[CUDA] bindless_images tests failed after f9c8c01d38f8fb

[jsji]

Describe the bug

With the new optimization in llvm/llvm-project#119730

[NVPTX] Aggressively try to replace image handles with references

The following test failed for CUDA.

Failed Tests (7):
  SYCL :: bindless_images/examples/example_1_1D_read_write.cpp
  SYCL :: bindless_images/examples/example_2_2D_dynamic_read.cpp
  SYCL :: bindless_images/examples/example_4_1D_array_read_write.cpp
  SYCL :: bindless_images/read_2D_dynamic.cpp
  SYCL :: bindless_images/read_write_1D.cpp
  SYCL :: bindless_images/read_write_1D_subregion.cpp
  SYCL :: bindless_images/user_types/read_write_user_type.cpp

f9c8c01 is reverted temporaily to unblock the pulldown, please investigate/fix and reland.

See details in
https://github.com/intel/llvm/actions/runs/12563446905/job/35025604202

FAIL: SYCL :: bindless_images/examples/example_1_1D_read_write.cpp (2117 of 2239)
******************** TEST 'SYCL :: bindless_images/examples/example_1_1D_read_write.cpp' FAILED ********************
Exit Code: 1

Command Output (stdout):
--
# RUN: at line 3
/__w/llvm/llvm/toolchain/bin//clang++  -Werror  -fsycl -fsycl-targets=nvptx64-nvidia-cuda  /__w/llvm/llvm/llvm/sycl/test-e2e/bindless_images/examples/example_1_1D_read_write.cpp -o /__w/llvm/llvm/build-e2e/bindless_images/examples/Output/example_1_1D_read_write.cpp.tmp.out
# executed command: /__w/llvm/llvm/toolchain/bin//clang++ -Werror -fsycl -fsycl-targets=nvptx64-nvidia-cuda /__w/llvm/llvm/llvm/sycl/test-e2e/bindless_images/examples/example_1_1D_read_write.cpp -o /__w/llvm/llvm/build-e2e/bindless_images/examples/Output/example_1_1D_read_write.cpp.tmp.out
# .---command stderr------------
# | Unknown instruction operating on handle
# | UNREACHABLE executed at /__w/llvm/llvm/src/llvm/lib/Target/NVPTX/NVPTXReplaceImageHandles.cpp:1854!
# | PLEASE submit a bug report to https://github.com/intel/llvm/issues and include the crash backtrace, preprocessed source, and associated run script.
# | Stack dump:
# | 0.	Program arguments: /__w/llvm/llvm/toolchain/bin/clang-20 -cc1 -triple nvptx64-nvidia-cuda -aux-triple x86_64-unknown-linux-gnu -fsycl-is-device -fdeclare-spirv-builtins -fenable-sycl-dae -Wno-sycl-strict -O2 -fsycl-int-header=/tmp/lit-tmp-2a8clmef/example_1_1D_read_write-header-d831a9.h -fsycl-int-footer=/tmp/lit-tmp-2a8clmef/example_1_1D_read_write-footer-ef7000.h -D__SYCL_TARGET_NVIDIA_GPU_SM_50__ -sycl-std=2020 -fsycl-unique-prefix=uid8f5bf9866a496f15 -D__SYCL_ANY_DEVICE_HAS_ANY_ASPECT__=1 -D__SYCL_ALL_DEVICES_HAVE_ext_intel_device_id__=1 -D__SYCL_ALL_DEVICES_HAVE_ext_intel_device_info_uuid__=1 -D__SYCL_ALL_DEVICES_HAVE_ext_intel_free_memory__=1 -D__SYCL_ALL_DEVICES_HAVE_ext_intel_max_mem_bandwidth__=1 -D__SYCL_ALL_DEVICES_HAVE_ext_intel_memory_bus_width__=1 -D__SYCL_ALL_DEVICES_HAVE_ext_intel_memory_clock_rate__=1 -D__SYCL_ALL_DEVICES_HAVE_ext_intel_pci_address__=1 -D__SYCL_ALL_DEVICES_HAVE_ext_oneapi_ballot_group__=1 -D__SYCL_ALL_DEVICES_HAVE_ext_oneapi_bindless_images__=1 -D__SYCL_ALL_DEVICES_HAVE_ext_oneapi_bindless_images_1d_usm__=1 -D__SYCL_ALL_DEVICES_HAVE_ext_oneapi_bindless_images_2d_usm__=1 -D__SYCL_ALL_DEVICES_HAVE_ext_oneapi_bindless_images_shared_usm__=1 -D__SYCL_ALL_DEVICES_HAVE_ext_oneapi_cubemap__=1 -D__SYCL_ALL_DEVICES_HAVE_ext_oneapi_cubemap_seamless_filtering__=1 -D__SYCL_ALL_DEVICES_HAVE_ext_oneapi_external_memory_import__=1 -D__SYCL_ALL_DEVICES_HAVE_ext_oneapi_external_semaphore_import__=1 -D__SYCL_ALL_DEVICES_HAVE_ext_oneapi_fixed_size_group__=1 -D__SYCL_ALL_DEVICES_HAVE_ext_oneapi_graph__=1 -D__SYCL_ALL_DEVICES_HAVE_ext_oneapi_image_array__=1 -D__SYCL_ALL_DEVICES_HAVE_ext_oneapi_limited_graph__=1 -D__SYCL_ALL_DEVICES_HAVE_ext_oneapi_mipmap__=1 -D__SYCL_ALL_DEVICES_HAVE_ext_oneapi_mipmap_anisotropy__=1 -D__SYCL_ALL_DEVICES_HAVE_ext_oneapi_mipmap_level_reference__=1 -D__SYCL_ALL_DEVICES_HAVE_ext_oneapi_native_assert__=1 -D__SYCL_ALL_DEVICES_HAVE_ext_oneapi_opportunistic_group__=1 -D__SYCL_ALL_DEVICES_HAVE_ext_oneapi_unique_addressing_per_dim__=1 -D__SYCL_ALL_DEVICES_HAVE_fp64__=1 -D__SYCL_ALL_DEVICES_HAVE_gpu__=1 -D__SYCL_ALL_DEVICES_HAVE_online_compiler__=1 -D__SYCL_ALL_DEVICES_HAVE_online_linker__=1 -D__SYCL_ALL_DEVICES_HAVE_queue_profiling__=1 -D__SYCL_ALL_DEVICES_HAVE_usm_atomic_host_allocations__=1 -D__SYCL_ALL_DEVICES_HAVE_usm_atomic_shared_allocations__=1 -D__SYCL_ALL_DEVICES_HAVE_usm_device_allocations__=1 -D__SYCL_ALL_DEVICES_HAVE_usm_host_allocations__=1 -D__SYCL_ALL_DEVICES_HAVE_usm_shared_allocations__=1 -D__SYCL_ALL_DEVICES_HAVE_usm_system_allocations__=1 -S -dumpdir /__w/llvm/llvm/build-e2e/bindless_images/examples/Output/example_1_1D_read_write.cpp.tmp.out- -disable-free -clear-ast-before-backend -main-file-name example_1_1D_read_write.cpp -mrelocation-model static -mframe-pointer=all -ffp-contract=on -fno-rounding-math -no-integrated-as -aux-target-cpu x86-64 -internal-isystem /__w/llvm/llvm/toolchain/bin/../include/sycl/stl_wrappers -internal-isystem /__w/llvm/llvm/toolchain/bin/../include -mlink-builtin-bitcode /__w/llvm/llvm/toolchain/lib/clang/20/../../clc/remangled-l64-signed_char.libspirv-nvptx64-nvidia-cuda.bc -mlink-builtin-bitcode /usr/local/cuda-12.6/nvvm/libdevice/libdevice.10.bc -target-sdk-version=12.6 -target-cpu sm_50 -target-feature +ptx85 -debugger-tuning=gdb -fno-dwarf-directory-asm -fdebug-compilation-dir=/__w/llvm/llvm/build-e2e/bindless_images/examples -resource-dir /__w/llvm/llvm/toolchain/lib/clang/20 -Werror -ferror-limit 19 -fgpu-rdc -fgnuc-version=4.2.1 -fskip-odr-check-in-gmf -o /tmp/lit-tmp-2a8clmef/example_1_1D_read_write-sm_50-e69480-a1d095.s -x ir /tmp/lit-tmp-2a8clmef/example_1_1D_read_write-sm_50-468b68_0.bc
# | 1.	Code generation
# | 2.	Running pass 'Function Pass Manager' on module '/tmp/lit-tmp-2a8clmef/example_1_1D_read_write-sm_50-468b68_0.bc'.
# | 3.	Running pass 'NVPTX Replace Image Handles' on function '@_ZTSN4sycl3_V16detail18RoundedRangeKernelINS0_4itemILi1ELb1EEELi1EZZ4mainENKUlRNS0_7handlerEE_clES6_EUlNS0_2idILi1EEEE_EE'
# | Stack dump without symbol names (ensure you have llvm-symbolizer in your PATH or set the environment var `LLVM_SYMBOLIZER_PATH` to point to it):
# | 0  clang-20  0x000000000[26](https://github.com/intel/llvm/actions/runs/12563446905/job/35025604202#step:24:27)31d98 llvm::sys::PrintStackTrace(llvm::raw_ostream&, int) + 40
# | 1  clang-20  0x000000000262f84e llvm::sys::RunSignalHandlers() + 238
# | 2  clang-20  0x0000000002632428
# | 3  libc.so.6 0x00007fcf5ca6f320
# | 4  libc.so.6 0x00007fcf5cac8b1c pthread_kill + 284
# | 5  libc.so.6 0x00007fcf5ca6f26e gsignal + 30
# | 6  libc.so.6 0x00007fcf5ca528ff abort + 223
# | 7  clang-20  0x00000000025a939f
# | 8  clang-20  0x0000000000c96fb0
# | 9  clang-20  0x0000000000c94257
# | 10 clang-20  0x0000000001ad1197 llvm::MachineFunctionPass::runOnFunction(llvm::Function&) + 695
# | 11 clang-20  0x00000000020[27](https://github.com/intel/llvm/actions/runs/12563446905/job/35025604202#step:24:28)aa5 llvm::FPPassManager::runOnFunction(llvm::Function&) + 629
# | 12 clang-20  0x000000000202fef2 llvm::FPPassManager::runOnModule(llvm::Module&) + 50
# | 13 clang-20  0x00000000020[28](https://github.com/intel/llvm/actions/runs/12563446905/job/35025604202#step:24:29)526 llvm::legacy::PassManagerImpl::run(llvm::Module&) + 1910
# | 14 clang-20  0x000000000284b0fe 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*) + 7694
# | 15 clang-20  0x0000000002e1f25a clang::CodeGenAction::ExecuteAction() + 2698
# | 16 clang-20  0x000000000[30](https://github.com/intel/llvm/actions/runs/12563446905/job/35025604202#step:24:31)c859f clang::FrontendAction::Execute() + 95
# | 17 clang-20  0x000000000303583d clang::CompilerInstance::ExecuteAction(clang::FrontendAction&) + 1149
# | 18 clang-20  0x000000000[31](https://github.com/intel/llvm/actions/runs/12563446905/job/35025604202#step:24:32)a0089 clang::ExecuteCompilerInvocation(clang::CompilerInstance*) + 473
# | 19 clang-20  0x0000000000c12b7c cc1_main(llvm::ArrayRef<char const*>, char const*, void*) + 6284
# | 20 clang-20  0x0000000000c0ecdf
# | 21 clang-20  0x0000000000c0ddfb clang_main(int, char**, llvm::ToolContext const&) + 5099
# | 22 clang-20  0x0000000000c1d087 main + 87
# | 23 libc.so.6 0x00007fcf5ca541ca
# | 24 libc.so.6 0x00007fcf5ca5428b __libc_start_main + 139
# | 25 clang-20  0x0000000000c0c705 _start + 37
# | llvm-foreach: Aborted (core dumped)
# | clang++: error: clang frontend command failed with exit code 254 (use -v to see invocation)
# | clang version 20.0.0git (https://github.com/intel/llvm 724f840609fce137a874dc6f1c846c3a3044b159)
# | Target: x86_64-unknown-linux-gnu
# | Thread model: posix
# | InstalledDir: /__w/llvm/llvm/toolchain/bin
# | Build config: +assertions
# | clang++: note: diagnostic msg: Error generating preprocessed source(s).
# `-----------------------------
# error: command failed with exit status: 1

--

[jsji]

FYI. @intel/bindless-images-reviewers

[ProGTX]

Thanks for reporting this, we're tracking this internally.