[err]Compiling cmsis dsp failed when using -O3.

[bshaokang]

i found that the problem was caused by the compilation option -o3.
Compiling cmsis dsp failed when using -O3.
Compiling cmsis dsp succeeded when not using -O3.
Is this a bug?

bsk@mcshyucs0008:~/clang_test$ clang-18 -c -Wstrict-prototypes -Wno-attributes -Wno-unknown-pragmas -Oz \
-fno-strict-aliasing -fomit-frame-pointer -fno-builtin -fshort-enums -fno-common -Wall -Wshadow -Wundef \ 
-ffunction-sections -fdata-sections -target armv8.1m.main_hard_fp.cfg -march=armv8.1-m.main+mve \ 
-mtune=cortex-m85 -mfpu=fpv5-d16 -mfloat-abi=hard -mthumb -Wa,-mthumb -Wa,-mimplicit-it=always \ 
--target=arm-none-eabi  -isystem ~/clang_test/mcu_common/nuttx/include -isystem \
~/clang_test/mcu_common/nuttx/include/newlib -D__NuttX__ -DNDEBUG -DXR_VERSION_FPGA \
-pipe -Werror -g -I~/clang_test/contexthub/../platform/tuling3 \
-I~/clang_test/contexthub/../platform/tuling3/fpgdafen/contexthub/../platform/common/pmu/include/tuling3/fpga \
-I~/clang_test/contexthub/../platform/common/pmu/include/tuling3 \
-I~/clang_test/contexthub/../platform/common/clk/include/tuling3 \
-I~/clang_test/contexthub/chips/mcu/cortex-m85/include \
-I~/clang_test/mcu_common/nuttx/../vendor/contexthub/sensor_drivers/include \
-I~/clang_test/contexthub/cmsis/CMSIS/CMSIS_5-5.9.0/Device/ARM/ARMCM85/Include \
-I~/clang_test/contexthub/cmsis/CMSIS/CMSIS_5-5.9.0/CMSIS/Core/Include \
-I~/clang_test/contexthub/cmsis/CMSIS/CMSIS-DSP-1.15.0/Include \
-I~/clang_test/contexthub/cmsis/CMSIS/CMSIS-NN-4.1.0/Include \
-I~/clang_test/contexthub/securec/include -DSECUREC_ENABLE_FSCANF=0 \
-DSECUREC_ENABLE_VFSCANF=0 -DSECUREC_ENABLE_WCHAR_FUNC=0 \
-DSECUREC_HAVE_WCHART=0 -DSECUREC_ENABLE_SPRINTF_FLOAT=0 \
-DSECUREC_HAVE_WCTOMB=0 -DSECUREC_IN_KERNEL=0 \
-I~/clang_test/contexthub/chips/include -DTODO_REG=0 \
-I ~/clang_test/mcu_common/apps/vendor/contexthub/cmsis/CMSIS/CMSIS-DSP-1.15.0/Include \
-I ~/clang_test/mcu_common/apps/vendor/contexthub/cmsis/CMSIS/CMSIS_5-5.9.0/CMSIS/Core/Include/ \
-I ~/clang_test/mcu_common/apps/include \
-I dafen/open-source/cmsis/CMSIS/CMSIS-DSP-1.15.0/PrivateInclude \
-I dafen/open-source/cmsis/CMSIS_5-5.9.0/CMSIS/DSP/Include/ \
-I dafen/open-source/cmsis/CMSIS_5-5.9.0/CMSIS/Core/Include/ \
-Wno-shadow -flax-vector-conversions \
**-O3** \
dafen/open-source/cmsis/CMSIS-DSP-1.15.0/Source/SupportFunctions/arm_float_to_f16.c -o bskarm_float_to_f16.o \
**-v**
clang version 18.1.3
Target: arm-none-unknown-eabi
Thread model: posix
InstalledDir: ~/clang_test/open-source/clang_for_arm/clang18/bin
 (in-process)
 "~/clang_test/open-source/clang_for_arm/clang18/bin/clang-18" -cc1 \
-triple thumbv8.1m.main-none-unknown-eabihf -emit-obj -mllvm -arm-implicit-it=always \
-disable-free -clear-ast-before-backend -disable-llvm-verifier -discard-value-names \
-main-file-name arm_float_to_f16.c -mrelocation-model static -mframe-pointer=none -relaxed-aliasing \
-fmath-errno -ffp-contract=on -fno-rounding-math -mconstructor-aliases -nostdsysteminc \
-target-cpu generic -target-feature +vfp2 -target-feature +vfp2sp -target-feature -vfp3 -target-feature +vfp3d16 \
-target-feature +vfp3d16sp -target-feature -vfp3sp -target-feature +fp16 \
-target-feature -vfp4 -target-feature +vfp4d16 -target-feature +vfp4d16sp \
-target-feature -vfp4sp -target-feature -fp-armv8 -target-feature +fp-armv8d16 \
-target-feature +fp-armv8d16sp -target-feature -fp-armv8sp -target-feature -fullfp16 \
-target-feature +fp64 -target-feature -d32 -target-feature -neon -target-feature \
-sha2 -target-feature -aes -target-feature +dsp -target-feature +mve -target-feature \
-fp16fml -target-feature +strict-align -target-abi aapcs -mfloat-abi hard -Wunaligned-access \
-debug-info-kind=constructor -dwarf-version=5 -debugger-tuning=gdb \
-fdebug-compilation-dir=/home/bsk/kasan_test -v -ffunction-sections -fdata-sections \
-fcoverage-compilation-dir=/home/bsk/kasan_test -resource-dir \
~/clang_test/open-source/clang_for_arm/clang18/lib/clang/18 -isystem ~/clang_test/mcu_common/nuttx/include\
 -isystem ~/clang_test/mcu_common/nuttx/include/newlib -D __NuttX__ -D \
NDEBUG -D XR_VERSION_FPGA -I ~/clang_test/contexthub/../platform/tuling3 \
-I ~/clang_test/contexthub/../platform/tuling3/fpgdafen/contexthub/../platform/common/pmu/include/tuling3/fpga \
-I ~/clang_test/contexthub/../platform/common/pmu/include/tuling3 \
-I ~/clang_test/contexthub/../platform/common/clk/include/tuling3 \
-I ~/clang_test/contexthub/chips/mcu/cortex-m85/include \
-I ~/clang_test/mcu_common/nuttx/../vendor/contexthub/sensor_drivers/include \
-I ~/clang_test/contexthub/cmsis/CMSIS/CMSIS_5-5.9.0/Device/ARM/ARMCM85/Include \
-I ~/clang_test/contexthub/cmsis/CMSIS/CMSIS_5-5.9.0/CMSIS/Core/Include \
-I ~/clang_test/contexthub/cmsis/CMSIS/CMSIS-DSP-1.15.0/Include \
-I ~/clang_test/contexthub/cmsis/CMSIS/CMSIS-NN-4.1.0/Include \
-I ~/clang_test/contexthub/securec/include -D SECUREC_ENABLE_FSCANF=0 \
-D SECUREC_ENABLE_VFSCANF=0 -D SECUREC_ENABLE_WCHAR_FUNC=0 -D SECUREC_HAVE_WCHART=0 -D SECUREC_ENABLE_SPRINTF_FLOAT=0 -D SECUREC_HAVE_WCTOMB=0 -D SECUREC_IN_KERNEL=0 -I ~/clang_test/contexthub/chips/include -D TODO_REG=0 \
-I ~/clang_test/mcu_common/apps/vendor/contexthub/cmsis/CMSIS/CMSIS-DSP-1.15.0/Include \
-I ~/clang_test/mcu_common/apps/vendor/contexthub/cmsis/CMSIS/CMSIS_5-5.9.0/CMSIS/Core/Include/ \
-I ~/clang_test/mcu_common/apps/include -I dafen//open-source/cmsis/CMSIS/CMSIS-DSP-1.15.0/PrivateInclude \
-I dafen//open-source/cmsis/CMSIS_5-5.9.0/CMSIS/DSP/Include/ \
-I dafen//open-source/cmsis/CMSIS_5-5.9.0/CMSIS/Core/Include/ \
-internal-isystem ~/clang_test/open-source/clang_for_arm/clang18/lib/clang/18/include -internal-isystem \
~/clang_test/open-source/clang_for_arm/clang18/bin/../lib/clang-runtimes/arm-none-eabi/armv8m.main_hard_fp_exn_rtti/include \
**-O3** \
-Wstrict-prototypes -Wno-attributes -Wno-unknown-pragmas \
-Wall -Wshadow -Wundef -Werror -Wno-shadow -ferror-limit 19 -flax-vector-conversions=integer \
-fno-builtin -fshort-enums -fno-signed-char -fgnuc-version=4.2.1 -fskip-odr-check-in-gmf \
-fcolor-diagnostics -vectorize-loops -vectorize-slp -faddrsig -D__GCC_HAVE_DWARF2_CFI_ASM=1 \
-o bskarm_float_to_f16.o -x c dafen//open-source/cmsis/CMSIS-DSP-1.15.0/Source/SupportFunctions/arm_float_to_f16.c \
clang -cc1 version 18.1.3 based upon LLVM 18.1.3 default target aarch64-linux-gnu
ignoring nonexistent directory "~/clang_test/contexthub/../platform/tuling3/fpgdafen/contexthub/../platform/common/pmu/include/tuling3/fpga"
ignoring nonexistent directory "~/clang_test/mcu_common/nuttx/../vendor/contexthub/sensor_drivers/include"
ignoring nonexistent directory "~/clang_test/contexthub/cmsis/CMSIS/CMSIS_5-5.9.0/Device/ARM/ARMCM85/Include"
ignoring nonexistent directory "~/clang_test/contexthub/cmsis/CMSIS/CMSIS_5-5.9.0/CMSIS/Core/Include"
ignoring nonexistent directory "~/clang_test/contexthub/cmsis/CMSIS/CMSIS-DSP-1.15.0/Include"
ignoring nonexistent directory "~/clang_test/contexthub/cmsis/CMSIS/CMSIS-NN-4.1.0/Include"
ignoring nonexistent directory "~/clang_test/mcu_common/apps/vendor/contexthub/cmsis/CMSIS/CMSIS-DSP-1.15.0/Include"
ignoring nonexistent directory "~/clang_test/mcu_common/apps/vendor/contexthub/cmsis/CMSIS/CMSIS_5-5.9.0/CMSIS/Core/Include/"
ignoring nonexistent directory "open-source/cmsis/CMSIS/CMSIS-DSP-1.15.0/PrivateInclude"
ignoring nonexistent directory "~/clang_test/mcu_common/nuttx/include/newlib"
ignoring duplicate directory "~/clang_test/open-source/clang_for_arm/clang18/lib/clang/18/include"
#include "..." search starts here:
#include <...> search starts here:
 ~/clang_test/contexthub/../platform/tuling3
 ~/clang_test/contexthub/../platform/common/pmu/include/tuling3
 ~/clang_test/contexthub/../platform/common/clk/include/tuling3
 ~/clang_test/contexthub/chips/mcu/cortex-m85/include
 ~/clang_test/contexthub/securec/include
 ~/clang_test/contexthub/chips/include
 ~/clang_test/mcu_common/apps/include
 dafen//open-source/cmsis/CMSIS_5-5.9.0/CMSIS/DSP/Include
 dafen//open-source/cmsis/CMSIS_5-5.9.0/CMSIS/Core/Include
 ~/clang_test/mcu_common/nuttx/include
 ~/clang_test/open-source/clang_for_arm/clang18/lib/clang/18/include
 ~/clang_test/open-source/clang_for_arm/clang18/bin/../lib/clang-runtimes/arm-none-eabi/armv8m.main_hard_fp_exn_rtti/include
End of search list.
fatal error: error in backend: Cannot select: 0x8883b60: v8f16 = ARMISD::VCVTN undef:v8f16, 0x8883150, Constant:i32<0>, dafen//open-source/cmsis/CMSIS-DSP-1.15.0/Source/SupportFunctions/arm_float_to_f16.c:117:17
  0x88830e0: v8f16 = undef
  0x8883150: v4f32,i32,ch = load<(load (s128) from %ir.29, align 4, !alias.scope !37), <post-inc>> 0x83a3ec0, 0x8883850, Constant:i32<16>, dafen//open-source/cmsis/CMSIS-DSP-1.15.0/Source/SupportFunctions/arm_float_to_f16.c:117:31
    0x8883850: i32,ch = CopyFromReg 0x83a3ec0, Register:i32 %6, dafen//open-source/cmsis/CMSIS-DSP-1.15.0/Source/SupportFunctions/arm_float_to_f16.c:117:31
      0x8883770: i32 = Register %6
    0x8883930: i32 = Constant<16>
  0x8882e40: i32 = Constant<0>
In function: arm_float_to_f16
PLEASE submit a bug report to https://github.com/ARM-software/LLVM-embedded-toolchain-for-Arm/issues and include the crash backtrace, preprocessed source, and associated run script.
Stack dump:
0.      Program arguments: clang-18 -c -Wstrict-prototypes -Wno-attributes -Wno-unknown-pragmas -Oz -fno-strict-aliasing -fomit-frame-pointer -fno-builtin -fshort-enums -fno-common -Wall -Wshadow -Wundef -ffunction-sections -fdata-sections -target armv8.1m.main_hard_fp.cfg -march=armv8.1-m.main+mve -mtune=cortex-m85 -mfpu=fpv5-d16 -mfloat-abi=hard -mthumb -Wa,-mthumb -Wa,-mimplicit-it=always --target=arm-none-eabi -isystem ~/clang_test/mcu_common/nuttx/include -isystem ~/clang_test/mcu_common/nuttx/include/newlib -D__NuttX__ -DNDEBUG -DXR_VERSION_FPGA -pipe -Werror -g -I~/clang_test/contexthub/../platform/tuling3 -I~/clang_test/contexthub/../platform/tuling3/fpgdafen/contexthub/../platform/common/pmu/include/tuling3/fpga -I~/clang_test/contexthub/../platform/common/pmu/include/tuling3 -I~/clang_test/contexthub/../platform/common/clk/include/tuling3 -I~/clang_test/contexthub/chips/mcu/cortex-m85/include -I~/clang_test/mcu_common/nuttx/../vendor/contexthub/sensor_drivers/include -I~/clang_test/contexthub/cmsis/CMSIS/CMSIS_5-5.9.0/Device/ARM/ARMCM85/Include -I~/clang_test/contexthub/cmsis/CMSIS/CMSIS_5-5.9.0/CMSIS/Core/Include -I~/clang_test/contexthub/cmsis/CMSIS/CMSIS-DSP-1.15.0/Include -I~/clang_test/contexthub/cmsis/CMSIS/CMSIS-NN-4.1.0/Include -I~/clang_test/contexthub/securec/include -DSECUREC_ENABLE_FSCANF=0 -DSECUREC_ENABLE_VFSCANF=0 -DSECUREC_ENABLE_WCHAR_FUNC=0 -DSECUREC_HAVE_WCHART=0 -DSECUREC_ENABLE_SPRINTF_FLOAT=0 -DSECUREC_HAVE_WCTOMB=0 -DSECUREC_IN_KERNEL=0 -I~/clang_test/contexthub/chips/include -DTODO_REG=0 -I ~/clang_test/mcu_common/apps/vendor/contexthub/cmsis/CMSIS/CMSIS-DSP-1.15.0/Include -I ~/clang_test/mcu_common/apps/vendor/contexthub/cmsis/CMSIS/CMSIS_5-5.9.0/CMSIS/Core/Include/ -I ~/clang_test/mcu_common/apps/include -I dafen//open-source/cmsis/CMSIS/CMSIS-DSP-1.15.0/PrivateInclude -I dafen//open-source/cmsis/CMSIS_5-5.9.0/CMSIS/DSP/Include/ -I dafen//open-source/cmsis/CMSIS_5-5.9.0/CMSIS/Core/Include/ -Wno-shadow -flax-vector-conversions -O3 dafen//open-source/cmsis/CMSIS-DSP-1.15.0/Source/SupportFunctions/arm_float_to_f16.c -o bskarm_float_to_f16.o -v
1.      <eof> parser at end of file
2.      Code generation
3.      Running pass 'Function Pass Manager' on module 'dafen//open-source/cmsis/CMSIS-DSP-1.15.0/Source/SupportFunctions/arm_float_to_f16.c'.
4.      Running pass 'ARM Instruction Selection' on function '@arm_float_to_f16'
 #0 0x0000000001efc018 llvm::sys::PrintStackTrace(llvm::raw_ostream&, int) (~/clang_test/open-source/clang_for_arm/clang18/bin/clang-18+0x1efc018)
 #1 0x0000000001efa294 llvm::sys::CleanupOnSignal(unsigned long) (~/clang_test/open-source/clang_for_arm/clang18/bin/clang-18+0x1efa294)
 #2 0x0000000001e53d83 llvm::CrashRecoveryContext::HandleExit(int) (~/clang_test/open-source/clang_for_arm/clang18/bin/clang-18+0x1e53d83)
 #3 0x0000000001ef3c9e llvm::sys::Process::Exit(int, bool) (~/clang_test/open-source/clang_for_arm/clang18/bin/clang-18+0x1ef3c9e)
 #4 0x0000000000ac64ed (~/clang_test/open-source/clang_for_arm/clang18/bin/clang-18+0xac64ed)
 #5 0x0000000001e5b775 llvm::report_fatal_error(llvm::Twine const&, bool) (~/clang_test/open-source/clang_for_arm/clang18/bin/clang-18+0x1e5b775)
 #6 0x0000000002eefaa1 llvm::SelectionDAGISel::CannotYetSelect(llvm::SDNode*) (~/clang_test/open-source/clang_for_arm/clang18/bin/clang-18+0x2eefaa1)
 #7 0x0000000002ef3882 llvm::SelectionDAGISel::SelectCodeCommon(llvm::SDNode*, unsigned char const*, unsigned int) (~/clang_test/open-source/clang_for_arm/clang18/bin/clang-18+0x2ef3882)
 #8 0x0000000000f26cc7 (~/clang_test/open-source/clang_for_arm/clang18/bin/clang-18+0xf26cc7)
 #9 0x0000000002eecb16 llvm::SelectionDAGISel::DoInstructionSelection() (~/clang_test/open-source/clang_for_arm/clang18/bin/clang-18+0x2eecb16)
#10 0x0000000002efa51c llvm::SelectionDAGISel::CodeGenAndEmitDAG() (~/clang_test/open-source/clang_for_arm/clang18/bin/clang-18+0x2efa51c)
#11 0x0000000002efce76 llvm::SelectionDAGISel::SelectAllBasicBlocks(llvm::Function const&) (~/clang_test/open-source/clang_for_arm/clang18/bin/clang-18+0x2efce76)
#12 0x0000000002eff706 (~/clang_test/open-source/clang_for_arm/clang18/bin/clang-18+0x2eff706)
#13 0x0000000000f2fd24 (~/clang_test/open-source/clang_for_arm/clang18/bin/clang-18+0xf2fd24)
#14 0x000000000147a000 (~/clang_test/open-source/clang_for_arm/clang18/bin/clang-18+0x147a000)
#15 0x00000000019e2608 llvm::FPPassManager::runOnFunction(llvm::Function&) (~/clang_test/open-source/clang_for_arm/clang18/bin/clang-18+0x19e2608)
#16 0x00000000019e277c llvm::FPPassManager::runOnModule(llvm::Module&) (~/clang_test/open-source/clang_for_arm/clang18/bin/clang-18+0x19e277c)
#17 0x00000000019e30ee llvm::legacy::PassManagerImpl::run(llvm::Module&) (~/clang_test/open-source/clang_for_arm/clang18/bin/clang-18+0x19e30ee)
#18 0x0000000002140cf3 clang::EmitBackendOutput(clang::DiagnosticsEngine&, clang::HeaderSearchOptions const&, clang::CodeGenOptions const&, clang::TargetOptions const&, clang::LangOptions const&, llvm::StringRef, llvm::Module*, clang::BackendAction, llvm::IntrusiveRefCntPtr<llvm::vfs::FileSystem>, std::unique_ptr<llvm::raw_pwrite_stream, std::default_delete<llvm::raw_pwrite_stream>>, clang::BackendConsumer*) (~/clang_test/open-source/clang_for_arm/clang18/bin/clang-18+0x2140cf3)
#19 0x00000000026c9b69 clang::BackendConsumer::HandleTranslationUnit(clang::ASTContext&) (~/clang_test/open-source/clang_for_arm/clang18/bin/clang-18+0x26c9b69)
#20 0x0000000003f821a9 clang::ParseAST(clang::Sema&, bool, bool) (~/clang_test/open-source/clang_for_arm/clang18/bin/clang-18+0x3f821a9)
#21 0x000000000293b179 clang::FrontendAction::Execute() (~/clang_test/open-source/clang_for_arm/clang18/bin/clang-18+0x293b179)
#22 0x00000000028be85b clang::CompilerInstance::ExecuteAction(clang::FrontendAction&) (~/clang_test/open-source/clang_for_arm/clang18/bin/clang-18+0x28be85b)
#23 0x00000000029f6063 clang::ExecuteCompilerInvocation(clang::CompilerInstance*) (~/clang_test/open-source/clang_for_arm/clang18/bin/clang-18+0x29f6063)
#24 0x0000000000ac81ce cc1_main(llvm::ArrayRef<char const*>, char const*, void*) (~/clang_test/open-source/clang_for_arm/clang18/bin/clang-18+0xac81ce)
#25 0x0000000000abfce8 _start (~/clang_test/open-source/clang_for_arm/clang18/bin/clang-18+0xabfce8)
#26 0x000000000270d999 (~/clang_test/open-source/clang_for_arm/clang18/bin/clang-18+0x270d999)
#27 0x0000000001e53c83 llvm::CrashRecoveryContext::RunSafely(llvm::function_ref<void ()>) (~/clang_test/open-source/clang_for_arm/clang18/bin/clang-18+0x1e53c83)
#28 0x000000000270dcf4 (~/clang_test/open-source/clang_for_arm/clang18/bin/clang-18+0x270dcf4)
#29 0x00000000026d99dd clang::driver::Compilation::ExecuteCommand(clang::driver::Command const&, clang::driver::Command const*&, bool) const (~/clang_test/open-source/clang_for_arm/clang18/bin/clang-18+0x26d99dd)
#30 0x00000000026da2ad clang::driver::Compilation::ExecuteJobs(clang::driver::JobList const&, llvm::SmallVectorImpl<std::pair<int, clang::driver::Command const*>>&, bool) const (~/clang_test/open-source/clang_for_arm/clang18/bin/clang-18+0x26da2ad)
#31 0x00000000026e93ec clang::driver::Driver::ExecuteCompilation(clang::driver::Compilation&, llvm::SmallVectorImpl<std::pair<int, clang::driver::Command const*>>&) (~/clang_test/open-source/clang_for_arm/clang18/bin/clang-18+0x26e93ec)
#32 0x0000000000ac4d48 clang_main(int, char**, llvm::ToolContext const&) (~/clang_test/open-source/clang_for_arm/clang18/bin/clang-18+0xac4d48)
#33 0x0000000000a06d34 main (~/clang_test/open-source/clang_for_arm/clang18/bin/clang-18+0xa06d34)
#34 0x00007f36b0e7c083 __libc_start_main /build/glibc-LcI20x/glibc-2.31/csu/../csu/libc-start.c:342:3
#35 0x0000000000abf347 _start (~/clang_test/open-source/clang_for_arm/clang18/bin/clang-18+0xabf347)
clang-18: error: clang frontend command failed with exit code 70 (use -v to see invocation)
clang version 18.1.3
Target: arm-none-unknown-eabi
Thread model: posix
InstalledDir: ~/clang_test/open-source/clang_for_arm/clang18/bin
clang-18: note: diagnostic msg:
********************

PLEASE ATTACH THE FOLLOWING FILES TO THE BUG REPORT:
Preprocessed source(s) and associated run script(s) are located at:
clang-18: note: diagnostic msg: /tmp/arm_float_to_f16-5fc438.c
clang-18: note: diagnostic msg: /tmp/arm_float_to_f16-5fc438.sh
clang-18: note: diagnostic msg:

[davemgreen]

Hello. I think we have improved the error since clang-18, but the problem might persist. The issue is likely in the use of -march=armv8.1-m.main+mve -mtune=cortex-m85 -mfpu=fpv5-d16. It tells the compiler that you have MVE from the architecture, but don't have MVE from the FPU. (And nothing has mve.fp). I would recommend just using -mcpu=cortex-m85 instead. If you leave off a -mfpu in clang it will pick a sensible default for the cpu. See https://godbolt.org/z/zaWxjd39P for a reproducer and https://developer.arm.com/documentation/ka005067/latest/ for different cortex-m85 configuration options.