[flang] compiler crashes in OptimizedBufferizationPass when dot_product is used inside parallel region

[shivaramaarao]

The issue is observed when compiling cp2k.

A reproducer is here:
 SUBROUTINE mysub()
      INTEGER  :: ns,  ne
      REAL(KIND=8), ALLOCATABLE, DIMENSION(:, :, :)     :: ar1, ar2
      REAL(KIND=8), DIMENSION(20)          :: d

!$OMP PARALLEL PRIVATE(n,ns,ne,ar1,ar2)
         d(ns:ne) = &
            (/(DOT_PRODUCT(ar1(:, n, n), ar2(:, n, n)),n=ns, ne)/)
!$OMP END PARALLEL
   END SUBROUTINE

Invocation and crash message:

$flang -O1 -fopenmp tm1.f90
flang-20: warning: OpenMP support in flang is still experimental [-Wexperimental-option]
flang: /home/shivaram/llvm_trunk/llvm-project/llvm/include/llvm/ADT/STLExtras.h:1289: ReferenceT llvm::detail::indexed_accessor_range_base<DerivedT, BaseT, T, PointerT, ReferenceT>::operator[](size_t) const [with DerivedT = mlir::OperandRange; BaseT = mlir::OpOperand*; T = mlir::Value; PointerT = mlir::Value; ReferenceT = mlir::Value; size_t = long unsigned int]: Assertion `Index < size() && "invalid index for value range"' failed.

at -O0 the testcase compiles. at O1 and above, the crash is observed.

[EugeneZelenko]

@shivaramaarao: Please provide full stack.

[shivaramaarao]

Sure: Full stack is here.

flang: /home/shivaram/llvm_trunk/llvm-project/llvm/include/llvm/ADT/STLExtras.h:1289: ReferenceT llvm::detail::indexed_accessor_range_base<DerivedT, BaseT, T, PointerT, ReferenceT>::operator[](size_t) const [with DerivedT = mlir::OperandRange; BaseT = mlir::OpOperand*; T = mlir::Value; PointerT = mlir::Value; ReferenceT = mlir::Value; size_t = long unsigned int]: Assertion `Index < size() && "invalid index for value range"' failed.
PLEASE submit a bug report to https://github.com/llvm/llvm-project/issues/ and include the crash backtrace.
 #0 0x00005615339093d0 llvm::sys::PrintStackTrace(llvm::raw_ostream&, int) (/home/shivaram/llvm_mir_flang/bin/flang+0x14953d0)
 #1 0x00005615339067df llvm::sys::RunSignalHandlers() (/home/shivaram/llvm_mir_flang/bin/flang+0x14927df)
 #2 0x0000561533906935 SignalHandler(int) Signals.cpp:0:0
 #3 0x00007f039da28520 (/lib/x86_64-linux-gnu/libc.so.6+0x42520)
 #4 0x00007f039da7c9fc __pthread_kill_implementation ./nptl/pthread_kill.c:44:76
 #5 0x00007f039da7c9fc __pthread_kill_internal ./nptl/pthread_kill.c:78:10
 #6 0x00007f039da7c9fc pthread_kill ./nptl/pthread_kill.c:89:10
 #7 0x00007f039da28476 gsignal ./signal/../sysdeps/posix/raise.c:27:6
 #8 0x00007f039da0e7f3 abort ./stdlib/abort.c:81:7
 #9 0x00007f039da0e71b _nl_load_domain ./intl/loadmsgcat.c:1177:9
#10 0x00007f039da1fe96 (/lib/x86_64-linux-gnu/libc.so.6+0x39e96)
#11 0x000056153571a295 _ZN4llvm12function_refIFvPN4mlir9OperationEEE11callback_fnIZNS1_6detail4walkILNS1_9WalkOrderE1ENS1_15ForwardIteratorEZN3firL13getPrivateArgINS1_3omp10ParallelOpEN5hlfir9DeclareOpEEENS1_5ValueERNSD_25BlockArgOpenMPOpInterfaceERT_RT0_EUlNSD_7YieldOpEE_SO_vEENSt9enable_ifIXaantsrSt11disjunctionIJSt7is_sameIT2_S3_ESS_IST_PNS1_6RegionEESS_IST_PNS1_5BlockEEEE5valuesrSS_IT3_vE5valueES12_E4typeES3_OT1_EUlS3_E_EEvlS3_ AliasAnalysis.cpp:0:0
#12 0x000056153571af43 void mlir::detail::walk<mlir::ForwardIterator>(mlir::Operation*, llvm::function_ref<void (mlir::Operation*)>, mlir::WalkOrder) (/home/shivaram/llvm_mir_flang/bin/flang+0x32a6f43)
#13 0x000056153571baca auto fir::AliasAnalysis::getSource(mlir::Value, bool)::'lambda7'(auto)::operator()<hlfir::DeclareOp>(auto) const AliasAnalysis.cpp:0:0
#14 0x000056153571eb74 fir::AliasAnalysis::getSource(mlir::Value, bool) (/home/shivaram/llvm_mir_flang/bin/flang+0x32aab74)
#15 0x000056153571f6e8 fir::AliasAnalysis::alias(mlir::Value, mlir::Value) (/home/shivaram/llvm_mir_flang/bin/flang+0x32ab6e8)
#16 0x00005615355801d4 (anonymous namespace)::containsReadOrWriteEffectOn(mlir::SideEffects::EffectInstance<mlir::MemoryEffects::Effect> const&, mlir::Value) OptimizedBufferization.cpp:0:0
#17 0x0000561535583326 (anonymous namespace)::ElementalAssignBufferization::matchAndRewrite(hlfir::ElementalOp, mlir::PatternRewriter&) const OptimizedBufferization.cpp:0:0
#18 0x00005615366689b8 mlir::PatternApplicator::matchAndRewrite(mlir::Operation*, mlir::PatternRewriter&, llvm::function_ref<bool (mlir::Pattern const&)>, llvm::function_ref<void (mlir::Pattern const&)>, llvm::function_ref<llvm::LogicalResult (mlir::Pattern const&)>) (/home/shivaram/llvm_mir_flang/bin/flang+0x41f49b8)
#19 0x0000561536633a6e (anonymous namespace)::GreedyPatternRewriteDriver::processWorklist() GreedyPatternRewriteDriver.cpp:0:0
#20 0x0000561536636255 mlir::applyPatternsAndFoldGreedily(mlir::Region&, mlir::FrozenRewritePatternSet const&, mlir::GreedyRewriteConfig, bool*) (/home/shivaram/llvm_mir_flang/bin/flang+0x41c2255)
#21 0x000056153557f65e (anonymous namespace)::OptimizedBufferizationPass::runOnOperation() OptimizedBufferization.cpp:0:0
#22 0x0000561538c2e5d9 mlir::detail::OpToOpPassAdaptor::run(mlir::Pass*, mlir::Operation*, mlir::AnalysisManager, bool, unsigned int) (/home/shivaram/llvm_mir_flang/bin/flang+0x67ba5d9)
#23 0x0000561538c2eac1 mlir::detail::OpToOpPassAdaptor::runPipeline(mlir::OpPassManager&, mlir::Operation*, mlir::AnalysisManager, bool, unsigned int, mlir::PassInstrumentor*, mlir::PassInstrumentation::PipelineParentInfo const*) (/home/shivaram/llvm_mir_flang/bin/flang+0x67baac1)
#24 0x0000561538c2ee66 mlir::detail::OpToOpPassAdaptor::runOnOperationAsyncImpl(bool)::'lambda'(mlir::detail::OpToOpPassAdaptor::runOnOperationAsyncImpl(bool)::OpPMInfo&)::operator()(mlir::detail::OpToOpPassAdaptor::runOnOperationAsyncImpl(bool)::OpPMInfo&) const Pass.cpp:0:0
#25 0x0000561538c2f04f std::_Function_handler<void (), llvm::LogicalResult mlir::failableParallelForEach<__gnu_cxx::__normal_iterator<mlir::detail::OpToOpPassAdaptor::runOnOperationAsyncImpl(bool)::OpPMInfo*, std::vector<mlir::detail::OpToOpPassAdaptor::runOnOperationAsyncImpl(bool)::OpPMInfo, std::allocator<mlir::detail::OpToOpPassAdaptor::runOnOperationAsyncImpl(bool)::OpPMInfo>>>, void mlir::parallelForEach<__gnu_cxx::__normal_iterator<mlir::detail::OpToOpPassAdaptor::runOnOperationAsyncImpl(bool)::OpPMInfo*, std::vector<mlir::detail::OpToOpPassAdaptor::runOnOperationAsyncImpl(bool)::OpPMInfo, std::allocator<mlir::detail::OpToOpPassAdaptor::runOnOperationAsyncImpl(bool)::OpPMInfo>>>, mlir::detail::OpToOpPassAdaptor::runOnOperationAsyncImpl(bool)::'lambda'(mlir::detail::OpToOpPassAdaptor::runOnOperationAsyncImpl(bool)::OpPMInfo&)>(mlir::MLIRContext*, __gnu_cxx::__normal_iterator<mlir::detail::OpToOpPassAdaptor::runOnOperationAsyncImpl(bool)::OpPMInfo*, std::vector<mlir::detail::OpToOpPassAdaptor::runOnOperationAsyncImpl(bool)::OpPMInfo, std::allocator<mlir::detail::OpToOpPassAdaptor::runOnOperationAsyncImpl(bool)::OpPMInfo>>>, __gnu_cxx::__normal_iterator<mlir::detail::OpToOpPassAdaptor::runOnOperationAsyncImpl(bool)::OpPMInfo*, std::vector<mlir::detail::OpToOpPassAdaptor::runOnOperationAsyncImpl(bool)::OpPMInfo, std::allocator<mlir::detail::OpToOpPassAdaptor::runOnOperationAsyncImpl(bool)::OpPMInfo>>>, mlir::detail::OpToOpPassAdaptor::runOnOperationAsyncImpl(bool)::'lambda'(mlir::detail::OpToOpPassAdaptor::runOnOperationAsyncImpl(bool)::OpPMInfo&)&&)::'lambda'(__gnu_cxx::__normal_iterator<mlir::detail::OpToOpPassAdaptor::runOnOperationAsyncImpl(bool)::OpPMInfo*, std::vector<mlir::detail::OpToOpPassAdaptor::runOnOperationAsyncImpl(bool)::OpPMInfo, std::allocator<mlir::detail::OpToOpPassAdaptor::runOnOperationAsyncImpl(bool)::OpPMInfo>>>&&)>(mlir::MLIRContext*, __gnu_cxx::__normal_iterator<mlir::detail::OpToOpPassAdaptor::runOnOperationAsyncImpl(bool)::OpPMInfo*, std::vector<mlir::detail::OpToOpPassAdaptor::runOnOperationAsyncImpl(bool)::OpPMInfo, std::allocator<mlir::detail::OpToOpPassAdaptor::runOnOperationAsyncImpl(bool)::OpPMInfo>>>, __gnu_cxx::__normal_iterator<mlir::detail::OpToOpPassAdaptor::runOnOperationAsyncImpl(bool)::OpPMInfo*, std::vector<mlir::detail::OpToOpPassAdaptor::runOnOperationAsyncImpl(bool)::OpPMInfo, std::allocator<mlir::detail::OpToOpPassAdaptor::runOnOperationAsyncImpl(bool)::OpPMInfo>>>, mlir::detail::OpToOpPassAdaptor::runOnOperationAsyncImpl(bool)::'lambda'(mlir::detail::OpToOpPassAdaptor::runOnOperationAsyncImpl(bool)::OpPMInfo&)&&)::'lambda'()>::_M_invoke(std::_Any_data const&) Pass.cpp:0:0
#26 0x0000561536639622 std::_Function_handler<std::unique_ptr<std::__future_base::_Result_base, std::__future_base::_Result_base::_Deleter> (), std::__future_base::_Task_setter<std::unique_ptr<std::__future_base::_Result<void>, std::__future_base::_Result_base::_Deleter>, std::thread::_Invoker<std::tuple<std::function<void ()>>>, void>>::_M_invoke(std::_Any_data const&) (/home/shivaram/llvm_mir_flang/bin/flang+0x41c5622)
#27 0x0000561536639d4d std::__future_base::_State_baseV2::_M_do_set(std::function<std::unique_ptr<std::__future_base::_Result_base, std::__future_base::_Result_base::_Deleter> ()>*, bool*) (/home/shivaram/llvm_mir_flang/bin/flang+0x41c5d4d)
#28 0x00007f039da7fee8 __pthread_once_slow ./nptl/pthread_once.c:118:7
#29 0x0000561536639887 std::__future_base::_Deferred_state<std::thread::_Invoker<std::tuple<std::function<void ()>>>, void>::_M_complete_async() (/home/shivaram/llvm_mir_flang/bin/flang+0x41c5887)
#30 0x000056153663a6a7 std::_Function_handler<void (), std::shared_future<void> llvm::ThreadPoolInterface::asyncImpl<void>(std::function<void ()>, llvm::ThreadPoolTaskGroup*)::'lambda'()>::_M_invoke(std::_Any_data const&) (/home/shivaram/llvm_mir_flang/bin/flang+0x41c66a7)
#31 0x0000561538f119f7 llvm::StdThreadPool::processTasks(llvm::ThreadPoolTaskGroup*) (/home/shivaram/llvm_mir_flang/bin/flang+0x6a9d9f7)
#32 0x0000561538f127be void* llvm::thread::ThreadProxy<std::tuple<llvm::StdThreadPool::grow(int)::'lambda'()>>(void*) ThreadPool.cpp:0:0
#33 0x00007f039da7aac3 start_thread ./nptl/pthread_create.c:442:8
#34 0x00007f039db0c850 ./misc/../sysdeps/unix/sysv/linux/x86_64/clone3.S:83:0
flang-20: error: unable to execute command: Aborted (core dumped)
flang-20: error: flang frontend command failed due to signal (use -v to see invocation)
flang version 20.0.0git (https://github.com/llvm/llvm-project.git cd418030de7ae75750bc4e48d1238baf03c675e5)
Target: x86_64-unknown-linux-gnu
Thread model: posix
InstalledDir: /home/shivaram/llvm_mir_flang/bin

[llvmbot]

@llvm/issue-subscribers-flang-ir

Author: None (shivaramaarao)

The issue is observed when compiling cp2k.

A reproducer is here:
SUBROUTINE mysub()
INTEGER :: ns, ne
REAL(KIND=8), ALLOCATABLE, DIMENSION(:, :, :) :: ar1, ar2
REAL(KIND=8), DIMENSION(20) :: d

!$OMP PARALLEL PRIVATE(n,ns,ne,ar1,ar2)
d(ns:ne) = &
(/(DOT_PRODUCT(ar1(:, n, n), ar2(:, n, n)),n=ns, ne)/)
!$OMP END PARALLEL
END SUBROUTINE

$flang -O1 -fopenmp tm1.f90
flang-20: warning: OpenMP support in flang is still experimental [-Wexperimental-option]
flang: /home/shivaram/llvm_trunk/llvm-project/llvm/include/llvm/ADT/STLExtras.h:1289: ReferenceT llvm::detail::indexed_accessor_range_base<DerivedT, BaseT, T, PointerT, ReferenceT>::operator const [with DerivedT = mlir::OperandRange; BaseT = mlir::OpOperand*; T = mlir::Value; PointerT = mlir::Value; ReferenceT = mlir::Value; size_t = long unsigned int]: Assertion `Index < size() && "invalid index for value range"' failed.

at -O0 the testcase compiles. at O1 and above, the crash is observed.

[kiranchandramohan]

A marginally reduced test is the following.

 SUBROUTINE mysub(ns,ne)
      INTEGER  :: n
      REAL(KIND=8), DIMENSION(:), allocatable     :: ar1
      real(kind=8), dimension(20) :: ar2
      REAL(KIND=8), DIMENSION(20) :: d

!$OMP parallel PRIVATE(ar1)
         d(1:1) = (/(DOT_PRODUCT(ar1(1:n), ar2(1:n)),n=1, 1)/)
!$OMP END parallel
   END SUBROUTINE

@DominikAdamski this seems to be failing in fir::getPrivateArg

llvm-project/flang/lib/Optimizer/Analysis/AliasAnalysis.cpp

Line 474 in a4c3683

static Value getPrivateArg(omp::BlockArgOpenMPOpInterface &argIface,

[DominikAdamski]

@shivaramaarao This PR should fix the issue: #120243 .