summaryrefslogtreecommitdiff
path: root/source/slang
AgeCommit message (Collapse)Author
2025-07-29Fix CUDA backend missing U32_firstbitlow implementation (#7921)Copilot
* Initial plan * Add U32_firstbitlow implementation for CUDA and CPP backends Co-authored-by: bmillsNV <163073245+bmillsNV@users.noreply.github.com> * Add I32_firstbitlow and comprehensive testing for signed/unsigned firstbitlow Co-authored-by: bmillsNV <163073245+bmillsNV@users.noreply.github.com> * Convert firstbitlow test to use inline filecheck syntax Co-authored-by: ArielG-NV <159081215+ArielG-NV@users.noreply.github.com> * Add U32_firstbithigh and I32_firstbithigh implementations for CUDA and CPP backends Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> * Update prelude/slang-cpp-scalar-intrinsics.h * Update prelude/slang-cpp-scalar-intrinsics.h * Update prelude/slang-cpp-scalar-intrinsics.h * Refactor Metal bit intrinsics to handle zero case correctly Co-authored-by: ArielG-NV <159081215+ArielG-NV@users.noreply.github.com> * Update slang-cuda-prelude.h remove fake links * Update hlsl.meta.slang * if -1, return -1 due to implicit hlsl rule * -1 or 0 is ~0u as per hlsl implictly * 0 or -1 as per hlsl * fix the math to map to hlsl * fix compile error * forgot `31 - clz` * format code (#7943) Co-authored-by: slangbot <186143334+slangbot@users.noreply.github.com> * Update source/slang/hlsl.meta.slang * Update source/slang/hlsl.meta.slang * Update source/slang/hlsl.meta.slang * Update source/slang/hlsl.meta.slang --------- Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com> Co-authored-by: bmillsNV <163073245+bmillsNV@users.noreply.github.com> Co-authored-by: ArielG-NV <159081215+ArielG-NV@users.noreply.github.com> Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> Co-authored-by: ArielG-NV <aglasroth@nvidia.com> Co-authored-by: slangbot <ellieh+slangbot@nvidia.com> Co-authored-by: slangbot <186143334+slangbot@users.noreply.github.com>
2025-07-28Fix issue in multi-level break elimination by handling multi-level continue ↵Sai Praveen Bangaru
statements (#7953)
2025-07-28Update slang-rhi to fix CI error on debug. (#7946)Yong He
* Update slang-rhi * Fix profile lookup on `nullptr`. * update rhi
2025-07-25Fix mesh shader reflection JSON output (#7868)pdeayton-nv
* Fix mesh shader reflection JSON output Fixes issue #7736 where mesh shaders were showing stage "UNKNOWN" and missing type information in reflection JSON output. Changes: - Add missing SLANG_STAGE_MESH and SLANG_STAGE_AMPLIFICATION cases to shader stage switch statements in emitReflectionVarBindingInfoJSON and emitReflectionEntryPointJSON - Add missing MeshOutput to TypeReflection::Kind enum in slang.h - Add missing type kind cases (OutputStream, MeshOutput, Specialized, None) to emitReflectionTypeInfoJSON Testing: - Simple mesh shader now correctly shows "stage": "mesh" instead of "UNKNOWN" - Complex mesh shader with parameters shows proper stage and input type information - Output parameters show "kind": "None" instead of crashing with assertion 🤖 Generated with [Claude Code](https://claude.ai/code) Co-authored-by: pdeayton-nv <pdeayton-nv@users.noreply.github.com> * Remove test files from mesh shader reflection fix As requested, removed test files since there is no testing infrastructure for reflection JSON output. Focus is now only on the core mesh and amplification shader reflection fixes. Co-authored-by: pdeayton-nv <pdeayton-nv@users.noreply.github.com> * Address review comments: move MeshOutput enum to end and restore assert - Move MeshOutput enum member to end of TypeReflection::Kind enum for backward compatibility - Replace fprintf with SLANG_ASSERT for unhandled type kinds Co-authored-by: pdeayton-nv <pdeayton-nv@users.noreply.github.com> --------- Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com> Co-authored-by: pdeayton-nv <pdeayton-nv@users.noreply.github.com> Co-authored-by: Yong He <yonghe@outlook.com>
2025-07-25Fix nv_viewport_mask semantic to emit SpvBuiltInViewportMaskNV (#7904)davli-nv
Fixes #7903 SPV_NV_viewport_array2 says ViewportMaskNV corresponds to gl_ViewportMask
2025-07-25Fix for Generic Function Redefinition Error (#7891)Gangzheng Tong
* emit literal values in getTypeNameHint for bool, str etc. * add test for specializing generics with bool literals * fix build error * add specializing with Enum type test
2025-07-25Add combined texture-sampler flag to reflection API to differentiate ↵Copilot
Texture2D from Sampler2D (#7901) * Initial plan * Add SLANG_TEXTURE_COMBINED_FLAG to differentiate combined texture-samplers Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> * Fix regression in hlsl-to-vulkan-combined test by updating expected output Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> --------- Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com> Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com>
2025-07-25Fix compiler crash when enum is used as vertex output data (#7915)Copilot
* Initial plan * Initial investigation and plan for enum vertex output fix Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> * Fix compiler crash when enum is used as vertex output data Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> * Fix compiler crash when enum is used as vertex output data Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> * Address reviewer feedback: use SLANG_ASSERT and improve CHECK directives Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> --------- Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com> Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com>
2025-07-25Fix SPIRV OpMemberName member indices (#7912)davli-nv
OpMemberName instructions were all using member index 0 instead of incrementing indices (0, 1, 2, ...) as required by the SPIR-V spec. The bug was that the member index increment (id++) was only happening for physical struct types, but OpMemberName emission occurs for all struct types when they have name decorations. This fix ensures member indices are properly incremented for both physical and non-physical struct types. Fixes #7909 Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com> Co-authored-by: davli-nv <davli-nv@users.noreply.github.com>
2025-07-24Avoid early specialization for witness tables in SimplifyIR (#7636)Jay Kwak
* Avoid early specialization for witness tables in SimplifyIR Prevents SimplifyIR from prematurely specializing witness tables before the main specialization pass. Witness tables are hoistable immutable objects that must maintain consistent signatures to avoid incorrect deduplication. SimplifyIR was incorrectly transforming expressions like "witness_table_t(%IFoo)(specialize(%7, %GenericValue4))" into "witness_table_t(%IFoo)(%Foo)" even when %GenericValue4 was unused. Fixes #7233 * Add a missing test file
2025-07-24Organize code better by splitting some big files (#7890)Theresa Foley
* Organize code better by splitting some big files The basic change here is that the majority of the declarations in `slang-compiler.h` have been split out into a set of smaller and more focused files. As a result, the implement of those declarations have been moved from `slang-compiler.cpp` and `slang.cpp` over to those new files when the proper home for code is obvious. I have tried as much as possible to *not* make any edits to the code along the way, and just copy-paste declarations from one place to another as-is. The exceptions I am aware of are: * In some cases a function that used to be file-scope `static` was used by code that landed in two or more different `.cpp` files. In these cases, I changed the function to be non-`static` (removing the `_` prefix from its name, if it had one, per our naming conventions), and put a declaration for the function into the most appropriate header I could identify. * I added a few comments in places where I saw ugly or unfortunate things in the code I was moving, and wanted to tag them with `TODO`s so we can hopefully get to them in the fullness of time. * I added top-level comments to each of the new `.h` files that was introduced to try to explain the logic for what goes into that file. * In cases where one of the new header files mostly existed to declare a single type, I sometimes added more detail to the doc comment on that type, to better explain the type and its role in the compiler (this is text that otherwise might have gone into the comment at the top leve lof the file, but I figured that the doc comment would have higher discoverability). I expect that the most contentious choice here is that the `Session` class lands in `slang-global-session.h` while `slang-session.h` holds the `Linkage` class. The names used in this change are consistent with how the relevant concepts in the public Slang API are named, and are consistent with how we *intend* to rename the classes themselves in time. * format code * fixup --------- Co-authored-by: slangbot <186143334+slangbot@users.noreply.github.com>
2025-07-24Fix confusing error messages for interface return type mismatches (#7854)Copilot
* Initial plan * Add improved diagnostic for interface return type mismatches Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> * Complete fix for interface return type mismatch error reporting Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> * Move diagnostic to synthesis phase for better interface return type mismatch errors Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> * Remove extraneous test file and update .gitignore Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> * Add diagnostic test for interface return type mismatch and apply formatting Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> * Address feedback: restore whitespace and use filecheck for diagnostic test Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> * Fix logic error in return type mismatch detection Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> * Remove unnecessary flag by using out parameter for diagnostic tracking Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> * Refactor witness synthesis failure reporting to use structured approach Replace ad-hoc `outSpecificDiagnosticEmitted` parameter with `WitnessSynthesisFailureReason` enum and `MethodWitnessSynthesisFailureDetails` struct as requested in code review. This provides: - Clear taxonomy of failure reasons (General, MethodResultTypeMismatch, MethodParameterMismatch) - Centralized diagnostic emission in findWitnessForInterfaceRequirement - Better extensibility for future failure types - Improved maintainability by removing state tracking flags The return type mismatch diagnostic continues to work correctly, showing error 38106 with precise location information. Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> * Remove unused MethodParameterMismatch enum and duplicate code Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> * Remove redundant requiredMethod field from MethodWitnessSynthesisFailureDetails Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> * Address feedback: add outFailureDetails guard and remove unnecessary hasReturnTypeError variable Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> * Fix regression: restore original diagnostic message for mutating method mismatch Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> * Fix. * Fix. * Remove `innerSink`. * Print candidates considered for interface match upon error. * Fix tests. --------- Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com> Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> Co-authored-by: Yong He <yonghe@outlook.com>
2025-07-24Include Inst UID in some assert failure messages. (#7882)Yong He
2025-07-24Skip creating transitive witnesses for interface conformances that are ↵aidanfnv
inherited through structs (#7597) * Eliminate null-pointer reference in multilevel struct interface inheritance * Add test * Replace witness nullptr guard with conformance check to skip the witness * Update test * Skip creating transitive witness entirely for struct-struct-interface * format code (#14) Co-authored-by: slangbot <186143334+slangbot@users.noreply.github.com> * Removes casts, check for identity witness --------- Co-authored-by: slangbot <ellieh+slangbot@nvidia.com> Co-authored-by: slangbot <186143334+slangbot@users.noreply.github.com>
2025-07-24Document how to use LLDB in the Slang codebase (#7809)Sam Estep
* Document how to use LLDB in the Slang codebase * Include `slang_lldb.py` in `.lldbinit` * Switch from GCC to Clang * Include a VS Code task to build before debugging * Fix clangd
2025-07-23Add LLDB data formatters for Slang IR (#7828)Sam Estep
2025-07-23fix #7869 (#7871)ArielG-NV
2025-07-22Fix visibility of synthesized Differential typedefs. (#7865)Yong He
* Fix visibility of synthesized `Differential` typedefs. * Delete incorrect test.
2025-07-22Fix public unscoped enum constants not visible across module boundaries (#7864)Copilot
* Initial plan * Fix public unscoped enum constants visibility across module boundaries Add visibility modifier copying in CompleteDecl for unscoped enum cases. When synthesizing static const declarations for unscoped enum cases, copy the visibility modifiers from the original enum declaration to ensure they have the same visibility scope across module boundaries. Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> * Create new visibility modifier instances instead of sharing existing ones Address reviewer feedback to avoid sharing modifier instances between declarations since modifiers form a linked list. Now creates new instances of the appropriate visibility modifier type (Public, Private, or Internal) instead of reusing the existing instance. Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> * Move unscoped enum visibility tests into subdirectory structure Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> * Use createByNodeType for visibility modifier creation as suggested Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> * format code (#7867) Co-authored-by: slangbot <186143334+slangbot@users.noreply.github.com> --------- Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com> Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> Co-authored-by: slangbot <ellieh+slangbot@nvidia.com> Co-authored-by: slangbot <186143334+slangbot@users.noreply.github.com>
2025-07-22Fix scalar to array conversion for tessellation factors (#7837)Yong He
* Fix scalar to array conversion for tessellation factors in GLSL legalization Add scalar-to-array conversion support in adaptType() function to handle cases where users declare scalar tessellation factors (e.g., float TessLevelInner) but GLSL requires arrays (float[2] for gl_TessLevelInner). This prevents the generation of BuiltinCast instructions that crash the SPIR-V emitter. Fixes crash: "unimplemented: Unhandled local inst in spirv-emit: BuiltinCast" - Add scalar-to-array case in slang-ir-glsl-legalize.cpp adaptType() - Fill all array elements with the scalar value for tessellation factors - Add test case for scalar tessellation factor conversion Fixes #7000 Co-authored-by: Yong He <csyonghe@users.noreply.github.com> * Apply review feedback fixes - Change test directive to TEST:SIMPLE - Use IRArrayType instead of IRArrayTypeBase - Use MakeArrayFromElement for cleaner scalar-to-array conversion Co-authored-by: Yong He <csyonghe@users.noreply.github.com> * Fix type conversion in scalar-to-array tessellation factor conversion Convert scalar value to array element type before creating array to handle cases where scalar type differs from array element type (e.g., int to float[3]). Co-authored-by: Yong He <csyonghe@users.noreply.github.com> * Use CHECK-DAG for order-independent tessellation factor checks Co-authored-by: Yong He <csyonghe@users.noreply.github.com> --------- Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com> Co-authored-by: Yong He <csyonghe@users.noreply.github.com>
2025-07-22Fix segfault when using -separate-debug-info with unsupported targets (#7777)Copilot
* Initial plan * Fix segfault when using -separate-debug-info with unsupported targets Add validation to emit a diagnostic error when -separate-debug-info is used with targets other than SPIR-V binary. Previously, this would cause a segfault because the separate debug info logic is only implemented for SPIR-V targets. Changes: - Added new diagnostic error (ID 18) for unsupported separate debug info usage - Added validation in OptionsParser::_parse() to check target compatibility - Created test cases for HLSL and GLSL targets to verify the fix - Updated error message to clarify only SPIR-V binary targets are supported The fix prevents segfaults and provides clear feedback to users about target limitations for the -separate-debug-info option. Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> * Address feedback: fix segfault properly instead of preventing it Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> --------- Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com> Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com>
2025-07-22Fix crash when private ctor is used for coercion. (#7858)Yong He
* Fix crash when private ctor is used for coercion. * Fix tests. * Fix. * Fix test error.
2025-07-22Fix findFunctionByNameInType to preserve functions with different signatures ↵Copilot
(#7827) findFunctionByNameInType was only returning one function when multiple functions existed with the same name but different signatures. This broke reflection functionality for extension methods. Fix the issue by changing findDeclFromStringInType by not calling maybeResolveOverloadedExpr if checkedTerm is overloaded functions. We still call maybeResolveOverloadedExpr when any candidates in the overloaded list is not DeclRefExpr referencing a function.
2025-07-22Add CI to check ir module versioning (#7821)Ellie Hermaszewska
2025-07-22Fix Conditioanl<T, false> fields with a semantic. (#7855)Yong He
* Fix Conditioanl<T, false> fields with a semantic. * Add unit test. * Fix test.
2025-07-21Fix C-style casts in GLSL pointer cast operations (#7841)Copilot
* Initial plan * Fix C-style casts in GLSL pointer cast operations Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> --------- Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com> Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com>
2025-07-21Add utility to trace creation of problematic IRInsts to assist LLM in ↵Copilot
debugging (#7820) * Initial plan * Add SLANG_DEBUG_IR_BREAK environment variable support Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> * Apply code formatting to SLANG_DEBUG_IR_BREAK implementation Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> * Improve stack trace debugging with -rdynamic flag and backtrace_symbols Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> * Address PR feedback: use PlatformUtil::getEnvironmentVariable, remove -rdynamic flag, and delete fallback branch Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> * Address PR feedback: simplify env var parsing, move backtrace to PlatformUtil, use #if for SLANG_LINUX_FAMILY Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> * Address PR feedback: remove unneeded include, make backtrace() more generic by removing uid parameter Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> * Fix and clone source tracking. * Add python script to dump traces. * Update instructions. * Batch calls to addr2line * Cleanup claude instructions. * update claude action. * Remove duplicated build instructions from claude.yml workflow Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> * fix build error. * Fix build errors --------- Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com> Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> Co-authored-by: Yong He <yonghe@outlook.com> Co-authored-by: Gangzheng Tong <tonggangzheng@gmail.com>
2025-07-18Fix enum array indexing by adding implicit conversion support (#7815)Copilot
* Initial plan * Fix enum array indexing by adding implicit conversion support Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> * Update enum array indexing test to support GPU backends Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> --------- Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com> Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com>
2025-07-18Fix debug info generation for let variables in SPIR-V output (#7743)Copilot
* Initial plan * Fix debug info for let variables Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> * Fix parameter count for emitDebugVar function call Fixed regression where let variable debug info generation was missing the optional argIndex parameter in emitDebugVar call. Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> * Add location validity check for debug info generation Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> * Don't insert debug value for nondebuggable types. --------- Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com> Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> Co-authored-by: Yong He <yonghe@outlook.com>
2025-07-18Lower int/uint/bool matrices to arrays for SPIRV (#7687)venkataram-nv
* Add tests for expected behaviour * Allow matrix types in logical or/and * Legalize int/bool matrix types and construction with makeMatrix * Legalize uint matrices and operations * Limit testing to only SPIRV * Better tests for int and bool * Add test for uint * Remove GLSL tests * Remove old test for diagnosing int matrices * Emit SPIRV directly in tests * format code * Address PR comments * Improve testing * Address PR comments * format code * Add tests for matrix intrinsic operations * Move matrix lowering to dedicated legalization pass * Fix compiler warning * Remove signal again * Reorder matrix and vector legalization * Fix formatting * Add shift and comparison tests --------- Co-authored-by: slangbot <186143334+slangbot@users.noreply.github.com>
2025-07-18Add bounds checking for out-of-bounds array access with constant indices (#7814)Copilot
* Initial plan * Implement out-of-bounds array access checking Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> * Add tests and format code for array bounds checking Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> * Move checkForOutOfBoundAccess to separate file and refactor using InstPassBase Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> * Optimize out-of-bounds checker to use single IR traversal Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> * Fix DiagnosticSink forward declaration from struct to class Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> * Update 0-array-1 test to use runtime indices to avoid bounds checking diagnostic Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> * Use SV_DispatchThreadID for truly runtime array access in 0-array-1 test Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> --------- Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com> Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com>
2025-07-17Fix GLSL global const diagnostic regression (#7808)Copilot
* Initial plan * Fix GLSL global const diagnostic regression - add test exclusion for GLSL module Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> --------- Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com> Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com>
2025-07-17Prelink ForceInlined functions during lowering. (#7812)Yong He
* Prelink ForceInlined functions during lowering. * Fixes and cleanups. * Fix warning. * Fix crash.
2025-07-17slang: Add support for generating getters for IR struct defs. (#7725)Sruthik P
This change expands the IR struct definition generation logic in slang-ir.h.lua to code generate the getters for the operands of an IR. To facilitate the above, the schema for the IR definitions in slang-ir-insts.lua is updated to allow for explicit specification of the operands of an IR, with Fiddle code generating the getters for them. slang-ir.h is updated to remove the hardcoded getters of the IRs since they are now generated by Fiddle. Fixes part of #7185
2025-07-17fix const qualifier for specialization constants (#7640)Sirox
Co-authored-by: Yong He <yonghe@outlook.com>
2025-07-17Fix BitCast emitting bit shifts with too small integer types (#7792)Julius Ikkala
Co-authored-by: Yong He <yonghe@outlook.com>
2025-07-17Merge NamePool and RootNamePool into a single type (#7797)Copilot
* Initial plan * Merge NamePool and RootNamePool into single NamePool class Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> * Remove unnecessary comment from slang-fiddle-scrape.cpp Co-authored-by: Theresa Foley <tangent-vector@users.noreply.github.com> * Address review feedback: initialize namePool to nullptr and remove unnecessary comments Co-authored-by: Theresa Foley <tangent-vector@users.noreply.github.com> --------- Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com> Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com> Co-authored-by: Theresa Foley <tangent-vector@users.noreply.github.com> Co-authored-by: Yong He <yonghe@outlook.com>
2025-07-17Perf improvements to IR serialization (#7751)Ellie Hermaszewska
* option to use riff as serialization backend * option to use riff as serialization backend * perf * shuffle code * perf improvements to deserialization * formatting * remove bit_cast * correct IR verification * neaten serialized format * fix peek module info * formatting * remove temporary profiling code * cleanup * fix wasm build * more explicit sizes * deserialize via fossil on 32 bit wasm * Make serialized modules Int size agnostic * reorder stable names to allow range based check for 64 bit constants * format * review comments * fix build * fix * c++17 compat slang-common.h
2025-07-17Improve lookup performance. (#7798)Yong He
* Improve lookup performance. * Cleanup. * Improve autocompletion latency.
2025-07-17Fix duplicate mangled names for interface requirements (#7764)Yong He
* Fix duplicate mangled names for interface requirements Remove linkage decorations from interface method requirement values to prevent duplicate mangled names and allow DCE to clean up unused functions. Interface requirements only need the type information, not the linkage, so the generated IRFunc with export decorations was causing conflicts. Fixes #7761 Co-authored-by: Yong He <csyonghe@users.noreply.github.com> * Move linkage decoration removal into default case of switch Move the removeLinkageDecorations call into the default case of the switch statement so it only applies to general requirements and not special cases like associated types. Co-authored-by: Yong He <csyonghe@users.noreply.github.com> * Remove conditional check for linkage decoration removal Apply removeLinkageDecorations unconditionally to interface requirement values to ensure all requirement values have their linkage decorations removed. Co-authored-by: Yong He <csyonghe@users.noreply.github.com> * Change removeLinkageDecorations parameter type from IRGlobalValueWithCode* to IRInst* This fixes build errors by allowing the function to accept the broader IRInst* type, making it compatible with the unconditional call in the interface requirement processing. Co-authored-by: Yong He <csyonghe@users.noreply.github.com> * format code (#7787) Co-authored-by: slangbot <186143334+slangbot@users.noreply.github.com> --------- Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com> Co-authored-by: Yong He <csyonghe@users.noreply.github.com> Co-authored-by: slangbot <ellieh+slangbot@nvidia.com> Co-authored-by: slangbot <186143334+slangbot@users.noreply.github.com>
2025-07-17Document IR changes (#7781)Ellie Hermaszewska
* Document IR changes * Remove duplication
2025-07-16Fix duplicate DiffPair struct generation for row_major matrices in autodiff ↵Copilot
(#7728) * Initial plan * Fix duplicate DiffPair struct generation for row_major matrices in autodiff Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> * Fix matrix layout conversion to use BuiltinCastExpr Address root cause in slang-check-conversion.cpp by creating proper cast expressions for matrix layout conversions instead of reusing expressions. This ensures autodiff sees proper type conversions and generates consistent DiffPair structs. Reverted the band-aid fix in autodiff system and implemented the proper front-end fix as suggested in code review. Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> * Fix test to prevent dead code elimination and make it executable on CPU Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> * Fix spirv emit of matrix layout cast insts. * Update test. * cleanup test. * Improve test with meaningful values that verify correct gradient computation Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> --------- Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com> Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> Co-authored-by: Yong He <yonghe@outlook.com>
2025-07-16Strip debug info only for the SpvOpExtInst of debug import type (#7779)Gangzheng Tong
2025-07-16Fix CUDA issues with texture reads and surface writes (#7780)Mukund Keshava
* Fix 1D texture reads in CUDA target Fixes #7570: 1D surface writes don't work The issue was that the Load function for read-only textures (hlsl.meta.slang lines 3629-3656) only supported 2D and 3D textures for CUDA targets, causing 1D texture reads to fall through to <invalid intrinsic>. This affected the srcTexture[tid.x] read operation in the reproduction case. Changes: - Updated static_assert to include SLANG_TEXTURE_1D support - Added tex1DArrayfetch_int<T> for 1D array texture reads - Added tex1Dfetch_int<T> for regular 1D texture reads 🤖 Generated with [Claude Code](https://claude.ai/code) Co-Authored-By: Mukund Keshava <mkeshavaNV@users.noreply.github.com> * Add 1D texture read support for CUDA target - Add tex1Dfetch_int template specializations for float2, float4, uint, uint2, uint4 - Remove TODO comment about 1D PTX not being supported - Enable 1D texture test in texture-subscript-cuda.slang - Fix assembly code issues in original template specializations 🤖 Generated with [Claude Code](https://claude.ai/code) Co-Authored-By: Mukund Keshava <mkeshavaNV@users.noreply.github.com> * Update slang-cuda-prelude.h * Fix texture3d ptx issue * undo 1D texture changes * Update hlsl.meta.slang * Update hlsl.meta.slang * Update hlsl.meta.slang * Update hlsl.meta.slang * Extend texture-subscript-cuda.slang test with uint and int format variants Add test cases for newly supported texture formats in CUDA: - 2D textures with uint, uint2, uint4 - 2D textures with int, int2, int4 - 3D textures with uint, uint2, uint4 - 3D textures with int, int2, int4 This ensures the texture subscript operations work correctly for all the format variants added in the CUDA texture fixes. Co-authored-by: Mukund Keshava <mkeshavaNV@users.noreply.github.com> * update expected file --------- Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com> Co-authored-by: Mukund Keshava <mkeshavaNV@users.noreply.github.com>
2025-07-16Diagnose on initExpr of a global `const` decl (#7711)Yong He
* Add diagnostic for const/uniform global variables with initializers Global const and uniform variables without the static modifier are treated as uniform parameters in HLSL, which cannot have default values. This change adds error 31224 to catch this case early and provide clear guidance to users. Fixes #7701 🤖 Generated with [Claude Code](https://claude.ai/code) Co-authored-by: Yong He <csyonghe@users.noreply.github.com> * Add test case for global const/uniform variable diagnostic Add test for diagnostic 31224 which catches global const and uniform variables with initializers that aren't declared static. The test verifies both error cases and valid cases to ensure the diagnostic works correctly. 🤖 Generated with [Claude Code](https://claude.ai/code) Co-authored-by: Yong He <csyonghe@users.noreply.github.com> * Fix failing tests due to new global const diagnostic Add 'static' keyword to global const variables with initializers to resolve diagnostic 31224: "global const variable with initializer must be declared static" Co-authored-by: Jay Kwak <jkwak-work@users.noreply.github.com> * Allow specialization constants with initializers without static - Modified diagnostic logic in slang-check-decl.cpp to allow specialization constants (SpecializationConstantAttribute and VkConstantIdAttribute) to have initializers without requiring the static keyword - Updated 8 SPIRV test files to remove unnecessary static keywords from specialization constant declarations - Enhanced diagnostic test case to include specialization constant examples - All tests passing with the new behavior 🤖 Generated with [Claude Code](https://claude.ai/code) Co-authored-by: Yong He <csyonghe@users.noreply.github.com> * format code (#7765) Co-authored-by: slangbot <186143334+slangbot@users.noreply.github.com> * Remove static keyword from specialization constants and add comprehensive test cases - Remove static from specialization constants in test files as they are semantically meaningless - Update tests to use existing diagnostic 31219 for static specialization constant errors - Add comprehensive test cases that verify the error occurs when static is used with specialization constants - Fixes suggested in PR review to make static specialization constants an error Co-authored-by: ArielG-NV <ArielG-NV@users.noreply.github.com> * Add error for static specialization constants Static specialization constants are semantically meaningless and should produce an error. This change adds logic to trigger diagnostic 31219 when both 'static' and specialization constant attributes are present. The existing diagnostic 31219 'push or specialization constants cannot be static' is now correctly triggered for: - [SpecializationConstant] static const variables - [vk::constant_id] static const variables Co-authored-by: ArielG-NV <ArielG-NV@users.noreply.github.com> * Update source/slang/slang-check-decl.cpp * Update tests/glsl/compute-shader-layout-id.slang * Update tests/glsl/compute-shader-layout-id.slang * Fix test case Claude broke * Update global-uniform.slang * Update global-uniform.slang.expected --------- Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com> Co-authored-by: Yong He <csyonghe@users.noreply.github.com> Co-authored-by: ArielG-NV <159081215+ArielG-NV@users.noreply.github.com> Co-authored-by: Jay Kwak <jkwak-work@users.noreply.github.com> Co-authored-by: slangbot <ellieh+slangbot@nvidia.com> Co-authored-by: slangbot <186143334+slangbot@users.noreply.github.com> Co-authored-by: ArielG-NV <ArielG-NV@users.noreply.github.com>
2025-07-16Correct help text for `-dump-intermediate-prefix` (#7768)Sam Estep
2025-07-16Document why failing VVL coop-vec tests cannot be enabled & minor ↵ArielG-NV
cleanup/bug-fix (#7754) * Changes Fix a coop-vec bug, fix incorrect test synax, clean up test, comment why some coop-vec tests are stil disabled * Changes Fix a coop-vec bug, fix incorrect test synax, clean up test, comment why some coop-vec tests are stil disabled * disable failing tests * push changes * add to failing GLSL test list
2025-07-16Fix Metal pointer type emission in entry point parameters (#7759)Jay Kwak
* Fix Metal pointer type emission in entry point parameters Add missing default case in Metal emitter's address space switch to preserve pointer types. Previously, unrecognized address spaces would fall through without emitting pointer syntax, causing uint64_t* to become ulong instead of ulong constant*. Fixes #7605, #6174 * Treat AddressSpace::UserPointer as Global in Metal Also adding another test for `uniform` keyword
2025-07-16Fix GLSL memory layout qualifiers not applied to uniform buffers or cbuffers ↵pdeayton-nv
(#7740) * Fix GLSL memory layout qualifiers not applied to uniform buffers or cbuffers The `layout(scalar)` qualifier was being ignored for GLSL `uniform` blocks and HLSL `cbuffer` declarations, causing them to use std140 layout instead of the requested scalar layout. **Root Cause**: The parsing logic in `slang-parser.cpp` handled layout qualifiers inconsistently: - GLSL `buffer` blocks correctly used `getLayoutArg()` to map layout modifiers - GLSL `uniform` blocks and HLSL `cbuffer` skipped layout arguments entirely **Solution**: - Enhanced `parseHLSLCBufferDecl()` to check for GLSL layout qualifiers - Added same `getLayoutArg()` logic used by buffer blocks - Modified uniform block parsing to pass layout arguments through **Testing**: - Added comprehensive test case: `tests/glsl/layout-scalar-qualifier.slang` - Verified fix works for both `uniform` blocks and `cbuffer` declarations - Confirmed no regressions in existing test suite **Before**: `layout(scalar) uniform {...}` → std140 layout (32 bytes, offset 0,16) **After**: `layout(scalar) uniform {...}` → scalar layout (16 bytes, offset 0,4) Fixes #7735 🤖 Generated with [Claude Code](https://claude.ai/code) Co-authored-by: pdeayton-nv <pdeayton-nv@users.noreply.github.com> * Apply code formatting to slang-parser.cpp Fixed line wrapping, trailing whitespace, and parameter formatting according to repository style guidelines. Co-authored-by: Harsh Aggarwal (NVIDIA) <szihs@users.noreply.github.com> * format code (#7742) Co-authored-by: slangbot <186143334+slangbot@users.noreply.github.com> --------- Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com> Co-authored-by: pdeayton-nv <pdeayton-nv@users.noreply.github.com> Co-authored-by: Harsh Aggarwal (NVIDIA) <szihs@users.noreply.github.com> Co-authored-by: slangbot <ellieh+slangbot@nvidia.com> Co-authored-by: slangbot <186143334+slangbot@users.noreply.github.com> Co-authored-by: Yong He <yonghe@outlook.com>
2025-07-16Fix WGSL sign function to return correct int type instead of float (#7739)Copilot
* Initial plan * Fix WGSL sign function to return int type properly Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> * Add vector version check for WGSL sign function test Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> * Fix FileCheck order in WGSL sign test - use DAG for order-independent checking Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com> --------- Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com> Co-authored-by: csyonghe <2652293+csyonghe@users.noreply.github.com>
generated by cgit v1.2.3 (git 2.39.1) at 2026-06-13 16:01:52 +0000