https://gcc.gnu.org/bugzilla/show_bug.cgi?id=124820
--- Comment #8 from GCC Commits <cvs-commit at gcc dot gnu.org> --- The master branch has been updated by Jeff Law <[email protected]>: https://gcc.gnu.org/g:309d3c344a47fc3fbc1f4e256fc2437dbbe7c323 commit r17-5786-g309d3c344a47fc3fbc1f4e256fc2437dbbe7c323 Author: Merlin Warner-Huish <[email protected]> Date: Sun Nov 1 10:55:31 2026 -0700 [PATCH] testsuite: Add -fno-signed-zeros to more FMA tests [PR124820] r17-5664 restricted the negated FMA folds in match.pd to !HONOR_SIGNED_ZEROS. These tests expect those folds to apply with the default options, so they started failing on aarch64. Add -fno-signed-zeros to them, as that commit did for the i386 FMA tests. --- These are the 65 aarch64 failures reported by the Linaro CI after r17-5664. Tested with an aarch64-linux-gnu cross compiler (they are all compile-only tests): all of them fail before this patch and pass after it. Also checked gcc.dg/fma-[567].c on x86_64-pc-linux-gnu. The unpacked_cond_* and vcond_* tests cover the IFN_COND_* folds the original patch changed, so those changes are tested after all. My write access request is still pending, so I'd be grateful if someone could commit this once approved. PR middle-end/124820 gcc/testsuite/ChangeLog: * gcc.dg/fma-5.c: Add -fno-signed-zeros. * gcc.dg/fma-6.c: Likewise. * gcc.dg/fma-7.c: Likewise. * gcc.target/aarch64/sve/unpacked_cond_fnmla_1.c: Likewise. * gcc.target/aarch64/sve/unpacked_cond_fnmla_2.c: Likewise. * gcc.target/aarch64/sve/unpacked_cond_fnmls_1.c: Likewise. * gcc.target/aarch64/sve/unpacked_cond_fnmls_2.c: Likewise. * gcc.target/aarch64/sve/unpacked_fnmla_1.c: Likewise. * gcc.target/aarch64/sve/unpacked_fnmla_2.c: Likewise. * gcc.target/aarch64/sve/unpacked_fnmls_1.c: Likewise. * gcc.target/aarch64/sve/unpacked_fnmls_2.c: Likewise. * gcc.target/aarch64/sve/vcond_13.c: Likewise. * gcc.target/aarch64/sve/vcond_14.c: Likewise. * gcc.target/aarch64/sve/vcond_15.c: Likewise. * gcc.target/aarch64/sve/vcond_16.c: Likewise.
