINE_VERSIONS_WITHOUT_UNDERSCORE=0 -DCHROMIUM_BUILD -DCR_CLANG_REVISION=245965-1 -DUSE_LIBJPEG_TURBO=1 -DENABLE_ONE_CLICK_SIGNIN -DENABLE_PRE_SYNC_BACKUP -DENABLE_WEBRTC=1 -DENABLE_MEDIA_ROUTER=1 -DUSE_PROPRIETARY_CODECS -DENABLE_PEPPER_CDMS -DENABLE_CONFIGURATION_POLICY -DENABLE_NOTIFICATIONS -DENABLE_HIDPI=1 -DSYSTEM_NATIVELY_SIGNALS_MEMORY_PRESSURE -DDONT_EMBED_BUILD_METADATA -DDCHECK_ALWAYS_ON=1 -DFIELDTRIAL_TESTING_ENABLED -DENABLE_TASK_MANAGER=1 -DENABLE_EXTENSIONS=1 -DENABLE_PDF=1 -DENABLE_PLUGIN_INSTALLATION=1 -DENABLE_PLUGINS=1 -DENABLE_SESSION_SERVICE=1 -DENABLE_THEMES=1 -DENABLE_AUTOFILL_DIALOG=1 -DENABLE_BACKGROUND=1 -DENABLE_GOOGLE_NOW=1 -DENABLE_PRINTING=1 -DENABLE_BASIC_PRINTING=1 -DENABLE_PRINT_PREVIEW=1 -DENABLE_SPELLCHECK=1 -DUSE_BROWSER_SPELLCHECKER=1 -DENABLE_CAPTIVE_PORTAL_DETECTION=1 -DENABLE_APP_LIST=1 -DENABLE_SETTINGS_APP=1 -DENABLE_SUPERVISED_USERS=1 -DENABLE_SERVICE_DISCOVERY=1 -DENABLE_WIFI_BOOTSTRAPPING=1 -DV8_USE_EXTERNAL_STARTUP_DATA -DFULL_SAFE_BROWSING -DSAFE_BROWSING_CSD -DSAFE_BROWSING_DB_LOCAL -DSAFE_BROWSING_SERVICE -DUSE_LIBPCI=1 -DUSE_OPENSSL=1 -DNDEBUG -DNVALGRIND -DDYNAMIC_ANNOTATIONS_ENABLED=0 -Igen -isysroot /Applications/Xcode5.1.1.app/Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs/MacOSX10.10.sdk -O2 -gdwarf-2 -fvisibility=hidden -Werror -Wnewline-eof -mmacosx-version-min=10.6 -arch x86_64 -Wendif-labels -Wno-unused-parameter -Wno-missing-field-initializers -Wno-selector-type-mismatch -Wpartial-availability -Wheader-hygiene -Wno-char-subscripts -Wno-unneeded-internal-declaration -Wno-covered-switch-default -Wstring-conversion -Wno-c++11-narrowing -Wno-deprecated-register -Wno-inconsistent-missing-override -Wno-shift-negative-value -Wno-unused-function -Wno-unused-variable -std=c++11 -fno-rtti -fno-exceptions -fvisibility-inlines-hidden -fno-threadsafe-statics -Xclang -load -Xclang /b/build/slave/GPU_Mac_Builder/build/src/third_party/llvm-build/Release+Asserts/lib/libFindBadConstructs.dylib -Xclang -add-plugin -Xclang find-bad-constructs -Xclang -plugin-arg-find-bad-constructs -Xclang check-templates -fcolor-diagnostics -fno-strict-aliasing -c ../../third_party/angle/src/compiler/preprocessor/ExpressionParser.cpp -o obj/third_party/angle/src/compiler/preprocessor/preprocessor.ExpressionParser.o
../../third_party/angle/src/compiler/preprocessor/ExpressionParser.cpp:1372:35: error: use of logical '||' with constant operand [-Werror,-Wconstant-logical-operand]
(yyval) = (yyvsp[-3]) || 0;
^ ~
../../third_party/angle/src/compiler/preprocessor/ExpressionParser.cpp:1372:35: note: use '|' for a bitwise operation
(yyval) = (yyvsp[-3]) || 0;
^~
|
../../third_party/angle/src/compiler/preprocessor/ExpressionParser.cpp:1406:35: error: use of logical '&&' with constant operand [-Werror,-Wconstant-logical-operand]
(yyval) = (yyvsp[-3]) && 0;
^ ~
../../third_party/angle/src/compiler/preprocessor/ExpressionParser.cpp:1406:35: note: use '&' for a bitwise operation
(yyval) = (yyvsp[-3]) && 0;
^~
&
../../third_party/angle/src/compiler/preprocessor/ExpressionParser.cpp:1406:35: note: remove constant to silence this warning
(yyval) = (yyvsp[-3]) && 0;
~^~~~
BUG=angleproject:347
This reverts commit 6ffe613518482b966b913013c51221ce06ca7c33.
Change-Id: I6d81666cca573f320bfb1164a6c794b6f75f7463
Reviewed-on: https://chromium-review.googlesource.com/298020
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>
|
|
|
6ffe6135
|
2015-08-26T14:30:57
|
|
Don't evaluate short-circuited preprocessor expressions
ESSL 3.00 spec section 3.4 mentions that the second operand in a logical
&& or || preprocessor operation is evaluated only if the first operand
doesn't short-circuit the expression. The non-evaluated part of a
preprocessor expression may also have undefined identifiers.
Make the expression parser follow the spec by ignoring errors that are
generated inside short-circuited expressions. This includes undefined
identifiers and divide by zero.
BUG=angleproject:347
TEST=dEQP-GLES3.functional.shaders.preprocessor.undefined_identifiers.*
angle_unittests
Change-Id: Ieed02a71298af838f784a5d1197d4f4a9ba0e3c8
Reviewed-on: https://chromium-review.googlesource.com/295033
Reviewed-by: Zhenyao Mo <zmo@chromium.org>
Tested-by: Olli Etuaho <oetuaho@nvidia.com>
|
|
|
3766a40d
|
2015-06-18T15:47:46
|
|
Remove dynamic indexing of matrices and vectors in HLSL
Re-landing after fixing D3D9 specific issues.
HLSL doesn't support dynamic indexing of matrices and vectors, so replace
that with helper functions that unroll dynamic indexing into switch/case
and static indexing.
Both the indexed vector/matrix expression and the index may have side
effects, and these will be evaluated correctly. If necessary, index
expressions that have side effects will be written to a temporary
variable that will replace the index.
Besides dEQP tests, this change is tested by a WebGL 2 conformance test.
In the case that a dynamic index is out-of-range, the base ESSL 3.00 spec
allows undefined behavior. KHR_robust_buffer_access_behavior adds the
requirement that program termination should not occur and that
out-of-range reads must return either a value from the active program's
memory or zero, and out-of-range writes should only affect the active
program's memory or do nothing. This patch clamps out-of-range indices so
that either the first or last item of the matrix/vector is accessed.
The code is not transformed in case the it fits within the limited subset
of ESSL 1.00 given in Appendix A of the spec. If the code isn't within
the restricted subset, even ESSL 1.00 shaders may require this
workaround.
BUG=angleproject:1116
TEST=dEQP-GLES3.functional.shaders.indexing.* (all pass after change)
WebGL 2 conformance tests (glsl3/vector-dynamic-indexing.html)
Change-Id: I66a5e5a8d7f4267da0045f1cc2ba6b0dc7eb3f5d
Reviewed-on: https://chromium-review.googlesource.com/296671
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tryjob-Request: Jamie Madill <jmadill@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Zhenyao Mo <zmo@chromium.org>
|
|
|
bdcc54a4
|
2015-09-02T13:09:48
|
|
Support writing interface blocks in OutputGLSL.
BUG=angleproject:882
BUG=angleproject:1149
Change-Id: Iea0b1d0bc586ec9517a06793386c91890b7a5115
Reviewed-on: https://chromium-review.googlesource.com/297086
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Tested-by: Geoff Lang <geofflang@chromium.org>
|
|
|
d8edb518
|
2015-09-02T11:52:56
|
|
Allow OutputGLSL to write non-square matrices.
Fixes assertion failures when ShaderGL fails to compile the generated shaders.
BUG=angleproject:882
BUG=angleproject:1149
Change-Id: Ib7c23a54bdb81852864d792f29c89644084859a3
Reviewed-on: https://chromium-review.googlesource.com/297085
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Geoff Lang <geofflang@chromium.org>
|
|
|
6e36042f
|
2015-09-02T15:54:36
|
|
Implement translation of EOpIndexDirectInterfaceBlock in OutputGLSL.
BUG=angleproject:882
BUG=angleproject:1149
Change-Id: I5e0808dd3576339aa87aade3a349586d01b28f74
Reviewed-on: https://chromium-review.googlesource.com/297087
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Geoff Lang <geofflang@chromium.org>
|
|
|
a0a9e12d
|
2015-09-02T15:54:30
|
|
translator: Add sh::OutputVariable type.
This replaces the dual-use of sh::Attribute, which can be a bit
confusing to people expecting a literal output variable.
Currently not used in Chromium, so should be safe to land.
BUG=angleproject:1146
Change-Id: I436f2bc9dc4ddc3709369cb2baa344c6b13a21a2
Reviewed-on: https://chromium-review.googlesource.com/296683
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Reviewed-by: Zhenyao Mo <zmo@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>
|
|
|
48705cad
|
2015-09-02T20:40:51
|
|
Revert "Move Uniform and UBO info to the gl::Program layer."
Seems to be failing dEQP-GLES2.functional.uniform_api.value.unused_uniforms.* on Linux (possibly Win as well)
BUG=angleproject:1123
This reverts commit 54f882c9167b1aff5b3700187a2048bd886e8b17.
Change-Id: I7dbbf40aae8dd9ebd35895df0dd338a3b6b9cc96
Reviewed-on: https://chromium-review.googlesource.com/297051
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>
|
|
|
54f882c9
|
2015-09-02T14:33:48
|
|
Move Uniform and UBO info to the gl::Program layer.
This data was previously stored entirely in the Impl level. Move
as much as possible to the GL level, using a read-only view in the
Impl level. Some information in D3D-specific, and should be stored
separately in the Impl.
This patch has a lot of refactoring that splits the D3D and GL info,
and moves as much validation as possible to the GL layer, where it
is shared between the back-ends.
BUG=angleproject:1123
TEST=end2end_tests,Canary WebGL (D3D11/GL), dEQP-GLES2+3.functional.uniform_api
Change-Id: I8b7e1380a61590a46ccee646a47ffe9bda287f6b
Reviewed-on: https://chromium-review.googlesource.com/295124
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Tryjob-Request: Jamie Madill <jmadill@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>
|
|
|
477b243b
|
2015-08-31T10:41:16
|
|
Change the FLATTEN heuristic to "ifs with a loop with a gradient"
This heuristic makes more sense than the previous "ifs with a
discontinuous loop" as the reason we need to flatten is that we need
gradients to be in branchless code.
Change the UnrollFlatten test accordingly.
Tested with:
- the WebGL CTS
- dev.miaumiau.cat/rayTracer "Skull Demo"
- THe turbulenz engine GPU particle demo
- Lots of ShaderToy Samples (inc. Volcanic, Metropolis and Hierarchical
Voronoi)
- Google Maps Earth mode
- Lots of Chrome experiments
- madebyevan.com/webgl-water
BUG=524297
Change-Id: Iaa727036fffcfde3952716a1ef33b6ee0546b69d
Reviewed-on: https://chromium-review.googlesource.com/296442
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Corentin Wallez <cwallez@chromium.org>
|
|
|
80ecac9e
|
2015-08-25T15:37:48
|
|
Remove unused mOptimize/mDebug flags from TIntermAggregate
These flags were written but they were never read.
TEST=compile
BUG=angleproject:1116
Change-Id: I41e3e89f13861ebda4828c76c753ca17c74c4358
Reviewed-on: https://chromium-review.googlesource.com/294931
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Zhenyao Mo <zmo@chromium.org>
Tested-by: Olli Etuaho <oetuaho@nvidia.com>
|
|
|
384f1fdd
|
2015-08-26T16:27:07
|
|
Fix a warning on Linux
BUG=
Change-Id: I5c4f5df14321644d7ac42af6059400ae43d0fe4d
Reviewed-on: https://chromium-review.googlesource.com/295721
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Tryjob-Request: Geoff Lang <geofflang@chromium.org>
Tested-by: Corentin Wallez <cwallez@chromium.org>
|
|
|
3c192a78
|
2015-08-26T20:32:53
|
|
Revert "Remove dynamic indexing of matrices and vectors in HLSL"
Seems to be failing a WebGL/ES2 CTS test in D3D9:
conformance/ogles/GL/mat3/mat3_001_to_006
BUG=angleproject:1116
BUG=525188
This reverts commit 83f3411da456faac8570892e3dd7d76edf4095e5.
Change-Id: Ic186f51240dbdd96ccab3f5470329cdc9727c618
Reviewed-on: https://chromium-review.googlesource.com/295730
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>
|
|
|
83f3411d
|
2015-06-18T15:47:46
|
|
Remove dynamic indexing of matrices and vectors in HLSL
HLSL doesn't support dynamic indexing of matrices and vectors, so replace
that with helper functions that unroll dynamic indexing into switch/case
and static indexing.
Both the indexed vector/matrix expression and the index may have side
effects, and these will be evaluated correctly. If necessary, index
expressions that have side effects will be written to a temporary
variable that will replace the index.
Besides dEQP tests, this change is tested by a WebGL 2 conformance test.
In the case that a dynamic index is out-of-range, the base ESSL 3.00 spec
allows undefined behavior. KHR_robust_buffer_access_behavior adds the
requirement that program termination should not occur and that
out-of-range reads must return either a value from the active program's
memory or zero, and out-of-range writes should only affect the active
program's memory or do nothing. This patch clamps out-of-range indices so
that either the first or last item of the matrix/vector is accessed.
The code is not transformed in case the it fits within the limited subset
of ESSL 1.00 given in Appendix A of the spec. If the code isn't within
the restricted subset, even ESSL 1.00 shaders may require this
workaround.
BUG=angleproject:1116
TEST=dEQP-GLES3.functional.shaders.indexing.* (all pass after change)
WebGL 2 conformance tests (glsl3/vector-dynamic-indexing.html)
Change-Id: I024722ef4ca1e14d5ad47fdc540397e18858bed6
Reviewed-on: https://chromium-review.googlesource.com/290515
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Olli Etuaho <oetuaho@nvidia.com>
|
|
|
d7a25243
|
2015-08-18T13:49:45
|
|
Add deep copying support for typed AST nodes
Resubmit with inconsistent override warnings fixed.
Removing dynamic indexing of vectors and matrices will require copying
the indexed nodes in case they are written. Any type of l-value node that
doesn't have side effects may need to be copied. Add a copying function
for all typed node classes so that this copying can be performed.
Private copy constructors are used to implement the deepCopy function in
order to make maintenance easier. With copy constructors, each subclass
only needs to take care of copying its own members, and not the base
class members, which reduces the possibility of errors. Copy constructors
are disabled for all node classes that don't support deep copying by
inheriting TIntermNode from angle::NonCopyable.
Assignment operator is disabled for all node classes through inheriting
angle::NonCopyable. This applies also to classes that now get the private
copy constructor.
Explicit copy constructor and assignment operator declarations are added
to some classes which show up in node member variables to make code
clearer.
BUG=angleproject:1116
TEST=angle_unittests
Change-Id: I7964976f5dac7dfd745b8c6612ca06fb01d271c4
Reviewed-on: https://chromium-review.googlesource.com/295080
Tested-by: Olli Etuaho <oetuaho@nvidia.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Zhenyao Mo <zmo@chromium.org>
|
|
|
4dfe8094
|
2015-08-21T17:44:35
|
|
Always collect variables when generating HLSL
HLSL output needs uniform information generated by the collectVariables()
step to be able to write uniform registers.
Tested manually by compiling a shader with a uniform with
shader_translator.
BUG=angleproject:1132
Change-Id: I91d19b5fa789b7b33cf76a654ffbbd17d279db01
Reviewed-on: https://chromium-review.googlesource.com/294962
Tested-by: Olli Etuaho <oetuaho@nvidia.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Zhenyao Mo <zmo@chromium.org>
|
|
|
e5a1f271
|
2015-08-21T02:58:25
|
|
Use override in all the places where it is possible
This will avoid -Winconsistent-overrides in the future. Done using the
-Wsuggest-override warning of GCC 5.1
BUG=
Change-Id: I707a649dc368f5dd1e139fd144370abcac0b6263
Reviewed-on: https://chromium-review.googlesource.com/294920
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Corentin Wallez <cwallez@chromium.org>
|
|
|
b5d61dac
|
2015-08-20T13:43:02
|
|
Revert "Add deep copying support for typed AST nodes"
This reverts commit d0d59aa4cbce2da92deaa0f4021dfed8885e8d01 and commit
760379532b02ae039790f1e8f1c89e03481aab4b
To be fixed will be the newline at the end of IntermNode_tests.cpp and
the -Winconsistent-missing-override
BUG=
Change-Id: I57693bdff01f9e3f03b2fbb5dc53fa21e68c2789
Reviewed-on: https://chromium-review.googlesource.com/294820
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Tested-by: Corentin Wallez <cwallez@chromium.org>
|
|
|
a7b6db7f
|
2015-08-19T14:26:30
|
|
Only apply Appendix A limitations to ESSL 1.00 shaders
ESSL 1.00 specifies a set of minimum functionality, and ANGLE
automatically checks that WebGL shaders stay within this minimum
functionality. However, this should only apply to ESSL 1.00. ESSL 3.00
shaders compiled for WebGL 2.0 should not be subject to these
restrictions, since there is no similar spec for minimum functionality
for ESSL 3.00.
In case a non-WebGL based shader spec is used, the restrictions can be
toggled from outside by specifying the SH_VALIDATE_LOOP_INDEXING flag,
same as before this patch.
BUG=angleproject:1116
TEST=WebGL 2 conformance tests
Change-Id: Idaec0fb4c7c85cd72020d0b23112fddb1b020571
Reviewed-on: https://chromium-review.googlesource.com/293933
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Olli Etuaho <oetuaho@nvidia.com>
|
|
|
d0d59aa4
|
2015-08-18T13:49:45
|
|
Add deep copying support for typed AST nodes
Removing dynamic indexing of vectors and matrices will require copying
the indexed nodes in case they are written. Any type of l-value node that
doesn't have side effects may need to be copied. Add a copying function
for all typed node classes so that this copying can be performed.
Private copy constructors are used to implement the deepCopy function in
order to make maintenance easier. With copy constructors, each subclass
only needs to take care of copying its own members, and not the base
class members, which reduces the possibility of errors. Copy constructors
are disabled for all node classes that don't support deep copying by
inheriting TIntermNode from angle::NonCopyable.
Assignment operator is disabled for all node classes through inheriting
angle::NonCopyable. This applies also to classes that now get the private
copy constructor.
Explicit copy constructor and assignment operator declarations are added
to some classes which show up in node member variables to make code
clearer.
BUG=angleproject:1116
TEST=angle_unittests
Change-Id: Ia757b69397837f8309f0e7511c0cd24ca2c7a721
Reviewed-on: https://chromium-review.googlesource.com/293931
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Olli Etuaho <oetuaho@nvidia.com>
|
|
|
26e355b8
|
2015-08-14T14:16:19
|
|
Add full support for line continuation in the preprocessor
Re-landing earlier change with constant signedness fixed (was causing
build issues on Linux).
Line continuation in ESSL 3.00 needs to be processed before tokenization,
since tokens can span the line continuation. On the other hand, ANGLE's
tokenizer keeps track of line numbers, and whenever a line continuation
appears the line number still needs to be incremented by one, just like
on a regular newline.
That's why line continuation is now implemented as follows: when the
shader strings are concatenated in Input, they are also checked for line
continuation. Whenever line continuation is encountered, the string
is cut before that point. When the tokenizer asks for more input, the
string starting from the character after line continuation is passed
to it, and the line number is incremented from Input. This way the
tokenizer can parse tokens that span multiple lines - it never sees the
line continuation - but still keeps track of the line number correctly.
Relevant spec is in ESSL 3.00 section 3.2 "Source strings".
Support for line continuation also applies to ESSL 1.00. ESSL 3.00
spec section 1.5 says that line continuation support is mandated when
an ESSL 1.00 shader is used with the OpenGL ES 3.0 API, and is optional
when ESSL 1.00 is used with the OpenGL ES 2.0 API.
TEST=dEQP-GLES3.functional.shaders.preprocessor.line_continuation.*
(all pass),
angle_unittests
BUG=angleproject:1125
Change-Id: Ic086aacac53cd75bf93c0fda782416501d2f842b
Reviewed-on: https://chromium-review.googlesource.com/294200
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Olli Etuaho <oetuaho@nvidia.com>
|
|
|
3b040eb8
|
2015-08-17T16:51:33
|
|
Revert "Add full support for line continuation in the preprocessor"
Warning in the Linux/Mac builders:
In file included from ../../third_party/angle/src/tests/preprocessor_tests/input_test.cpp:7:
In file included from ../../third_party/angle/src/tests/preprocessor_tests/PreprocessorTest.h:7:
../../testing/gtest/include/gtest/gtest.h:1392:16: error: comparison of integers of different signs: 'const int' and 'const unsigned long' [-Werror,-Wsign-compare]
if (expected == actual) {
~~~~~~~~ ^ ~~~~~~
../../testing/gtest/include/gtest/gtest.h:1422:12: note: in instantiation of function template specialization 'testing::internal::CmpHelperEQ<int, unsigned long>' requested here
return CmpHelperEQ(expected_expression, actual_expression, expected,
^
../../third_party/angle/src/tests/preprocessor_tests/input_test.cpp:171:5: note: in instantiation of function template specialization 'testing::internal::EqHelper<false>::Compare<int, unsigned long>' requested here
EXPECT_EQ(3, input.read(buf, maxSize, &lineNo));
^
BUG=angleproject:1125
This reverts commit c1157d1963170c7411eb6c32e2b2fbce02c5a170.
Change-Id: Ic6fa286d190b006cccc5154d86e21ecc03175763
Reviewed-on: https://chromium-review.googlesource.com/294080
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>
|
|
|
c1157d19
|
2015-08-14T14:16:19
|
|
Add full support for line continuation in the preprocessor
Line continuation in ESSL 3.00 needs to be processed before tokenization,
since tokens can span the line continuation. On the other hand, ANGLE's
tokenizer keeps track of line numbers, and whenever a line continuation
appears the line number still needs to be incremented by one, just like
on a regular newline.
That's why line continuation is now implemented as follows: when the
shader strings are concatenated in Input, they are also checked for line
continuation. Whenever line continuation is encountered, the string
is cut before that point. When the tokenizer asks for more input, the
string starting from the character after line continuation is passed
to it, and the line number is incremented from Input. This way the
tokenizer can parse tokens that span multiple lines - it never sees the
line continuation - but still keeps track of the line number correctly.
Relevant spec is in ESSL 3.00 section 3.2 "Source strings".
Support for line continuation also applies to ESSL 1.00. ESSL 3.00
spec section 1.5 says that line continuation support is mandated when
an ESSL 1.00 shader is used with the OpenGL ES 3.0 API, and is optional
when ESSL 1.00 is used with the OpenGL ES 2.0 API.
TEST=dEQP-GLES3.functional.shaders.preprocessor.line_continuation.*
(all pass),
angle_unittests
BUG=angleproject:1125
Change-Id: I1c1de49602e7cd755d6072c3c0aa5524cd0313b4
Reviewed-on: https://chromium-review.googlesource.com/293721
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Olli Etuaho <oetuaho@nvidia.com>
|
|
|
59f9a641
|
2015-08-06T20:38:26
|
|
Remove EOpInternalFunctionCall
It's cleaner to mark internal functions by using the TName class,
similarly to TIntermSymbol.
TEST=angle_unittests
BUG=angleproject:1116
Change-Id: I12a03a3dea42b3fc571fa25a1b11d0161f24de72
Reviewed-on: https://chromium-review.googlesource.com/291621
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Olli Etuaho <oetuaho@nvidia.com>
|
|
|
f5cfc8df
|
2015-08-06T16:36:39
|
|
Track whether a name is internal to ANGLE in a separate class
The AST contains identifiers in a few different places: besides symbols,
there are also function names, which show up in function signatures
and function calls. Any of these can be coming either from the original
shader or from inside ANGLE. A class that encapsulates a string and its
internalness will be useful for implementing a unified way of handling
all names in shader translation. Start implementing this by splitting
the functionality out of TSymbol.
TEST=angle_unittests
BUG=angleproject:1116
Change-Id: I0a1b5936dcccd0d5fc1c0c13c712102fbfff2a79
Reviewed-on: https://chromium-review.googlesource.com/291280
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Olli Etuaho <oetuaho@nvidia.com>
|
|
ec17d54a
|
2015-08-13T12:19:28
|
|
Fix angle miscompiling shaders on x86_64 Android.
BUG=487341
Change-Id: I792e0c9419566facb0bec0ad93f3646294e5a2a7
Reviewed-on: https://chromium-review.googlesource.com/293500
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Nico Weber <thakis@chromium.org>
Reviewed-by: Nico Weber <thakis@chromium.org>
|
|
|
217fe6ec
|
2015-08-05T13:25:08
|
|
Detect when built-in modf requires an l-value in AST traversal
This fixes an omission that out parameter tracking had inherited from
EmulatePrecision. Accurate tracking of when values are written is
required for converting dynamic indexing of vectors and matrices to
function calls.
A new test covering this is added to angle_unittests.
TEST=angle_unittests
BUG=angleproject:1116
Change-Id: I05c5fd60355117d0053b84110748ae221375a790
Reviewed-on: https://chromium-review.googlesource.com/290562
Reviewed-by: Zhenyao Mo <zmo@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Olli Etuaho <oetuaho@nvidia.com>
|
|
|
3fc9337f
|
2015-08-11T14:50:59
|
|
Refactor l-value tracking to a separate traverser parent class
This makes TIntermTraverser implementation easier to understand and
removes the overhead of maintaining the user-defined GLSL function table
from the traversers that don't need it.
Some logic is duplicated between TIntermTraverser and its new subclass
TLValueTrackingTraverser, but duplication is hard to eliminate completely
since there are some differences scattered throughout the code.
BUG=angleproject:1116
TEST=angle_unittests
Change-Id: Iab4a0c1d4320ecfafaf18ea3a45824d756890774
Reviewed-on: https://chromium-review.googlesource.com/292721
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Olli Etuaho <oetuaho@nvidia.com>
|
|
|
6cb4c7f0
|
2015-08-13T11:27:17
|
|
Set __VERSION__ macro when the #version directive is parsed
__VERSION__ macro needs to be set to 300 when #version 300 es is on the
first line of the shader, since section 3.4 of ESSL 3.00.4 spec mentions
that the value of __VERSION__ should match the shading language being
parsed. The value from parsing the version directive replaces the default
value 100.
BUG=angleproject:524
TEST=dEQP-GLES3.functional.shaders.preprocessor.predefined_macros.*
(4 tests start passing, 2 still fail)
dEQP-GLES3.functional.shaders.preprocessor.conditionals.*
(2 tests start passing)
dEQP-GLES2.functional.shaders.preprocessor.predefined_macros.*
(no regression)
Change-Id: I15bfdeb73d1e343d131ded56b1fd52ca5ef32408
Reviewed-on: https://chromium-review.googlesource.com/293440
Reviewed-by: Zhenyao Mo <zmo@chromium.org>
Tested-by: Olli Etuaho <oetuaho@nvidia.com>
|
|
4d61f7ed
|
2015-08-12T10:56:50
|
|
Reland Fixed compiler warning C4267 'conversion from 'size_t' to 'type', possible loss of data'
Additional warnings found with more testing and added C4267 warning disable only for angle_libpng
BUG=angleproject:1120
Change-Id: Ic403dcff5a8018056fa51a8c408e64207f3362eb
Reviewed-on: https://chromium-review.googlesource.com/293028
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Tested-by: Geoff Lang <geofflang@chromium.org>
|
|
|
8b76e9fb
|
2015-08-12T15:22:58
|
|
Fix warnings on Linux
BUG=
Change-Id: I24ed2a2ee85a152ef79fc9cb2c3b067bbbaedf05
Reviewed-on: https://chromium-review.googlesource.com/293310
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Corentin Wallez <cwallez@chromium.org>
|
|
|
b195643c
|
2015-08-12T17:35:20
|
|
Revert "Fixed compiler warning C4267 'conversion from 'size_t' to 'type', possible loss of data'"
Seems to have quite a few warnings in 64-bit on my machine.
BUG=angleproject:1120
This reverts commit c5cf9bc47d0ee028adbbf9e9f94ca567eec601dc.
Change-Id: I86768b900aeba52e7a2242d9ae8949f93f1a5ba9
Reviewed-on: https://chromium-review.googlesource.com/293280
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>
|
|
|
56193ce3
|
2015-08-12T15:55:09
|
|
Clean up parseMatrixFields
Applying field selection directly on matrices is not mentioned in ESSL
1.00 or 3.00 specs. Remove erroneous code that generated odd error
messages when a shader tried to apply certain kinds of field selection
on a matrix.
BUG=angleproject:1118
TEST=angle_unittests, dEQP-GLES3.functional.shaders.swizzles.*
Change-Id: I7bbf5d0cbaee3f21d20b830d904c0feef445dd78
Reviewed-on: https://chromium-review.googlesource.com/293190
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Olli Etuaho <oetuaho@nvidia.com>
|
|
|
391befef
|
2015-08-12T16:30:38
|
|
Revert "Add pragma errors for malformed pragmas."
Since this commit was made, dEQP tests were fixed to check that
unrecognized pragma tokens only generate warnings, not errors.
This applies to both ESSL1.00 and ESSL3.00, which specify this
behavior in section 3.4 Preprocessor.
BUG=angleproject:989
TEST=dEQP-GLES2.functional.shaders.preprocessor.pragmas.*
This reverts commit d3c29f57aaeb451b149bbb9fd17b3f1f99101c52.
Change-Id: Ie4e0ec061fa3164d6f3872ac0016a063056ed110
Reviewed-on: https://chromium-review.googlesource.com/293181
Tested-by: Olli Etuaho <oetuaho@nvidia.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
|
|
|
c5cf9bc4
|
2015-08-06T10:46:48
|
|
Fixed compiler warning C4267 'conversion from 'size_t' to 'type', possible loss of data'
BUG=angleproject:1120
Change-Id: I01ef10bea7f487c2b394d030c76628f38d2ea645
Reviewed-on: https://chromium-review.googlesource.com/292780
Tested-by: Cooper Partin <coopp@microsoft.com>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
|
27446bda
|
2015-08-10T14:59:53
|
|
Move traverse functions under TIntermTraverser
This enables implementing specialized variants of TIntermTraverser.
Intermediate subclasses of TIntermTraverser are expected to maintain
contextual information in their traverse* functions, which is then used
by the visit* functions in the actual traversers.
This approach is better than adding a flag to TIntermTraverser, since
it will make the code cleaner and easier to understand.
Traverse() functions in TIntermNode subclasses are still kept around to
redirect calls to TIntermTraverser traversal functions. This is less
hacky than choosing the function with switch/case in TIntermTraverser
would be, and if new TIntermNode subclasses get added, it's not as likely
that adding the traversal support would be forgotten.
TEST=angle_unittests
BUG=angleproject:1116
Change-Id: Ie6889be1d7e955518f13cd3390dce17871ba49b5
Reviewed-on: https://chromium-review.googlesource.com/292720
Tested-by: Olli Etuaho <oetuaho@nvidia.com>
Reviewed-by: Zhenyao Mo <zmo@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
|
|
|
4f488494
|
2015-08-07T16:05:25
|
|
Fixed compiler warning C4457 'declaration of 'node' hides function parameter'.
BUG=angleproject:1119
Change-Id: Ie3de5a776b5860b1ca502cee9e2b19c41cd3bfb2
Reviewed-on: https://chromium-review.googlesource.com/292051
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Cooper Partin <coopp@microsoft.com>
|
|
|
149e6e69
|
2015-08-07T16:18:18
|
|
Fixed compiler warning C4458 'declaration of variable hides class member'.
BUG=angleproject:1119
Change-Id: Ibc7cfdea72abe402cbfa1c10e0ada7576fa1cfa2
Reviewed-on: https://chromium-review.googlesource.com/292052
Tested-by: Cooper Partin <coopp@microsoft.com>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
|
|
|
8afe1e1b
|
2015-08-05T18:00:01
|
|
Track that indices of l-values are not required to be l-values
In an expression like a[ind]++, a[ind] is required to be an l-value but
ind is not. Reset the l-value required flags before traversing the index
of an indexing operation, so that this is accurately tracked. After the
index has been traversed, the previous state of the l-value required
flags is restored.
New tests are added to angle_unittests cover this functionality.
TEST=angle_unittests
BUG=angleproject:1116
Change-Id: I8929ec01e85e672c83ef7d385e455b7df8682f4b
Reviewed-on: https://chromium-review.googlesource.com/290561
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Zhenyao Mo <zmo@chromium.org>
Tested-by: Olli Etuaho <oetuaho@nvidia.com>
|
|
|
a26ad58d
|
2015-08-04T13:51:47
|
|
Track where l-values are required in AST traversal
This functionality is refactored out of EmulatePrecision to be a common
feature of TIntermTraverser. This is done since tracking where l-values
are required will be useful for other traversers. For example, it will
be needed for converting dynamic indexing of matrices and vectors to
function calls. This change adds some overhead to all tree traversers,
but the overhead is expected to be small for typical shaders which don't
contain too many user-defined functions.
BUG=angleproject:1116
TEST=angle_unittests
Change-Id: I54d34c2b5093ef028f2b24d854c11c0195dc1dbb
Reviewed-on: https://chromium-review.googlesource.com/290514
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Olli Etuaho <oetuaho@nvidia.com>
Reviewed-by: Zhenyao Mo <zmo@chromium.org>
|
|
|
b18609b9
|
2015-07-16T14:13:11
|
|
Implement ES 2,3 parts of EXT_blend_func_extended for shader translation
Exposes gl_SecondaryFragColor, glSecondaryFragData[] and
gl_MaxDualSourceDrawBuffers to GLES SL 1.0.
Relaxes rules for undefined output locations for GLES SL 3.0
and exposes gl_MaxDualSourceDrawBuffers.
If the output GL context is GL ES 2.0 or 3.0:
The emulation layer is expected to turn on EXT_blend_func_extended
if the output GL context supports it.
If the output GL context is GL:
The emulation layer is expected to turn on EXT_blend_func_extended
if the output GL context supports ARB_blend_func_extended or if GL
context is 3.2 or later.
If the source shader spec is GLES SL 2.0: The emulation layer is
expected to inspect the shader compilation output variables upon
linking. If output target is GL SL, the emulation layer should bind
color location 0, index 1 to "angle_SecondaryFragColor" if variable
"gl_SecondaryFragColorEXT" is used. Alternatively, emulation layer
should bind "angle_SecondaryFragData" to locations 0,1,2,3,..., all
color index 1, if "gl_SecondaryFragData" array is used.
(The caller can not bind the locations or specify output variables.)
If the source shader spec is GLES SL 3.0:
The emulation layer is expected to do location auto-resolve of the
the output variables that have undefined output locations that have
not been bound by the caller.
(The caller can not use gl_ built-ins, so nothing to do with those.)
BUG=angleproject:1085
TEST=angle_unittest
Change-Id: I5cafe205b0c29478b0dcd24aa89a7b0000f5d046
Reviewed-on: https://chromium-review.googlesource.com/287580
Reviewed-by: Zhenyao Mo <zmo@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Kimmo Kinnunen <kkinnunen@nvidia.com>
|
|
|
43613b03
|
2015-08-04T11:02:21
|
|
Make sure that the AST root is always a sequence node
This enables inserting helper functions as an AST transformation.
BUG=angleproject:1116
TEST=angle_unittests
Change-Id: I169d4d3a726d0e389cb3444fe9dfb4c6c5d80155
Reviewed-on: https://chromium-review.googlesource.com/290513
Tested-by: Olli Etuaho <oetuaho@nvidia.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Zhenyao Mo <zmo@chromium.org>
|
|
|
cc36b983
|
2015-07-10T14:14:18
|
|
Implement ESSL 3.00 shader input/output variable type rules
ESSL 3.00 allows a wider variety of types of input/output variables
than ESSL 1.00, but there are still specific restrictions on structs,
matrices and arrays.
Some of the checks need to be implemented twice: once for array syntax
where the brackets are after the type, and another time for array syntax
where the brackets are after the variable name.
This requires fixes to constant folding unit tests which were previously
incorrectly using matrix outputs in fragment shaders.
New unit tests are added for several of the rules introduced, but some
cases are also covered by dEQP.
TEST=angle_unittests, dEQP-GLES.functional.shaders.linkage.varying.rules.*
BUG=angleproject:1061
Change-Id: I655b054cfe56d376db775b96a2bb41b3ac5740b0
Reviewed-on: https://chromium-review.googlesource.com/285482
Tested-by: Olli Etuaho <oetuaho@nvidia.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Zhenyao Mo <zmo@chromium.org>
|
|
|
13e7c7e6
|
2015-07-30T14:17:29
|
|
Revert "Names of built-in functions cannot be redeclared as functions"
Causing failures in the WebGL1 test: conformance/glsl/misc/shader-with-non-reserved-words.html
This reverts commit b5f88853ea80ea112368bb15999b363db0e4c648.
Change-Id: I2105c8040057665abda00435e8c0ff8a83af3645
Reviewed-on: https://chromium-review.googlesource.com/289192
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Tested-by: Geoff Lang <geofflang@chromium.org>
|
|
b5f88853
|
2015-07-23T17:37:39
|
|
Names of built-in functions cannot be redeclared as functions
With ESSL 3.00, names of built-in functions cannot be redeclared as
functions and therefore overloading or redefining builtin functions
is an error.
This is fixed by inserting unmangled built-ins into the symbol table
and then checking if the new function declaration matches any of the
built-in in symbol table.
BUG=angleproject:1066
TEST=angle_unittests(new: ParseESSLFunctionsTest), dEQP Tests
dEQP-GLES3.functional.shaders.functions.invalid.overload_builtin_function*
(2 tests started passing with this change)
Change-Id: I0e027e588664e604f29c130028178315c3e21631
Reviewed-on: https://chromium-review.googlesource.com/287801
Reviewed-by: Zhenyao Mo <zmo@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>
|
|
|
b98c3a82
|
2015-07-23T14:26:04
|
|
Reformat style in ParseContext.cpp.
Using git cl format.
BUG=None
Change-Id: Ia0ce03ae8115f45c3bcc4b022acd7e3e88518062
Reviewed-on: https://chromium-review.googlesource.com/287833
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Zhenyao Mo <zmo@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>
|
|
|
1a4b1b3c
|
2015-07-23T18:27:13
|
|
Fix use-after-delete in ParseContext.cpp.
This bug was exposed after a refactoring from the translator.
BUG=None
TEST=angle_unittests,angle_end2end_tests
Change-Id: I13fddcbe84f87826068a557f139f6e35c674571e
Reviewed-on: https://chromium-review.googlesource.com/287832
Reviewed-by: Zhenyao Mo <zmo@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>
|
|
|
e2bfe2cd
|
2015-07-23T21:25:45
|
|
Revert "Implement gl_FragDepth for GLES SL 3.0"
assertion failures in WebGL2 CTS.
This reverts commit 544809610a131fe7040f2212789c62e212bbaf24.
Change-Id: I2e0c7045c5b6ef9031a6e6c5916504fe68f51077
Reviewed-on: https://chromium-review.googlesource.com/287910
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Tested-by: Geoff Lang <geofflang@chromium.org>
|
|
|
54480961
|
2015-07-22T10:30:35
|
|
Implement gl_FragDepth for GLES SL 3.0
Makes it an error to access gl_FragDepthEXT in #version 300 es shader.
TODO:
Lacks the feature to make "#extension GL_EXT_frag_depth : require" an
error for #version 300 es.
BUG=angleproject:1102
TEST=angle_unittest
Change-Id: Ic313a0e1ed6369550d53885290f300fab1aaf304
Reviewed-on: https://chromium-review.googlesource.com/287570
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Zhenyao Mo <zmo@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>
|
|
|
0932df61
|
2015-07-21T14:35:11
|
|
Expose GL built-in output variables in ShGetOutputVariables
Expose GL built-in output variables in ShGetOutputVariables.
Currently gl_FragColor, gl_FragData and gl_FragDepthEXT are exposed.
The output variable names in the returned array are the input shader names,
not the output shader names.
This is needed in future features in which the emulation layer (command
buffer/libANGLE) needs to know which output variables caller used.
Example of such a feature is EXT_blend_func_extended, where
gl_SecondaryFragColorEXT and gl_SecondaryFragDataEXT cause the need
to bind the emulated output variables to their respective color
indices.
BUG=angleproject:1085
TEST=angle_unittests
Change-Id: I7ca3e0fe6bdd3e3c66113518aa771cbb013fc014
Reviewed-on: https://chromium-review.googlesource.com/287230
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Zhenyao Mo <zmo@chromium.org>
Tested-by: Kimmo Kinnunen <kkinnunen@nvidia.com>
|
|
|
01c796ac
|
2015-07-21T18:53:58
|
|
Revert "Emulate the pack/unpack functions for unorms."
Causing MSAN failures on Linux.
This reverts commit 1915652ee82109d7756dc8349562c3c11ea39b77.
Change-Id: Ib23bec16eab22288930be0b41186e54cd8d1f921
Reviewed-on: https://chromium-review.googlesource.com/287127
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Tested-by: Geoff Lang <geofflang@chromium.org>
|
|
|
c4c7442d
|
2015-07-20T13:09:26
|
|
Fix missing uint cases for GLSL and ESSL output.
These were causing assertions in several WebGL2 conformance tests.
BUG=angleproject:882
BUG=angleproject:1084
BUG=483282
Change-Id: I7651c359a528c95469dcbfb5da2715ae4616b49b
Reviewed-on: https://chromium-review.googlesource.com/286592
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Zhenyao Mo <zmo@chromium.org>
Tested-by: Geoff Lang <geofflang@chromium.org>
|
|
|
9a4c8c46
|
2015-07-21T11:03:51
|
|
Initialize all members in the default TType constructor.
This is an attempt to fix use-of-uninitialized-value errors.
Example failure: https://build.chromium.org/p/tryserver.chromium.linux/builders/linux_chromium_msan_rel_ng/builds/68/steps/cc_unittests%20%28with%20patch%29/logs/EnlargedTextureWithAlphaThresholdFilter.GL
BUG=angleproject:1044
Change-Id: I5906c67a55da553e2e5fd15320a5dd4186644e87
Reviewed-on: https://chromium-review.googlesource.com/287191
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Geoff Lang <geofflang@chromium.org>
|
|
|
8518e7f6
|
2015-07-16T14:54:08
|
|
Make all fragment shader out variables require location layout qualifier
Make all fragment shader out variables require location layout
qualifier. Previously, the last variable did not need a location layout
qualifier if the previous variables had those.
TEST=angle_unittests
BUG=angleproject:1070
Change-Id: Ifb66ee52b811409f5278eaad330d6cd9b8ea059f
Reviewed-on: https://chromium-review.googlesource.com/287100
Tested-by: Kimmo Kinnunen <kkinnunen@nvidia.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>
|