[Bug tree-optimization/123196] [16 regression] gcc.dg/vect/vect-105.c FAILs

[ro at CeBiTec dot Uni-Bielefeld.DE]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=123196

--- Comment #3 from ro at CeBiTec dot Uni-Bielefeld.DE <ro at CeBiTec dot Uni-Bielefeld.DE> ---
> --- Comment #2 from ro at CeBiTec dot Uni-Bielefeld.DE <ro at CeBiTec dot Uni-Bielefeld.DE> ---
>> profile is hard, we tend to remove those scans on testcases where it breaks. 
>> Unless you want to bisect?  Apart from some profile change that look unrelated
>> there's r16-1624-g6deab186535a5a that looks suspicious.
>>
>> Can you confirm this causes it?
>
> I cannot revert that patch cleanly, so I'll bisect.

Confirmed: this is the first bad one:

commit 6deab186535a5aa9f930e2db637089865d0bc4ff
Author: Pengfei Li <Pengfei.Li2@arm.com>
Date:   Wed Jun 11 15:01:36 2025 +0000

    vect: Use combined peeling and versioning for mutually aligned DRs

I'm attaching the good and bad dumps.