[Bug target/69459] [5/6 Regression] wrong code with -O2 and vector arithmetics @ x86_64

[zsojka at seznam dot cz]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=69459

--- Comment #3 from Zdenek Sojka <zsojka at seznam dot cz> ---
Created attachment 37455
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=37455&action=edit
original testcase

(In reply to Zdenek Sojka from comment #0)
...
> 
> Tested revisions:
> trunk r232760 - FAIL
> 5-branch r232545 - FAIL
> 4_[987]-branch - OK

Actually, the unreduced testcase fails at 4_9-branch as well; and there are
further similar testcases, some of them failing since 4_7-branch.