| Age | Commit message (Collapse) | Author |
|---|
|
* Defer immutable buffer loads when emitting spirv.
* Fix.
* Fix.
* Fix.
* Fix tests.
* Fix test.
|
|
* Fix false negative result for CUDA with recent versions
From CUDA version 12.8 and above, nvrtc returns an exit code treated as
an error. Some of slang-test test cases had to change from TEST to
DIAGONOSTIC_TEST to handle it properly.
|
|
* Add additional completion keywords.
* LanguageServer: Enhance auto completion for `override`.
|
|
* Require `override` keyword for overriding default interface methods.
* Update doc.
* Fix test.
|
|
* Diagnose on use of struct inheritance.
* fix test.
* Fix tests.
* fix.
---------
Co-authored-by: ArielG-NV <159081215+ArielG-NV@users.noreply.github.com>
|
|
This PR replaces enable/disable style C function calls with C++ RAII style code.
In debug build, when an assertion failed in between enable and disable functions, an exception is thrown and the disable function is not called. RAII style code is safer for an exception
|
|
* Fix#6993 - Emit Diagnostic Warning and Fix SIGSEGV
* Update external/slang-rhi submodule
* Add checks for valid stage names for paq in SemanticsVisitor check
* format code
---------
Co-authored-by: slangbot <186143334+slangbot@users.noreply.github.com>
Co-authored-by: Ellie Hermaszewska <ellieh@nvidia.com>
|
|
* Add legalization for 0-sized arrays.
* Allow 0-sized arrays in the front-end.
* More tests.
* Add `Conditional<T, hasValue>` type to core module.
* Update toc.
* Fix wording.
* Update test.
|
|
|
|
Fixes #6987
|
|
* Implement default initializer list for C-Style type member
Close #6189.
Previsouly, for the C-Style member in a struct, if it doesn't have any initialize
expression, when we synthesize the ctor, we will not associate the
default value for the parameter corresponding to that member.
This bring some trouble that existing slang users has to add '= {}' to
every struct fields in order to make all the parameters in the synthesized ctor having
a default value, so people can still use `Struct a = {}` to create a
struct.
To make this use case convenience, we will automatically associated a
'= {}' as the default value for this case.
This PR also add support for empty initializing link-time sized vector/matrix by "= {}".
In addition, this PR also fix a bug in auto diff where we should not report error when proccessing
transpose on an empty struct.
|
|
The user can explicitly use Vulkan memory model, or it will be
automatically used when cooperative-matrix is used.
When vulkan memory model is used, two keywords, "Coherent" and
"Volatile", are not allowed.
There are many differences regarding atomic and texture but
this PR has changes limited to support `globallycoherent`
keyword. When variables with `globallycoherent` is used with `OpLoad`, it
will use additional options, `MakePointerAvailable|NonPrivatePointer`,
that will provide the same effect. For `OpStore`, it will use
`MakePointerVisible|NonPrivatePointer`.
|
|
* Add checking for hlsl register semantic.
* Fix.
* Fix test.
* Fix switch error.
* Fix tests.
|
|
* Rename 'main' on some backednds
Fixes #5542
Some backends like cuda, metal, cpu do not allow 'main' as the entry
point. This commit adds a new warning that is emitted when a program
uses main as the entry point. In addition to emitting the warning, it
internally renames the entry point to "main_". It also adds a test to
check these for the three backends.
* format code
* move test to diagnostics
* rename test to diagnostic
* generate unique name
---------
Co-authored-by: slangbot <186143334+slangbot@users.noreply.github.com>
|
|
* Error out on invalid vector sizes
* Remove unnecessary include
* Fix incorrect assert
* Add test
|
|
Fixes #7049
The root cause of the problem in #7049 is simply that newer NVRTC versions produce a warning when asked to generate code for older CUDA SM versions, and the default that Slang was requesting compilation for was old enough to trigger that warning, and thus trip up the test case (which only looks at the first diagnostic produced by the downstream compiler).
Superficially, the fix was easy: change the test case in question (`tests/diagnostics/local-line.slang`) to request `-capability cuda_sm_8_0`, the minimum version supported by current NVRTC.
Unfortunately, the simple fix required some other fixes in order to actually work.
The capability system includes capability names of the form `cuda_sm_*_*`, but specifying such a capability had *no* impact on the CUDA SM version passed in when invoking NVRTC.
Instead, only the CUDA SM versions requested in the implementation of intrinsics in the core module were affecting the version number passed down.
This change adds logic to `slang-compiler.cpp` to take explicitly requested capabilities into account when inferring the CUDA SM version to be passed downstream.
A more complete fix would also add similar logic for all the other targets.
Unfortunately... yet again... that fix wasn't enough to make things work as expect.
Now I had the problem that requesting `-capability cuda_sm_8_0` was actually causing the NVRTC invocation to request CUDA SM version **9.0**!
The underlying problem *there* was that the `slang-capabilities.capdef` file has defined certain capability names in a way that implies atomic capabilities much higher than one would expect.
E.g., the `cuda_sm_8_0` alias was including HLSL `sm_5_0`, but then `sm_5_0` in turn included `_cuda_sm_9_0`.
The fix, for now, is to change the definitions in `slang-capabilities.capdef` to not have the counter-intuitive definitions for `cuda_sm_*_*`.
With this set of fixes, the test failure in the original bug report no longer occurs.
The work that went into this change suggests several larger-scope fixes that would be good to pursue:
* Ideally the capability definitions would have some sort of validation checking to make sure that counter-intuitive results like `cuda_sm_8_0` requesting CUDA SM 9.0 do not occur.
* The translation of capabilities over to version numbers for a downstream compiler should be expanded to cover other targets, and not just CUDA. It might be better/simpler to just pass the capabilities themselves to the downstream compiler, since it is possible that a downstream compiler could have more fine-grained enable/disable options than a simple version number.
* The entire approach to computing version numbers required for downstream compilation should be cleaned up so that we don't have this duplication between the capabilities that represent those versions and separate syntactic constructs that are used to "request" those versions as part of code generation.
* We are very much at the point where we should consider dropping the current behavior where a profile name or capability like `sm_5_0`, that is specific to a single target or a subset of targets, also implies a set of comparable capabilities for other targets.
|
|
* Fix #6544: Properly format nested type names in extensions
Modify DeclRefBase::toText to properly handle types defined
in extensions by qualifying them with their parent type name.
This ensures getFullName() returns the full name like
'FullPrecisionOptimizer<half>.State' instead of just '.State'.
Also handle other nested types in structs/classes similarly.
* Update extension reflection handling - with generics args and namespaces
- stopping namespace inclusion for extension members
- Update to use getTargetType() to handle the generic arguments
- update test cases
* Simplify code to remove using parentDecl
|
|
* Initial implementation for SPIRV, GLSL and Metal
* test add bool test
* Fix and improve subgroup rotate tests
* Add proper GLSL extensions and proper Metal type checking
* Clean up tests and add diagnostics test for subgroup type for Metal
* Update wave-intrinsics docs
|
|
* Add struct member offset qualifier for SPIRV
* Implement for GLSL target and add tests
* clean up
* fix formatting
* fix typo
* renamed GLSLStructOffset to VkStructOffset and added emit-spirv-via-glsl test case
|
|
* Fix matrix division by scalar for Metal and WGSL targets
* Add tests
* Minor fix
* Fix compilation error
* Convert to multiplication for WGSL
* Minor cleanup
---------
Co-authored-by: Yong He <yonghe@outlook.com>
|
|
interface-typed output parameter (#6788)
* More specific diagnostic for invalid concrete-to-interface arg coercion
* Add test for the new error message
* Fix typo in expected test result
|
|
* Add support for Ray Payload Access Qualifiers (PAQs) (#3448)
- Added [raypayload] attribute for struct declarations
- Implemented field validation requiring read/write access qualifiers
- Added diagnostic error for missing qualifiers
- Enabled PAQs in DXC compiler and HLSL emission
- Added new test demonstrating PAQ syntax
- Implemented proper handling of ray payload attributes in IR generation
* format code
* Cleanup: Remove unused vars
* Add check to enablePAQ only for profile >= lib_6_7
* Review Fix - Add PAQ support for DX Raytracing
add enablePAQ flag to DownstreamCompileOpitons, improve PAQ handling
update raypayload-attribute-paq.slang to ensure hlsl and dxil is
validated
* Add diagnostic test for missing paq for lib_6_7
Compile using `-disable-payload-qualifiers` aka lib_6_6 profile
raypayload-attribute-no-struct.slang and
raypayload-attribute.slang
---------
Co-authored-by: slangbot <186143334+slangbot@users.noreply.github.com>
Co-authored-by: Ellie Hermaszewska <ellieh@nvidia.com>
|
|
* initial wip
* more WIP
* preserve old lower behavior
* remove unnecessary includes
* add test
* add no target case in test
* fix broken test
---------
Co-authored-by: Ellie Hermaszewska <ellieh@nvidia.com>
|
|
* initial wip
* more wip
* add test
* add unexpected for invalid target
* fixups and improve error message
* fixups and improve error message
* remove incorrect comment
---------
Co-authored-by: Ellie Hermaszewska <ellieh@nvidia.com>
|
|
* Allow partial specialization of existential arguments.
* Fix.
* Add test case for improved diagnostics.
* Fix compile error.
* Fix tests.
* Fix.
* Fix test.
* Fix compile issue.
* Fix typo.
* Address comment.
|
|
* Make capability diagnostic message more friendly.
* Fix.
* Fix.
* Fix.
* Fix test.
* Update expected fail setting for aarch64/linux
* Fix.
|
|
* Allow `.member` syntax on vector and scalars.
* Fix.
* fix.
* Fix.
* update comment.
* Fix tests.
* Fix warning.
* Add more tests.
|
|
* Clone name hint decoration when emiting Undefined
When emiting "Undefined", we lost the information of where it was
synthasized from. This prevents us from providing more helpful error
messages.
The issue was the when we handle "IRLoop", the inputs parameters to the
Phi didn't clone the name hint decoration. This commit clones them when
emiting "Undefined".
* Adding more test case
---------
Co-authored-by: Yong He <yonghe@outlook.com>
|
|
* Legalize array size of SV_TessFactor and SV_InsideTessFactor
When targeting SPIR-V or GLSL, the type for SV_TessFactor must be
float[4]; note that it is not a vector but an array.
Similarly the type of SV_InsideTessFactor has to be an array of
float[2].
When the user shader declare them as arrays smaller than the required
size, Slang will legalize them to the required sizes.
Note that it is not the user mistake to declare floar[3] when the hull
mode is in "tri", as an example. The unused components are expected to
be ignored.
Note also that HLSL allows the type to be float[2|3|4] whereas GLSL and
SPIR-V requires it to be always float[4].
* Change from "quad" domain mode to "tri"
* Handle error case
---------
Co-authored-by: Yong He <yonghe@outlook.com>
|
|
* Fix zero size array handling in slangc
Fixes #2890
1. Fix zero size array handling in slangc
2. Add new zero size array diagnostic test.
* format code
* fix review comments
---------
Co-authored-by: slangbot <186143334+slangbot@users.noreply.github.com>
Co-authored-by: Ellie Hermaszewska <ellieh@nvidia.com>
|
|
Closes #2890
Co-authored-by: Ellie Hermaszewska <ellieh@nvidia.com>
|
|
* Neaten metal parameter block checking
* Disallow only resources in constant buffers in parameterblocks on metal
closes https://github.com/shader-slang/slang/issues/6200
* add unit test for metal parameterblock cbuffer
---------
Co-authored-by: Yong He <yonghe@outlook.com>
|
|
* Fix crash when handling uninitialized resource type
close #6328.
When declare a var with struct type, if the struct has a resource
type field and it doesn't provide explicit constructor, because
slang won't implicit construct such variable at declare site if user
doesn't explicitly call the initializer list, we should report the
resource type field uninitialized as an error to prevent future possible
crash when legalize any use of such variable.
|
|
Co-authored-by: Yong He <yonghe@outlook.com>
Co-authored-by: Ellie Hermaszewska <ellieh@nvidia.com>
|
|
* SP004: implement initialize list translation to ctor
- We synthesize a member-wise constructor for each struct follow
the rules described in SP004.
- Add logic to translate the initialize list to constructor invoke
- Add cuda-host decoration for the synthesized constructor
- Remove the default constructor when we have a valid member init constructor
- Disable -zero-initialize option, will re-implement it in followup (#6109).
- Fix the overload lookup issue
When creating invoke expression for ctor, we need to call
ResolveInvoke() to find us the best candidates, however
the existing lookup logic could find us the base constructor
for child struct, we should eliminate this case by providing
the LookupOptions::IgnoreInheritance to lookup, this requires
us to create a subcontext on SemanticsVisitor to indicate that
we only want to use this option on looking the constructor.
- Do not implicit initialize a struct that doesn't have explicit default
constructor.
Co-authored-by: slangbot <186143334+slangbot@users.noreply.github.com>
|
|
* Fix warning about push constants, shaderRecordEXT
These resources should not warn about missing vk::binding.
* format code
---------
Co-authored-by: slangbot <186143334+slangbot@users.noreply.github.com>
|
|
* Fix combined sampler documentation and warning
* Update comment, show detailed '-fvk-t-shift' message in warning instead of generic '-fvk-xxx-shift'
* format code
---------
Co-authored-by: slangbot <186143334+slangbot@users.noreply.github.com>
|
|
* add brga8 format
* add tests
* minor fixes
* cleanup
* maybe fix broken quad control test
* add missing xslang flag on test
---------
Co-authored-by: Yong He <yonghe@outlook.com>
|
|
Co-authored-by: Yong He <yonghe@outlook.com>
|
|
* Fix incorrect resolve of specialization instance
While checking the uninitialized variables, we were not resolving the
specialized instance correctly. This commit repeats the resolve while
the result is a specialization instance. A new test is added for this:
tests/diagnostics/uninitialized-generic.slang
After the problem is fixed, it revealed another problem in existing
tests:
tests/compute/nested-generics2.slang
tests/diagnostics/uninitialized-local-variables.slang
When a struct has a member variable whose type is a generic type,
we cannot iterate over its member variables yet, because the type
is unknown until the generic function/struct is specialized. We will
have to give up checking for these cases.
|
|
* Refactor _Texture to constrain on texel types.
* Fix tests.
* Fix.
* Disable glsl texture test because rhi can't run it correctly.
|
|
* Inline global constants that contains opaque handles for legalization.
* Add diagnostics on opaque type global variables.
* Fix.
* Fix test.
|
|
test (#6067)
Co-authored-by: Yong He <yonghe@outlook.com>
|
|
There are a few tests that are currently passing but could fail when
conditions change little bit.
When slang-test runs with `test-server`, the spirv validation fails and
the test gets reported as failed even though they are actually passing.
To avoid the potential problem, this commit adds an empty entry point.
Co-authored-by: Yong He <yonghe@outlook.com>
|
|
* Use disassemble API from SPIRV-Tools
This commit uses C API version of SPIRV disassemble function rather than
calling spirv-dis.exe.
This allows us to use a correct version of SPIRV disassble function that
Slangc.exe is using.
The implementation is mostly copied from external/spirv-tools/tools/dis/dis.cpp, which is a source file for building spirv-dis.exe.
This commit also includes a fix for a bug in RPC communication to `test-server`.
When an RPC connection to `test-server.exe` is reused and the second
test abruptly fails due to a compile error or SPIRV validation error,
the output from the first test run was incorrectly reused as the output
for the second test.
This commit resets the RPC result before waiting for the response so
that even when the RPC connection is erratically disconnected, the
result from the previous run will not be reused incorrectly.
Some of the tests appear to be relying on this type of behavior. By
using an option, `-skip-spirv-validation`, the RPC connection will
continue without an interruption.
|
|
|
|
(#5964)
|
|
|
|
* Add verification logic on push and specialization constants.
* format code
---------
Co-authored-by: slangbot <186143334+slangbot@users.noreply.github.com>
|
|
* Allow glsl set and binding layout qualifiers to be compile time integer exprs
* add new tests
* add comments
* cleanup on asserts
* addressed review comments and cleanup
* fix missing set expr in test
* fixed tests and cleanup
---------
Co-authored-by: Yong He <yonghe@outlook.com>
|
