[Bug target/123089] [16 regression] [x86] Excessive run time with -march=znver3

[dcb314 at hotmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=123089

David Binderman <dcb314 at hotmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|needs-bisection             |
                 CC|                            |tamar.christina at arm dot com

--- Comment #2 from David Binderman <dcb314 at hotmail dot com> ---
As expected:

trunk $ git bisect bad 65a3849eb46df2fb
65a3849eb46df2fbac6b41ff78dae13c85387f9e is the first bad commit
commit 65a3849eb46df2fbac6b41ff78dae13c85387f9e (HEAD)
Author: Tamar Christina <tamar.christina@arm.com>
Date:   Sun Nov 30 07:29:50 2025 +0000

    vect: support vectorization of early break forced live IVs as scalar



For reduction, I got most of a cvise script done:

(/usr/bin/gcc -w bug1158.c) && 
        (./a.out | grep "checksum = 34099839") &&
        (~/gcc/results/bin/gcc -w -fno-strict-aliasing -O3 -march=znver3
-freport-bug bug1
158.c) &&
        (ulimit -t 10; ./a.out)

but I haven't looked at how to capture that the final run of a.out
gets killed. A return code of 137 is produced.

Maybe someone with more bash knowledge would find this easy.