Re: [PATCH 0/4] nolibc: add support for the s390 platform

[Willy Tarreau]

On Tue, Jan 10, 2023 at 08:32:42AM +0100, Willy Tarreau wrote:
> On Mon, Jan 09, 2023 at 11:15:34AM -0800, Paul E. McKenney wrote:
> > On Mon, Jan 09, 2023 at 09:09:06AM +0100, Willy Tarreau wrote:
> > > Hello Paul,
> > > 
> > > this series adds support for the s390x platform to nolibc and rcutorture,
> > > reducing the init size from ~600kB to ~1kB. The work was contributed by
> > > Sven Schnelle. It depends on the fixes series I sent previously:
> > > 
> > >   https://lore.kernel.org/lkml/20230109075442.25963-1-w@xxxxxx/
> > > 
> > > It passes the self-tests correctly and the patches are clean, please
> > > consider queuing it.
> > 
> > "80 test(s) passed" and successful exit code.  I will assume that the
> > decrease in tests from 81 to 80 is intended.  I have these queued,
> > but it may be some hours before they are externally visible.
> 
> Interesting, I didn't notice and I'm not observing this when running
> qemu-s390x (the userland version), I'm currently rebuilding the kernel
> to compare and will let you know. Thanks for checking!

So after a re-check, I'm constantly seeing 82 tests passed and a success
here. I don't understand what's differing, and would be interested in a
copy of your "run.out" in the "tools/testing/selftests/nolibc" subdir to
compare with mine, as it's possible we have a bug somewhere in the way
tests are run (or maybe we're still facing a mangled qemu output).

You can just send this to me privately, no need to spam the list, the
file is moderately large and uninteresting.

Thanks!
Willy