[DRE-maint] Bug#996724: ruby3.0: FTBFS on ppc64el: Segmentation fault

[Antonio Terceiro]

On Mon, Oct 18, 2021 at 01:22:25PM +0200, John Paul Adrian Glaubitz wrote:
> On 10/18/21 13:16, Antonio Terceiro wrote:
> > I don't think we are on the right track here. I need to read the failing
> > test correctly. this "test case" alone is bogus, as
> > 
> > ruby -e 'END {Process.kill :SEGV, $$}'
> > 
> > is _expected_ to segfault, i.e. the process is sending a SEGV signal to
> > itself. This test case is not complete.

Those tests pass on the upstream master branch, but I couldn't really
bisect the issue due to other build failures in the intermediary
commits.

In the end I "solved" this particular issue by skipping the tests that
cause that failure. I'm not really happy about that, but the other 21k+
tests pass so I guess I will have to live with it.

> It might be an idea to look at the Fedora package:
> 
> > https://src.fedoraproject.org/rpms/ruby/blob/rawhide/f/ruby.spec
> 
> It has a number of patches against 3.x:
> 
> > https://src.fedoraproject.org/rpms/ruby/tree/rawhide
> 
> It builds fine on ppc64el:
> 
> > https://koji.fedoraproject.org/koji/buildinfo?buildID=1801321

The test run here runs less tests than we do so I assume there's also a
skip list somewhere.

> OTOH, ruby3.0 builds fine in openSUSE Factory on ppc64el, but they're still at
> version 3.0.1, so it might be an issue that will only show with 3.0.2?
> 
> > https://build.opensuse.org/package/show/openSUSE:Factory:PowerPC/ruby3.0

the build log for ppc64el also has Segmentation faults, that I assume is
not being considered fatal.

Thanks for your help.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: not available
URL: <http://alioth-lists.debian.net/pipermail/pkg-ruby-extras-maintainers/attachments/20211019/e41685bd/attachment.sig>