Build issues with latest Triton LLVM

[coleramos425]

When attempting to build from main with the latest version of Triton LLVM on HPCFund I encountered the following error. Looks related to the new function signature of registerPipelineEarlySimplificationEPCallback() (llvm/llvm-project@dc45ff1).

Should be fixed by #6.

(instrument) [colramos@login1 build]$ cmake -DCMAKE_C_COMPILER=/opt/rocm-6.2.1/bin/amdclang -DCMAKE_CXX_COMPILER=/opt/rocm-6.2.1/bin/amdclang++ -DLLVM_INSTALL_DIR=$HOME/.triton/llvm/llvm-86b69c31-ubuntu-x64 ..
-- The C compiler identification is Clang 18.0.0
-- The CXX compiler identification is Clang 18.0.0
-- Detecting C compiler ABI info
-- Detecting C compiler ABI info - done
-- Check for working C compiler: /opt/rocm-6.2.1/bin/amdclang - skipped
-- Detecting C compile features
-- Detecting C compile features - done
-- Detecting CXX compiler ABI info
-- Detecting CXX compiler ABI info - done
-- Check for working CXX compiler: /opt/rocm-6.2.1/bin/amdclang++ - skipped
-- Detecting CXX compile features
-- Detecting CXX compile features - done
-- Found ZLIB: /usr/lib64/libz.so (found version "1.2.11") 
-- Found LibXml2: /usr/lib64/libxml2.so (found version "2.9.13") 
-- Found LLVM 20.0.0git
-- Using LLVMConfig.cmake in: /home1/colramos/.triton/llvm/llvm-86b69c31-ubuntu-x64
LLVM STATUS:
  Definitions -D_GNU_SOURCE -D_DEBUG -D_GLIBCXX_ASSERTIONS -D__STDC_CONSTANT_MACROS -D__STDC_FORMAT_MACROS -D__STDC_LIMIT_MACROS
  Includes    /home1/colramos/.triton/llvm/llvm-86b69c31-ubuntu-x64/include
  Libraries   /home1/colramos/.triton/llvm/llvm-86b69c31-ubuntu-x64/lib
  Targets     NVPTX;AMDGPU;X86
-- Performing Test SUPPORTS_FVISIBILITY_INLINES_HIDDEN_FLAG
-- Performing Test SUPPORTS_FVISIBILITY_INLINES_HIDDEN_FLAG - Success
-- Configuring done
-- Generating done
-- Build files have been written to: /work1/amd/colramos/audacious/instrument-amdgpu-kernels/build
(instrument) [colramos@login1 build]$ make
[ 10%] Building CXX object lib/CMakeFiles/InjectAMDGCNFunction.dir/InjectAMDGCNFunction.cpp.o
/work1/amd/colramos/audacious/instrument-amdgpu-kernels/lib/InjectAMDGCNFunction.cpp:28:55: warning: 'getDeclaration' is deprecated: Use getOrInsertDeclaration instead [-Wdeprecated-declarations]
   28 |       Function *WorkItemXIDIntrinsicFunc = Intrinsic::getDeclaration(
      |                                                       ^~~~~~~~~~~~~~
      |                                                       getOrInsertDeclaration
/home1/colramos/.triton/llvm/llvm-86b69c31-ubuntu-x64/include/llvm/IR/Intrinsics.h:100:3: note: 'getDeclaration' has been explicitly marked deprecated here
  100 |   LLVM_DEPRECATED("Use getOrInsertDeclaration instead",
      |   ^
/home1/colramos/.triton/llvm/llvm-86b69c31-ubuntu-x64/include/llvm/Support/Compiler.h:234:50: note: expanded from macro 'LLVM_DEPRECATED'
  234 | #define LLVM_DEPRECATED(MSG, FIX) __attribute__((deprecated(MSG, FIX)))
      |                                                  ^
/work1/amd/colramos/audacious/instrument-amdgpu-kernels/lib/InjectAMDGCNFunction.cpp:46:9: error: no viable conversion from '(lambda at /work1/amd/colramos/audacious/instrument-amdgpu-kernels/lib/InjectAMDGCNFunction.cpp:46:9)' to 'const std::function<void (ModulePassManager &, OptimizationLevel, ThinOrFullLTOPhase)>' (aka 'const function<void (PassManager<Module> &, llvm::OptimizationLevel, llvm::ThinOrFullLTOPhase)>')
   46 |         [&](ModulePassManager &MPM, auto) {
      |         ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
   47 |           MPM.addPass(InjectAMDGCNFunc());
      |           ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
   48 |           return true;
      |           ~~~~~~~~~~~~
   49 |         });
      |         ~
/usr/lib/gcc/x86_64-redhat-linux/11/../../../../include/c++/11/bits/std_function.h:375:7: note: candidate constructor not viable: no known conversion from '(lambda at /work1/amd/colramos/audacious/instrument-amdgpu-kernels/lib/InjectAMDGCNFunction.cpp:46:9)' to 'nullptr_t' (aka 'std::nullptr_t') for 1st argument
  375 |       function(nullptr_t) noexcept
      |       ^        ~~~~~~~~~
/usr/lib/gcc/x86_64-redhat-linux/11/../../../../include/c++/11/bits/std_function.h:386:7: note: candidate constructor not viable: no known conversion from '(lambda at /work1/amd/colramos/audacious/instrument-amdgpu-kernels/lib/InjectAMDGCNFunction.cpp:46:9)' to 'const function<void (PassManager<Module> &, OptimizationLevel, ThinOrFullLTOPhase)> &' for 1st argument
  386 |       function(const function& __x)
      |       ^        ~~~~~~~~~~~~~~~~~~~
/usr/lib/gcc/x86_64-redhat-linux/11/../../../../include/c++/11/bits/std_function.h:404:7: note: candidate constructor not viable: no known conversion from '(lambda at /work1/amd/colramos/audacious/instrument-amdgpu-kernels/lib/InjectAMDGCNFunction.cpp:46:9)' to 'function<void (PassManager<Module> &, OptimizationLevel, ThinOrFullLTOPhase)> &&' for 1st argument
  404 |       function(function&& __x) noexcept
      |       ^        ~~~~~~~~~~~~~~
/usr/lib/gcc/x86_64-redhat-linux/11/../../../../include/c++/11/bits/std_function.h:435:2: note: candidate template ignored: requirement '_Callable<(lambda at /work1/amd/colramos/audacious/instrument-amdgpu-kernels/lib/InjectAMDGCNFunction.cpp:46:9), (lambda at /work1/amd/colramos/audacious/instrument-amdgpu-kernels/lib/InjectAMDGCNFunction.cpp:46:9), std::__invoke_result<(lambda at /work1/amd/colramos/audacious/instrument-amdgpu-kernels/lib/InjectAMDGCNFunction.cpp:46:9) &, llvm::PassManager<llvm::Module, llvm::AnalysisManager<llvm::Module>> &, llvm::OptimizationLevel, llvm::ThinOrFullLTOPhase>>::value' was not satisfied [with _Functor = (lambda at /work1/amd/colramos/audacious/instrument-amdgpu-kernels/lib/InjectAMDGCNFunction.cpp:46:9)]
  435 |         function(_Functor&& __f)
      |         ^
/home1/colramos/.triton/llvm/llvm-86b69c31-ubuntu-x64/include/llvm/Passes/PassBuilder.h:484:54: note: passing argument to parameter 'C' here
  484 |                                ThinOrFullLTOPhase)> &C) {
      |                                                      ^
1 warning and 1 error generated.
make[2]: *** [lib/CMakeFiles/InjectAMDGCNFunction.dir/build.make:76: lib/CMakeFiles/InjectAMDGCNFunction.dir/InjectAMDGCNFunction.cpp.o] Error 1
make[1]: *** [CMakeFiles/Makefile2:295: lib/CMakeFiles/InjectAMDGCNFunction.dir/all] Error 2
make: *** [Makefile:91: all] Error 2

• CMake version: 3.25.2
• Clang version: 18.0.0
• ROCm version: 6.2.1
• Triton LLVM (commit id): 86b69c31