Hi, These tests are incorrect. Therefore, we will correct the tests.
Details are as follows. Investigation shows that all failures are caused by assumptions in the test programs rather than compiler defects. Two independent issues were identified.
Issue 1: Fujitsu-C-0060_0008.test and Fujitsu-C-0076_0009.test These tests use NaN/Infinity related computations while being compiled with -ffast-math. The source files explicitly enable: - -fhonor-nans but do not enable: - -fhonor-infinities As a result, optimizations assuming that infinities are not significant remain legal under -ffast-math. Fujitsu-C-0060_0008.test generates Inf and -Inf through division by zero and compares them against expected values. The test assumes IEEE infinity semantics are preserved, but the compilation options do not require that behavior. Fujitsu-C-0076_0009.test exhibits the same problem.
Issue 2: Fujitsu-Fortran-0311_0026.test This test uses Cray pointers and an allocatable array that shares the same storage. However, the allocatable arrays participating in the aliasing relationship do not have the TARGET attribute. Under Flang's documented aliasing rules for Cray pointers, optimizations may assume that Cray-pointer targets do not alias ordinary variables unless the aliased variables are declared with the TARGET attribute. Consequently, the optimizer may legally treat the two objects as non-aliasing despite referring to the same storage at runtime. The failure therefore originates from the test program not expressing the required aliasing relationship.
Please ignore the mail. We'll update the test. We apologize for the false detection.
Best regards, Yasuyuki Ohno
Software Development Div., Advanced Technology Development Unit, Fujitsu Research, Fujitsu Limited, JAPAN. ________________________________ 差出人: ci_notify@linaro.org ci_notify@linaro.org 送信: 2026 年 8 月 29 日 (土曜日) 16:20 宛先: contact-ci-compiler-test-suite@cs.jp.fujitsu.com contact-ci-compiler-test-suite@cs.jp.fujitsu.com Cc: jfargnoli@nvidia.com jfargnoli@nvidia.com; maxim.kuvyrkov@linaro.org maxim.kuvyrkov@linaro.org 件名: [Linaro-TCWG-CI] llvmorg-24-init-6257-gc47197af5327: 3 regressions on aarch64
Dear contributor,
Our automatic CI has detected problems related to your patch(es). Please find some details below.
In tcwg_flang_test/main-aarch64-Ofast-sve_vls-lto-lld, after: | commit llvmorg-24-init-6257-gc47197af5327 | Author: Justin Fargnoli jfargnoli@nvidia.com | Date: Thu Aug 27 13:03:39 2026 -0700 | | [LTO] Enable `PrepareForLTO` for loop unrolling in pre-link pipelines (#192155) | | Currently, the unroller will not unroll loops with inlineable calls. | However, calls that may not appear to be inlineable during the pre-link | phase may become inlineable during the post-link phase. The unroller's | ... 4 lines of the commit log omitted.
Produces 3 regressions: | | regressions.sum: | Running test-suite:Fujitsu/C/0060 ... | FAIL: test-suite :: Fujitsu/C/0060/Fujitsu-C-0060_0008.test | Running test-suite:Fujitsu/C/0076 ... | FAIL: test-suite :: Fujitsu/C/0076/Fujitsu-C-0076_0009.test | Running test-suite:Fujitsu/Fortran/0311 ... | ... and 1 more | # "FAIL" means : the execution of the compiled binary failed / output of the binary differs from the expected one
Used configuration : * Toolchain : cmake -G Ninja ../llvm/llvm "-DLLVM_ENABLE_PROJECTS=clang;lld;flang;clang-tools-extra" "-DLLVM_ENABLE_RUNTIMES=openmp" -DCMAKE_BUILD_TYPE=Release -DLLVM_ENABLE_ASSERTIONS=True -DCMAKE_INSTALL_PREFIX=../llvm-install "-DLLVM_TARGETS_TO_BUILD=AArch64" -DCLANG_DEFAULT_LINKER=lld * Testsuite : export LD_LIBRARY_PATH=$\WORKSPACE/llvm-install/lib/aarch64-unknown-linux-gnu${LD_LIBRARY_PATH:+:$\LD_LIBRARY_PATH} cmake -GNinja -DCMAKE_C_COMPILER="$\WORKSPACE/llvm-install/bin2/clang" -DCMAKE_CXX_COMPILER="$\WORKSPACE/llvm-install/bin2/clang++" -DCMAKE_Fortran_COMPILER="$\WORKSPACE/llvm-install/bin2/flang-new" -DCMAKE_BUILD_TYPE=Release -DCMAKE_C_FLAGS= -DCMAKE_CXX_FLAGS= -DCMAKE_Fortran_FLAGS= -DCMAKE_C_FLAGS_RELEASE="-O3 -ffast-math -march=armv8.4-a+sve -msve-vector-bits=256 -flto -fuse-ld=lld -DNDEBUG" -DCMAKE_CXX_FLAGS_RELEASE="-O3 -ffast-math -march=armv8.4-a+sve -msve-vector-bits=256 -flto -fuse-ld=lld -DNDEBUG" -DCMAKE_Fortran_FLAGS_RELEASE="-O3 -ffast-math -march=armv8.4-a+sve -msve-vector-bits=256 -flto -fuse-ld=lld -DNDEBUG" -DTEST_SUITE_FORTRAN=ON -DTEST_SUITE_SUBDIRS=Fujitsu -DTEST_SUITE_FUJITSU_WITH_FAST_MATH=ON "$\WORKSPACE/test/test-suite"
We track this bug report under https://linaro.atlassian.net/browse/LLVM-2268. (Require Linaro login) Please let us know if you have a fix.
If you have any questions regarding this report, please ask on linaro-toolchain@lists.linaro.org mailing list.
-----------------8<--------------------------8<--------------------------8<--------------------------
The information below contains the details of the failures, and the ways to reproduce a debug environment:
You can find the failure logs in *.log.1.xz files in * http://54.172.246.49:9090/jobs/tcwg_flang_test--main-aarch64-Ofast-sve_vls-l... The full lists of regressions and improvements as well as configure and make commands are in * http://54.172.246.49:9090/jobs/tcwg_flang_test--main-aarch64-Ofast-sve_vls-l... The list of [ignored] baseline and flaky failures are in * http://54.172.246.49:9090/jobs/tcwg_flang_test--main-aarch64-Ofast-sve_vls-l...
Fujitsu testsuite : https://github.com/fujitsu/compiler-test-suite/
Current build : http://54.172.246.49:9090/jobs/tcwg_flang_test--main-aarch64-Ofast-sve_vls-l... Reference build : http://54.172.246.49:9090/jobs/tcwg_flang_test--main-aarch64-Ofast-sve_vls-l...
Instruction to reproduce the build : https://gitlab.com/LinaroLtd/tcwg/ci/interesting-commits/-/raw/master/llvm/s...
Full commit : https://github.com/llvm/llvm-project/commit/c47197af53270a4795040756c27802f2...