David Neto [Wed, 7 Dec 2016 15:33:27 +0000 (10:33 -0500)]
An OpVariable initializer can be a module-scope variable
Fixes https://github.com/KhronosGroup/SPIRV-Tools/issues/482
Ehsan Nasiri [Mon, 5 Dec 2016 20:11:51 +0000 (15:11 -0500)]
Fixing bug in OpAccessChain validation code.
The validation code for OpAccessChain was missing OpTypeRuntimeArray as
a possible type that can be indexed into.
This was caught by running the validator on VKCTS.
Also adding unit tests for it.
Ehsan Nasiri [Mon, 5 Dec 2016 19:00:18 +0000 (14:00 -0500)]
Add a test that exercises wrong usage of result id
Result <id> resulting from an instruction within a function may not be
used outside that function.
Ehsan Nasiri [Wed, 30 Nov 2016 18:29:12 +0000 (13:29 -0500)]
Adding validation code for OpAccessChain.
* Result Type must be an OpTypePointer. Its Type operand must be the
type reached by walking the Base’s type hierarchy down to the last
provided index in Indexes, and its Storage Class operand must be the
same as the Storage Class of Base.
* Base must be a pointer, pointing to the base of a composite object.
* Indexes walk the type hierarchy to the desired depth, potentially down
to scalar granularity. The first index in Indexes will select the
top-level member/element/component/element of the base composite. All
composite constituents use zero-based numbering, as described by their
OpType... instruction. The second index will apply similarly to that
result, and so on. Once any non-composite type is reached, there must
be no remaining (unused) indexes. Each of the Indexes must:
- be a scalar integer type,
- be an OpConstant when indexing into a structure.
* Check for the case where no indexes are passed to OpAccessChain.
Minor improvements based on code review.
Ehsan Nasiri [Thu, 1 Dec 2016 21:18:15 +0000 (16:18 -0500)]
No need to install cmake for new osx images.
It seems that the osx image has been updated and already includes ctest.
As a result, all jobs fail with this error:
Error: cmake-3.6.2 already installed
Ehsan Nasiri [Tue, 29 Nov 2016 20:50:34 +0000 (15:50 -0500)]
Validate the number of global and local variables.
According to the Universal Limits section of the SPIR-V Spec (2.17), the
number of global variables may not exceed 65,535 and the number of local
variables may not exceed 524,287.
Also added unit tests for each one.
Ehsan Nasiri [Mon, 28 Nov 2016 18:04:32 +0000 (13:04 -0500)]
Validate the number of arguments in OpTypeFunction
According to the SPIR-V spec (section 2.17: Universal Limits), the
OpTypeFunction instruction may not take more than 255 arguments for the
function. Also added unit tests for it.
David Neto [Wed, 30 Nov 2016 20:48:39 +0000 (15:48 -0500)]
Say how the validator checks universal limits
Indicate it's future work to allow parameterization of those
limits.
Ehsan Nasiri [Fri, 25 Nov 2016 14:26:26 +0000 (09:26 -0500)]
Adding validation code for OpSwitch limits
The number of (literal, label) pairs passed to OpSwitch may not exceed
16,383. Added code to validate this and added unit tests for it.
Also fixed a typo in another validor error message.
Ehsan Nasiri [Thu, 24 Nov 2016 20:37:22 +0000 (15:37 -0500)]
Validation for number of Struct members.
The number of members in a struct may not exceed 16,383.
Also Adding unit tests for Struct size limit check.
Ehsan Nasiri [Thu, 24 Nov 2016 20:37:22 +0000 (15:37 -0500)]
Checks that result IDs are within the ID bound specified in the SPIR-V header
This is described in Section 2.17 of the SPIR-V Spec.
* Updated existing unit test 'SemanticsIdIsAnIdNotALiteral' to pass by
manipulating the ID bound in its binary header.
* Fixed boundary check in the code.
* Added unit test to check the case that the largest ID is equal to the
ID bound.
Ehsan Nasiri [Tue, 22 Nov 2016 23:06:55 +0000 (18:06 -0500)]
Validation for OpSampledImage instruction.
This change implements the validation for usages of OpSampledImage
instruction as described in the Data Rules section of the Universal
Validation Rules of the SPIR-V Spec.
Ehsan Nasiri [Wed, 16 Nov 2016 16:04:13 +0000 (11:04 -0500)]
Adding validation for OpSpecConstantComposite.
SpecConstantComposite may specialize to a vector, matrix, array, or
struct. In each case, the number of components and type of components
that are being specialized to must match the expected result type.
Removed use of macros in these tests.
Now using the spvValidateBase class. Using CompileSuccessfully(), and
ValidateInstructions() to compile to binary and run the validator. Also
using getDiagnosticString() to check the proper error message string.
All the heavy lifting is done in ValidateBase class.
Ehsan Nasiri [Thu, 10 Nov 2016 20:12:26 +0000 (15:12 -0500)]
Adding validation code for OpTypeStruct.
According to the Data Rules section of 2.16.1. Universal Validation
Rules of the SPIR-V Spec:
Forward reference operands in an OpTypeStruct
* must be later declared with OpTypePointer
* the type pointed to must be an OpTypeStruct
* had an earlier OpTypeForwardPointer forward reference to the same <id>
David Neto [Wed, 16 Nov 2016 15:52:31 +0000 (10:52 -0500)]
Skip building example executable when asked
David Neto [Thu, 13 Oct 2016 20:22:04 +0000 (16:22 -0400)]
Make friendly number-based names for OpConstant
For example:
%int_42 = OpConstant %int 42
%int_n42 = OpConstant %int -42
%float_3_14 = OpConstant %float 3.14
David Neto [Thu, 13 Oct 2016 19:17:11 +0000 (15:17 -0400)]
Extract EmitNumericLiteral from disassembler
Test with disassembly of numeric literals.
Ehsan Nasiri [Thu, 10 Nov 2016 20:12:26 +0000 (15:12 -0500)]
Adding validation code for more data rules.
These rules are under "Data Rules" in 2.16.1 (Universal Validation
Rules) part of the SPIR-V 1.1 Specification document:
* Scalar floating-point types can be parameterized only as 32 bit, plus
any additional sizes enabled by capabilities.
* Scalar integer types can be parameterized only as 32 bit, plus any
additional sizes enabled by capabilities.
* Vector types can only be parameterized with numerical types or the
OpTypeBool type.
* Matrix types can only be parameterized with floating-point types.
* Matrix types can only be parameterized as having only 2, 3, or 4
columns.
* Specialization constants (see Specialization) are limited to integers,
Booleans, floating-point numbers, and vectors of these.
Lei Zhang [Tue, 8 Nov 2016 18:26:09 +0000 (10:26 -0800)]
Rename validation source files to comply with Google style guide.
Lei Zhang [Tue, 8 Nov 2016 18:11:07 +0000 (10:11 -0800)]
Rename validation tests to comply with Google style guide.
Ehsan Nasiri [Fri, 4 Nov 2016 22:31:21 +0000 (18:31 -0400)]
Adding validation for vector data rule.
Number of components in a vector can be 2 or 3 or 4. If Vector16
capability is used, 8 and 16 components are also allowed.
Also added unit tests for vector data rule.
Lei Zhang [Thu, 22 Sep 2016 15:05:30 +0000 (11:05 -0400)]
Renaming test files to comply with Google style guide.
Also posfixing test files with `_test' to make it more clear.
David Neto [Fri, 28 Oct 2016 02:47:20 +0000 (22:47 -0400)]
Remove source/instruction.cpp
Avoids linker problems on Windows and OSX with duplicate object file
names. Avoids collision with source/val/instruction.cpp
Fixes https://github.com/KhronosGroup/SPIRV-Tools/issues/451
Jamie Madill [Thu, 27 Oct 2016 20:09:06 +0000 (16:09 -0400)]
enum_set: Fix bool performance warning.
Implicit casts from int to bool cause a warning in visual studio.
Christopher Gautier [Fri, 14 Oct 2016 10:45:27 +0000 (12:45 +0200)]
Do not gitignore source/opt/build_module.*
David Neto [Wed, 12 Oct 2016 14:47:28 +0000 (10:47 -0400)]
Test SPV_KHR_shader_draw_parameters support
In assembler, disassembler, parser.
Corentin Wallez [Tue, 4 Oct 2016 21:47:19 +0000 (17:47 -0400)]
Fix compilation errors when building inside Chromium
Example of an error:
spirv-tools/source/validate_cfg.cpp:516:45: error: chosen constructor is
explicit in copy-initialization:
_.current_function().RegisterBlockEnd({}, opcode);
Lei Zhang [Thu, 29 Sep 2016 16:00:35 +0000 (12:00 -0400)]
Add begin() and end() for TypeManager.
David Neto [Thu, 29 Sep 2016 20:43:47 +0000 (16:43 -0400)]
Remove unnecessary #include
This will make it easier to port these tests to different contexts.
Lei Zhang [Wed, 28 Sep 2016 14:50:43 +0000 (10:50 -0400)]
Add examples to demo C++ library APIs.
Lei Zhang [Fri, 23 Sep 2016 19:13:24 +0000 (15:13 -0400)]
Avoid appending EOL in StringifyMessage().
EOL should be appended when outputing to streams by the tool.
Lei Zhang [Thu, 22 Sep 2016 15:28:48 +0000 (11:28 -0400)]
Fix operand class test coverage table.
* Fix a missing column.
* Convert line endings.
This will make the table beautiful and searchable on Github.
Lei Zhang [Wed, 21 Sep 2016 21:16:31 +0000 (17:16 -0400)]
Clean up header dependency.
David Neto [Thu, 22 Sep 2016 21:09:34 +0000 (17:09 -0400)]
Handle parse failure for setting spec constants
Umar Arshad [Thu, 22 Sep 2016 14:23:08 +0000 (10:23 -0400)]
Fix forward reference issues with forword pointer
* Allows OpTypeForwardPointer to reference IDs not yet declared in
the module
* Allows OpTypeStruct to reference IDs not yet declared in
the module
Possible Issue: OpTypeStruct should only allow forward references
if the ID is a pointer that is referenced by a forward pointer. Need
Type support in Validator which is currently a work in progress.
qining [Thu, 22 Sep 2016 15:50:12 +0000 (11:50 -0400)]
Command line option for set default value for spec constant
Format:
```
--set-spec-constant-default-value "<spec id A>:<default value A> <spec id
B>:<default value B> ..."
```
Example:
shader: `test.vert`
```
layout(constant_id = 100) const int myint = 10;
layout(constant_id = 101) const int myuint = 100;
layout(constant_id = 200) const float myfloat = 1.25;
layout(constant_id = 201) const double mydouble = 2.34;
void main() {}
```
command line:
```
spirv-opt --set-spec-const-default-value "100:12 101:200 200:1.2323
201:1.2345" test.vert -o output.spv
```
output:
```
...
OpDecorate %7 SpecId 100
OpDecorate %8 SpecId 101
OpDecorate %10 SpecId 200
OpDecorate %12 SpecId 201
%void = OpTypeVoid
%3 = OpTypeFunction %void
%int = OpTypeInt 32 1
%7 = OpSpecConstant %int 12
%8 = OpSpecConstant %int 200
%float = OpTypeFloat 32
%10 = OpSpecConstant %float 1.23232
%double = OpTypeFloat 64
%12 = OpSpecConstant %double 2.34232
...
```
Lei Zhang [Wed, 21 Sep 2016 20:45:53 +0000 (16:45 -0400)]
Use nullptr as the default message consumer to ignore all messages.
There is no difference between the previous IgnoreMessage() function
and a null std::function, from functionality's perspective.
The user can set nullptr as the MessageConsumer, so need to guard
against nullptr before calling the consumer anyway. It's better
we use it internally so that it may expose problems by us instead
of the user.
Lei Zhang [Wed, 21 Sep 2016 14:53:15 +0000 (10:53 -0400)]
Change some asserts to normal errors in IrLoader.
Lei Zhang [Wed, 21 Sep 2016 14:52:31 +0000 (10:52 -0400)]
Add two more utility functions for sending errors to consumer.
Lei Zhang [Wed, 14 Sep 2016 16:09:52 +0000 (12:09 -0400)]
Change parameter name from file to source in Log() and Logf().
Lei Zhang [Tue, 20 Sep 2016 22:03:37 +0000 (18:03 -0400)]
Add more Assemble(), Disassemble() and Validate() overloads.
Lei Zhang [Fri, 16 Sep 2016 20:12:04 +0000 (16:12 -0400)]
Publish the C++ interface.
Lei Zhang [Fri, 16 Sep 2016 19:56:30 +0000 (15:56 -0400)]
Rename SpvTools to SpirvTools.
Lei Zhang [Wed, 21 Sep 2016 16:44:37 +0000 (12:44 -0400)]
Use C style names for MessageLevel.
Lei Zhang [Wed, 21 Sep 2016 15:30:41 +0000 (11:30 -0400)]
Change to use enum instead of enum class for MessageLevel.
This allows us to create C interface with MessageLevel later.
Lei Zhang [Mon, 12 Sep 2016 18:11:46 +0000 (14:11 -0400)]
Add a C++ interface to the optimization functionalities.
Lei Zhang [Tue, 20 Sep 2016 20:48:00 +0000 (16:48 -0400)]
Allow changing MessageConsumer in Pass & PassManager.
Default-constructed Pass/PassManager will have a MessageConsumer
which ignores all messages. SetMessageConsumer() should be called
to supply a meaningful MessageConsumer.
Lei Zhang [Tue, 20 Sep 2016 20:44:24 +0000 (16:44 -0400)]
Change BuildModule() to accept pointer-size pair for binary.
Lei Zhang [Mon, 12 Sep 2016 16:39:44 +0000 (12:39 -0400)]
Change interface of Pass::Process() to return possible failures.
David Neto [Mon, 19 Sep 2016 14:13:04 +0000 (10:13 -0400)]
Provide friendly names for shader ballot BuiltIns
Umar Arshad [Thu, 15 Sep 2016 21:32:44 +0000 (17:32 -0400)]
Update validate interface to accept binary pointers
Lei Zhang [Fri, 16 Sep 2016 19:43:41 +0000 (15:43 -0400)]
Set the default source name as "input".
David Neto [Fri, 16 Sep 2016 20:40:00 +0000 (16:40 -0400)]
Start v2016.6-dev
David Neto [Fri, 16 Sep 2016 20:39:40 +0000 (16:39 -0400)]
Finalize v2016.5
David Neto [Fri, 16 Sep 2016 18:40:02 +0000 (14:40 -0400)]
Add basic support for SPV_KHR_shader_ballot
Requires use of SPIRV-Headers that has support
for SPV_KHR_shader_ballot.
Adds assembler, disassembler, binary parser support.
Adds general support for allowing an operand to be
only enabled by a set of extensions.
TODO: Validator support for extension checking.
Lei Zhang [Fri, 9 Sep 2016 14:46:23 +0000 (10:46 -0400)]
Prepare the C++ interface for publication.
* Use PIMPL idiom in the C++ interface.
* Clean up interface for assembling and disassembling.
* Add validation into C++ interface.
* Add more tests for the C++ interface.
Lei Zhang [Fri, 9 Sep 2016 18:45:18 +0000 (14:45 -0400)]
Move functions for building modules outside of the C++ interface.
Lei Zhang [Fri, 16 Sep 2016 16:10:47 +0000 (12:10 -0400)]
Let SPIRV_ASSERT() exit the program if the assertion fails.
Lei Zhang [Sat, 20 Aug 2016 03:17:44 +0000 (23:17 -0400)]
Add MessageConsumer to PassManager, Pass, and analysis interfaces.
Also convert some uses of assert() in optimization code to use
SPIRV_ASSERT(), SPIRV_UNIMPLEMENTED(), or SPIRV_UNREACHABLE()
accordingly.
Lei Zhang [Thu, 8 Sep 2016 15:43:48 +0000 (11:43 -0400)]
Add utility macros for logging messages.
Add the following macros for logging purpose:
* SPIRV_ASSERT
* SPIRV_DEBUG
* SPIRV_UNIMPLEMENTED
* SPIRV_UNREACHABLE
The last two is always turned on, while the first two can only
be turned on in debug build.
Lei Zhang [Fri, 2 Sep 2016 22:06:18 +0000 (18:06 -0400)]
Add a callback mechanism for communicating messages to callers.
Every time an event happens in the library that the user should be
aware of, the callback will be invoked.
The existing diagnostic mechanism is hijacked internally by a
callback that creates an diagnostic object each time an event
happens.
David Neto [Tue, 13 Sep 2016 16:38:42 +0000 (12:38 -0400)]
Make friendly names for built-in variables.
Handles only OpDecorate
Does not handle:
- decorations on struct members
- decoration via OpGroupDecorate
David Neto [Wed, 14 Sep 2016 19:22:47 +0000 (15:22 -0400)]
Update CHANGES to say #414 is fixed
David Neto [Wed, 14 Sep 2016 15:57:20 +0000 (11:57 -0400)]
Fix validator: OpUndef can be member of a constant composite
This was enabled in SPIR-V 1.0 Rev 7
Fixes: https://github.com/KhronosGroup/SPIRV-Tools/issues/414
David Neto [Wed, 14 Sep 2016 16:48:38 +0000 (12:48 -0400)]
Clarify: Ok to pass null diagnostic to spvBinaryParse
It's a valid use case: I might mostly trust my binaries and I don't
care to to collect detailed diagnostic feedback.
David Neto [Wed, 14 Sep 2016 15:04:19 +0000 (11:04 -0400)]
Fix validator SSA check: Phi can use its own value sometimes
Defer removal of a Phi's result id from the undefined-forward-reference
set until after you've scanned the arguments. The reordering is only
significant for Phi.
Fixes https://github.com/KhronosGroup/SPIRV-Tools/issues/415
qining [Fri, 9 Sep 2016 15:47:42 +0000 (11:47 -0400)]
Build "spec id->default val str" mapping from string
Add function `ParseDefaultValuesString()` to build the spec id->default
value string mapping required by `SetSpecConstantDefaultValuePass`.
qining [Tue, 13 Sep 2016 15:57:02 +0000 (11:57 -0400)]
Guard against nullptr in number parsing utilities
ParseNumber(): Returns false if the given string is a nullptr.
ParseAndEncodeXXXX(): Returns kInvalidText and populate error message:
"The given text is a nullptr", if the givne string is a nullptr.
qining [Tue, 30 Aug 2016 18:57:56 +0000 (14:57 -0400)]
Set default values for spec const decorated by SpecId
The pass instance is constructed with a map from spec id (uint32_t) to
default values in string format. The default value strings will be
parsed to numbers according to the target spec constant type.
If the Spec Id decoration is found to be applied on multiple different
target ids, that decoration instruction (OpDecorate or OpGroupDecorate)
will be skipped. But other decoration instrucitons may still be
processed.
David Neto [Mon, 12 Sep 2016 20:48:05 +0000 (16:48 -0400)]
Update CMake condition for Emacs load/save helper
Make SPIRV_TOOLS_INSTALL_EMACS_HELPERS a CMake option. Otherwise
I get a variable-defined-but-not-used error.
steve-lunarg [Fri, 2 Sep 2016 20:04:02 +0000 (14:04 -0600)]
add spirv-dis/spirv-as support for emacs when loading .spv binaries
qining [Thu, 8 Sep 2016 22:00:12 +0000 (18:00 -0400)]
Add IsSigned() to opt::Analysis::Integer.
qining [Thu, 1 Sep 2016 18:27:04 +0000 (14:27 -0400)]
Pull out the number parsing logic
Pull out the number parsing logic from
AssemblyContext::binaryEncodeNumericLiteral() to utilities.
The new utility function: `ParseAndEncodeNumber()` now accepts:
* number text to parse
* number type
* a emit function, which is a function which will be called with each
parsed uint32 word.
* a pointer to std::string to be overwritten with error messages.
(pass nullptr if expect no error message)
and returns:
* an enum result type to indicate the status
Type/Structs moved to utility:
* template<typename T> class ClampToZeroIfUnsignedType
New type:
* enum EncodeNumberStatus: success or error code
* NumberType: hold the number type information for the number to be parsed.
* several helper functions are also added for NumberType.
Functions moved to utility:
* Helpers:
* template<typename T> checkRangeAndIfHexThenSignExtend() -> CheckRangeAndIfHex....()
* Interfaces:
* template<typename T> parseNumber() -> ParseNumber()
* binaryEncodeIntegerLiteral() -> ParseAndEncodeIntegerNumber()
* binaryEncodeFloatingPointLiteral() -> ParseAndEncodeFloatingPointNumber()
* binaryEncodeNumericLiteral() -> ParseAndEncodeNumber()
Tests added/moved to test/ParseNumber.cpp, including tests for:
* ParseNumber(): This is moved from TextToBinary.cpp to ParseNumber.cpp
* ParseAndEncodeIntegerNumber(): New added
* ParseAndEncodeFloatingPointNumber(): New added
* ParseAndEncodeNumber(): New added
Note that the error messages are kept almost the same as before, but
they may be inappropriate for an utility function. Those will be fixed
in another CL.
Lei Zhang [Thu, 8 Sep 2016 21:26:53 +0000 (17:26 -0400)]
Merge diagnostic tests and single them out in a separate binary.
Lei Zhang [Wed, 7 Sep 2016 21:43:15 +0000 (17:43 -0400)]
Create separate binaries for validation tests.
Lei Zhang [Wed, 7 Sep 2016 21:26:18 +0000 (17:26 -0400)]
Prefix test binaries with `test_`, rename UnitSPIRV, update README.
Lei Zhang [Mon, 5 Sep 2016 20:17:33 +0000 (16:17 -0400)]
Travis CI: mark build as finished once a job failed.
Lei Zhang [Mon, 5 Sep 2016 21:11:44 +0000 (17:11 -0400)]
Kill the spvCheck macro which is just alias to if-statement.
Lei Zhang [Fri, 2 Sep 2016 14:15:47 +0000 (10:15 -0400)]
Update README about the optimization library and tool.
David Neto [Fri, 2 Sep 2016 14:29:14 +0000 (10:29 -0400)]
Say contributors have to click through the CLA
David Neto [Fri, 2 Sep 2016 13:29:52 +0000 (09:29 -0400)]
Add test for presence of copyrights and license terms
David Neto [Thu, 1 Sep 2016 19:38:53 +0000 (15:38 -0400)]
Start v2016.5-dev
David Neto [Thu, 1 Sep 2016 19:33:59 +0000 (15:33 -0400)]
Relicense SPIRV-Tools under Apache 2.0
Fixes https://github.com/KhronosGroup/SPIRV-Tools/issues/383
Finalize v2016.4
David Neto [Thu, 1 Sep 2016 18:49:14 +0000 (14:49 -0400)]
Add check_copyright to check or update copyright.
It will also replace an MIT or Khronos free use license with
the Apache license notice.
David Neto [Thu, 1 Sep 2016 17:04:00 +0000 (13:04 -0400)]
Fix Google copyright lines
Lei Zhang [Fri, 2 Sep 2016 13:51:44 +0000 (09:51 -0400)]
Kill the spvCheckReturn macro.
Lei Zhang [Fri, 2 Sep 2016 13:49:56 +0000 (09:49 -0400)]
Kill diagnostic_helper & the wrapping DIAGNOSTIC macro.
Lei Zhang [Wed, 31 Aug 2016 23:37:13 +0000 (19:37 -0400)]
Tests for preserving line debug insts even killing its affecting targets.
qining [Wed, 17 Aug 2016 21:19:48 +0000 (17:19 -0400)]
Unify constants pass
De-duplicate constants and unifies the uses of constants for a SPIR-V
module. If two constants are defined exactly the same, only one of them
will be kept and all the uses of the removed constant will be redirected
to the kept one.
This pass handles normal constants (defined with
OpConstant{|True|False|Composite}), some spec constants (those defined
with OpSpecConstant{Op|Composite}) and null constants (defined with
OpConstantNull).
There are several cases not handled by this pass:
1) If there are decorations for the result id of a constant defining
instruction, that instruction will not be processed. This means the
instruction won't be used to replace other instructions and other
instructions won't be used to replace it either.
2) This pass does not unify null constants (defined with
OpConstantNull instruction) with their equivalent zero-valued normal
constants (defined with OpConstant{|False|Composite} with zero as the
operand values or component values).
David Neto [Wed, 31 Aug 2016 18:35:58 +0000 (14:35 -0400)]
Refactor CapabilitySet into templated class EnumSet
qining [Wed, 31 Aug 2016 16:44:49 +0000 (12:44 -0400)]
Add forwarding so that passes' ctor can have args.
Also removed the default argument value of `skip_nop` for function
`SinglePassRunAndCheck()` and `SinglePassRunAndDisassemble()`. This is
required to support variadic arguments.
David Neto [Mon, 29 Aug 2016 18:49:00 +0000 (14:49 -0400)]
Handle capabilities larger than value 63
Use libspirv::CapabilitySet instead of a 64-bit mask.
Remove dead function spvOpcodeRequiresCapability and its tests.
The JSON grammar parser is simplified since it just writes the
list of capabilities as a braced list, and takes advantage of
the CapabilitySet intializer-list constructor.
David Neto [Sat, 27 Aug 2016 18:49:53 +0000 (14:49 -0400)]
Add libspirv::CapabilitySet
It's optimized for the common case, where capabilities have value
at most 63.
qining [Mon, 29 Aug 2016 22:05:24 +0000 (18:05 -0400)]
Get annotation instructions of a given id.
qining [Mon, 29 Aug 2016 20:09:40 +0000 (16:09 -0400)]
Add AppendNames() interface for AssemblyBuilder
Lei Zhang [Mon, 29 Aug 2016 19:46:09 +0000 (15:46 -0400)]
Create a separate header and source file for each pass.
qining [Thu, 4 Aug 2016 17:24:08 +0000 (13:24 -0400)]
Fold spec constants to normal constants (values fixed)
For the spec constants defined by OpSpecConstantOp and
OpSpecContantComposite, if all of their operands are constants with
determined values (normal constants whose values are fixed), calculate
the correct values of the spec constants and re-define them as normal
constants.
In short, this pass replaces all the spec constants defined by
OpSpecContantOp and OpSpecConstantComposite with normal constants when
possible. So far not all valid operations of OpSpecConstantOp are
supported, we have several constriction here:
1) Only 32-bit integer and boolean (both scalar and vector) are
supported for any arithmetic operations. Integers in other width (like
64-bit) are not supported.
2) OpSConvert, OpFConvert, OpQuantizeToF16, and all the
operations under Kernel capability, are not supported.
3) OpCompositeInsert is not supported.
Note that this pass does not unify normal constants. This means it is
possible to have new generatd constants defining the same values.
David Neto [Thu, 25 Aug 2016 22:36:03 +0000 (18:36 -0400)]
Pass manager recomputes Id bound automatically.
Fixes https://github.com/KhronosGroup/SPIRV-Tools/issues/371
in the sense that the id bound is correct after all the passes
have been run. But it might be inconsistent between passes.
Lei Zhang [Fri, 26 Aug 2016 20:52:16 +0000 (16:52 -0400)]
Create const overloads for begin() and end() in Instruction.
David Neto [Fri, 26 Aug 2016 20:25:40 +0000 (16:25 -0400)]
Fix typo in spirv-opt help