Created attachment 1603249 [details] root.log The EPEL7 build of openblas-0.3.7-1 fails with an internal compiler error on the ppc64le architecture. I don't have access to a ppc64le machine, so I can't supply the preprocessed source. The error should be readily reproducible from the srpm. koji build logs attached.
Created attachment 1603250 [details] build.log
original koji build at https://koji.fedoraproject.org/koji/taskinfo?taskID=36953437
Created attachment 1614604 [details] Preprocessed source from upstream, https://github.com/xianyi/OpenBLAS/issues/2254#issuecomment-530891558 While I don't have access to the hardware, upstream has produced a preprocessed source.
------- Comment From wschmidt.com 2019-09-24 14:45 EDT------- *** Bug 181567 has been marked as a duplicate of this bug. ***
Reduced testcase for -O2 -mvsx -mcpu=power8: typedef __attribute__((altivec(vector__))) unsigned int V; void foo (V *p) { register V a = {4,4,4,4}; register V b = a << 1; b = b << 1; *p = b; } I believe this has been fixed upstream with r198772 aka https://gcc.gnu.org/ml/gcc-patches/2013-05/msg00378.html
------- Comment From edelsohn.com 2019-09-24 15:29 EDT------- There appears to be one bug that already has been fixed, and the patch should be considered for GCC 4.8.5 RH. The OpenBLAS community uncovered a separate bug that reproduces in GCC 8.3 and is resolved with the same work-around that the OpenBLAS community applied for the first bug. I have asked the OpenBLAS community to create a reduced testcase and open an issue in GCC Bugzilla for the second issue.
Patch commmitted.
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory (gcc bug fix and enhancement update), and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHBA-2020:3926