Created attachment 1542027 [details] Reduced test case from gmp-ecm Description of problem: The gmp-ecm package failed the mass rebuild, on s390x only. I will attach a reduced version of the failing test. When built with "gcc -O2 -Wall -Wextra -fno-strict-aliasing -fwrapv -fno-aggressive-loop-optimizations", the code segfaults due to walking off the end of an allocated block of memory. Libusan is unavailable on s390x, so I cannot check with -fsanitize=undefined, but on x86_64 the sanitizer gives no errors. When built with "gcc -O2 -fno-code-hoisting", the code does not segfault, and neither valgrind nor the address sanitizer shows any errors. Version-Release number of selected component (if applicable): gcc-9.0.1-0.4.fc30.s390x How reproducible: Always Steps to Reproduce: 1. Build the attached code with -O2 2. 3. Actual results: The code segfaults unless -fno-code-hoisting is also given. Expected results: No segfault. Additional info:
This seems to be fixed with upstream http://gcc.gnu.org/r269302 aka http://gcc.gnu.org/PR89497 fix.
Great! Thank you, Jakub.
gcc-9.0.1-0.10.fc30 has been submitted as an update to Fedora 30. https://bodhi.fedoraproject.org/updates/FEDORA-2019-833ad0f15b
gcc-9.0.1-0.10.fc30 has been pushed to the Fedora 30 testing repository. If problems still persist, please make note of it in this bug report. See https://fedoraproject.org/wiki/QA:Updates_Testing for instructions on how to install test updates. You can provide feedback for this update here: https://bodhi.fedoraproject.org/updates/FEDORA-2019-833ad0f15b
gcc-9.0.1-0.10.fc30 has been pushed to the Fedora 30 stable repository. If problems still persist, please make note of it in this bug report.