Commit Graph

224 Commits

Author SHA1 Message Date
Lei Zhang
16f3ddfbb8 Use std::string instead of a huge array for storing literal strings. 2015-11-12 09:43:06 -05:00
Lei Zhang
2a0b773b27 Add missing copyright. 2015-11-12 09:43:03 -05:00
Lei Zhang
923f6c13fc Use quotation for libspirv.h and sort headers. 2015-11-12 09:43:01 -05:00
Lei Zhang
df920ecb71 Remove table parameters from API functions.
Opcode, operand, and external instruction tables are no longer
passed in as parameters to spvTextToBinary(), spvBinaryToText(),
and spvValidate().
2015-11-12 09:42:50 -05:00
Andrew Woloszyn
ec691fba2a Fixed permissions 2015-11-11 12:12:16 -05:00
Andrew Woloszyn
55ecc2e097 Fixed warnings on windows and constness of spv_binary.
Replaced uint64_t with size_t in the places that make sense and
added spv_const_binary{,_t} to allow the interface to accept non
modifiable spirv where appropriate.
2015-11-11 12:12:13 -05:00
Lei Zhang
9049bb4877 Get magic number and version from spirv.h.
This is a part of the effort to clean up libspirv.h.
2015-11-11 12:12:10 -05:00
David Neto
2119694775 Disassembler support for OpSpecConstantOp
Document the fact that we use names for extended instructions
and OpSpecConstantOp opcode operands.
2015-11-11 12:12:07 -05:00
David Neto
0f166be68d Assembler support for OpSpecConstantOp
Adds SPV_OPERAND_TYPE_SPEC_CONSTANT_OP_NUMBER.
2015-11-11 12:12:04 -05:00
David Neto
ce9cef71ac Fix Windows build errors.
- uint64_t != size_t sometimes
- don't use C99 runtime sized arrays
2015-11-11 09:55:51 -05:00
David Neto
003e288c35 Enable round trip tests for OpSwitch 2015-11-11 09:55:42 -05:00
Lei Zhang
aa3cd5abb9 Optional typed literal number should be concrete in BinaryParser. 2015-11-10 15:58:16 -05:00
David Neto
2ae4a68fe8 Fix handling of OpExtInstImport
The assembler tracks mapping of extended instruction import Id
to extended instruction type.

Adds a few new ways to fail.
2015-11-10 15:58:07 -05:00
David Neto
9e545d7968 Assembler supports hex float constants.
The bit pattern for a hex float is preserved through
assembly and disassembly.

You can use a hex float to express Inf and any kind of NaN
in a portable way.
2015-11-10 15:58:03 -05:00
David Neto
229b90f6f4 Print OpConstant values according to type.
Zero and normal floating point values are printed with enough
enough digits to reproduce all the bits exactly.
Other float values (subnormal, infinity, and NaN) are printed
as hex floats.

Fix a binary parse bug: Count partially filled words in a
typed literal number operand.

TODO: Assembler support for hex numbers, and therefore reading
infinities and NaNs.
2015-11-10 15:57:59 -05:00
Lei Zhang
8f6ba14b58 advance() should check current string index is in bound. 2015-11-10 15:57:52 -05:00
Lei Zhang
9042f40f7c Remove support for canonical assembly syntax format. 2015-11-10 15:57:49 -05:00
David Neto
201caf7001 spv_operand_type_t cleanup.
- Concrete operand types are never optional.
  Split them to make this so, e.g. add SPV_OPERAND_TYPE_IMAGE
  since there was SPV_OPERAND_TYPE_OPTIONAL_IMAGE.
  Similarly for SPV_OPERAND_TYPE_MEMORY_ACCESS.
  This entails duplicating two operand table entries.

- The above, plus some rearranging of enums, allows us to define
  first and last optional operand types, and first and last
  variable operand types.
  This lets us simplify the code for spvOperandIsOptional, and
  spvOperandIsVariable.

- Replace SPV_OPERAND_TYPE_MULTIWORD_LITERAL_NUMBER with the
  more accurately named SPV_OPERAND_TYPE_TYPED_LITERAL_NUMBER.
  Its special characteristic is that the type of the literal
  number is determined by some previous operand in the instruction.
  This is used for literals in OpSwitch, OpConstant, and OpSpecConstant.
  This lets us refactor operand parsing cases in the assembler.

- Remove the special required-thing-in-optional-tuple in favour of
  the corresponding concrete operand type:
        SPV_OPERAND_TYPE_ID_IN_OPTIONAL_TUPLE
    --> SPV_OPERAND_TYPE_ID
        SPV_OPERAND_TYPE_INTEGER_LITERAL_IN_OPTIONAL_TUPLE
    --> SPV_OPERAND_TYPE_INTEGER_LITERAL

- Constrain spvOpeandTypeStr to only have to work for non-variable
  operand types.  Add a test for this.
2015-11-10 15:57:45 -05:00
David Neto
0ca6b59bfd Refactor the binary parser.
The binary parser has a C API, described in binary.h.
Eventually we will make it public in libspirv.h.
The API is event-driven in the sense that a callback is called
when a valid header is parsed, and for each parsed instruction.

Classify some operand types as "concrete".  The binary parser uses
only concrete operand types to describe parsed instructions.

The old disassembler APIs are moved into disassemble.cpp

TODO: Add unit tests for spvBinaryParse.
2015-11-10 15:57:33 -05:00
Dejan Mircevski
0981b1514e Slay "undeflow." 2015-11-10 15:57:31 -05:00
Dejan Mircevski
fc10eac868 Refactor EncodeTestHelper() and DecodeTestHelper(). 2015-11-10 15:57:28 -05:00
David Neto
bae8851111 DiagnosticStream keeps a copy of the position.
Don't store the pointer, as the underlying storage
could disappear.  This a bit more flexible and robust.
2015-11-10 15:57:04 -05:00
Andrew Woloszyn
758b347983 Added double tests to HexFloat 2015-11-10 15:57:00 -05:00
David Neto
ea633a6427 Empty assembly text compiles to no instructions.
But it's still a valid module.
2015-11-10 15:56:57 -05:00
Lei Zhang
1a0334edee Run clang-format to enforce Google style globally.
Note that we are more strict than Google style for one aspect:
pointer/reference indicators are adjacent to their types, not
their variables.

find . -name "*.h" -exec clang-format -i {} \;
find . -name "*.cpp" -exec clang-format -i {} \;
2015-11-10 15:56:47 -05:00
Andrew Woloszyn
6dfd4186f5 Updated HexFloat parsing for windows.
It is valid for float values to be modified on copy if they are NaN,
so long as they remain the correct NaN. What this means is that
we can not rely on the float data-type for storing float values
if we want to retain bit patterns.

Added FloatProxy which stores data in an unsigned integer, and updated
the HexFloat template to deal with FloatProxy values instead.
2015-11-02 13:52:31 -05:00
Andrew Woloszyn
0b61c593a2 Added HexFloat reading to HexFloat.
This allows reading of hex-encoded floats.
2015-11-02 13:52:28 -05:00
Andrew Woloszyn
c2887f98bc Added HexFloat helper class to print out floating point numbers.
TODO Add double tests before we actually use this.
2015-11-02 13:52:25 -05:00
Lei Zhang
b36e704a47 Use spirv.h instead of spirv.hpp.
This is a part of the effort to make libspirv.h C-compatible.
2015-11-02 13:52:20 -05:00
David Neto
9f79d789ec Inclusion guards follow Google C++ style
Follow the scheme in
http://google-styleguide.googlecode.com/svn/trunk/cppguide.html#The__define_Guard
except:
 - 'include/' is dropped from the guard token
 - 'source/' is dropped from the guard token
2015-11-02 13:52:15 -05:00
David Neto
db901b622e Move endian utilities to their own file. 2015-11-02 13:52:12 -05:00
David Neto
fcc7d58353 Move AssemblyGrammar into its own source files 2015-11-02 13:52:05 -05:00
David Neto
e5e99b57cc Add DISABLED failing test for ext inst import confusion
Currently, the assembler does not track external import
IDs when there are two or more.

TODO(dneto): Fix this functionality.
2015-11-02 13:52:00 -05:00
David Neto
b6a43383e7 Use basic_string<char> for the UTF-8 test string
This is an attempt to fix the unit tests on DeveloperStudio 2013.
Currently, the size of the earth_africa string is reported as 2
on Windows.  But I think that may be 2 16-bit characters.
2015-11-02 13:51:57 -05:00
David Neto
1780fc4fcf Update to latest registered generator enums
The registry is at:
https://www.khronos.org/registry/spir-v/api/spir-v.xml
2015-11-02 13:51:54 -05:00
Andrew Woloszyn
f731cbf6c4 Added set_bits<> to bitutils.
This allows us to get a constant with the given bits set at
compile-time. This is needed for a future patch for HexFloat.
2015-11-02 13:51:51 -05:00
David Neto
e169a7cd86 Test cleanup: remove useless std::string constructor 2015-11-02 13:51:48 -05:00
Andrew Woloszyn
fe6527fd80 Fixed typo in utf8 string literal 2015-10-26 12:55:33 -04:00
Andrew Woloszyn
4274f93065 Fixed UTF-8 encoding issues on windows.
Also cleaned up some warnings related to implicit size_t/uint64_t
conversions.
2015-10-26 12:55:33 -04:00
Andrew Woloszyn
157e41bf57 Added type tracking to the disassembler.
TODO: Actually use the tracked types to make sure that we print out
values correctly.
2015-10-26 12:55:33 -04:00
David Neto
4942a318f4 Remaining tests for OpenCL extended instructions. 2015-10-26 12:55:33 -04:00
David Neto
21c4ad4b22 Add OpenCL extended instructions.
Versions 1.2, 2.0, and 2.1 all use the same
extended instruction list.

Updated the source code patch for the SPIR-V doc generator,
so it can both generate the core syntax table, and also the
OpenCL extended instructions table.

Tested the Math and Common functions.
TODO: test the remaining entries.
2015-10-26 12:55:33 -04:00
Lei Zhang
6483bd7132 Enforce suitable literal number operands to be unsigned integers.
Except for OpConstant and OpSpecConstant, all other literal number
operands are indeed unsigned integers. So,

* Rename all *LITERAL_NUMBER* operand types to *LITERAL_INTEGER*.
* Expect unsigned integers for *LITERAL_INTEGER* operands.
* Keep MULITPLE_WORD_LITERAL untouched since it is only used by
  OpConstant and OpSpecConstant.

And we want to provide the capability to specify floating-point
numbers after !<integer> in the alternate parsing mode. So,
OPTIONAL_LITERAL_NUMBER is reserved for OPTIONAL_CIV.
2015-10-26 12:55:33 -04:00
Lei Zhang
62f92cdae6 Use normal string instead of raw string for simple strings. 2015-10-26 12:55:33 -04:00
Lei Zhang
4c2db8347e Simplify methods for compilation failure in test framework.
Remove  CompileWithFormatFailure() and make CompileFailure() accept
a default argument.
2015-10-26 12:55:33 -04:00
David Neto
9e11c7d301 Fix Android build: use std::numeric_limits<T>::max and friends
UINT32_MAX and friends are not available.
2015-10-26 12:55:33 -04:00
David Neto
470405e0ed Add disabled tests for disassembler literal emission
Currently, negative numbers are printed as if they were
first statically cast to unsigned.
2015-10-26 12:55:33 -04:00
Andrew Woloszyn
e59e6b72f9 Added proper string escaping to the disassembler.
The disassembler now generates correct string when encountering
quotes and slashes.
2015-10-26 12:55:33 -04:00
David Neto
51013d15ee SPV_FAILED_MATCH error mutes DiagnosticStream.
The DiagnosticStream will not emit the accumulated message
text if the error is SPV_FAILED_MATCH.

Change various interfaces to accept the intended error
code instead of a boolean "is_optional".  This allows
us to avoid repeating the following type of logic deep
inside helper methods:

  if (is_optional) return SPV_FAILED_MATCH;
  return diagnostic() << " message text ";
2015-10-26 12:55:33 -04:00
Andrew Woloszyn
f08c679e97 Removed un-needed comments and fixed merge issue. 2015-10-26 12:55:33 -04:00