> I expected something like that.  In such case there should be XFAIL (or maybe
> a different pseudo-FAIL variant but XFAIL seems OK to me) like we did in:

Can you try the following? Looking at the output, providing the readelf
output is just going to clutter the testcase, IMO, so I left it out.
The ___XA type is missing, it's not malformed, and there isn't much
to add (again, IMO).

Thanks,
-- 
Joel