Bug#956324: Clustalo bus error on mipsel (Was: Bug#956324: python-biopython: FTBFS on mipsel)
Adrian Bunk <[email protected]>
| Newsgroups | gmane.linux.debian.devel.bugs.general,gmane.linux.debian.devel.mentors,gmane.linux.debian.ports.mips,gmane.linux.debian.devel.gcc |
|---|---|
| Message-ID | <20200511070424.GB19155@localhost> |
On Sun, May 10, 2020 at 08:56:50PM +0200, Andreas Tille wrote: >... > On Sun, May 10, 2020 at 11:19:27AM +0300, Adrian Bunk wrote: > > What does fix the problem is disabling OpenMP. > > I suspect OpenMP is somehow broken in gcc >= 8 on mipsel. > > I wonder how we could keep this finding to other packages. I bet it > would not the only issue but it has shown up due to intense testing. >... > I've now uploaded now with this patch closing the bug - but as I tried to > express I'd sleep a bit better if this issue would be recorded somewhere > else than in a closed bug. A bug buried somewhere in the Debian bts would not change anything. What would have to happen would be a Debian MIPS porter debugging this problem and then submitting a minimal testcase to gcc upstream. Assuming this will be confirmed to be a mipsel-specific gcc bug, this looks likely but it is still possible that there actually is a bug in clustalo that works by chance on other architectures. Unlikely, but not impossible. In a more positive note, I assume people would have already noticed if OpenMP on mipsel in stable and unstable would be completely broken. > Thanks again for your research > > Andreas. cu Adrian