I was wondering if I could ask an aarch64 maintainer/reviewer to take a look at the aarch64 specific pieces (improvements to rtx_costs) of my patch at: https://gcc.gnu.org/pipermail/gcc-patches/2026-March/709586.html
The middle-end and the x86_64 pieces have both been approved, so it's just the aarch64 specific bits holding up this regression fix. Alas, there's a catch. According to the linaroo CI tester, the new testcase gcc.target/aarch64/pr123238.c is failing on their set-up, but I've been unable to reproduce this on either a cross-build to aarch64-elf or a native build on iains' aarch64-apple-darwin24.3.0 tree. I believe that it's just a problem with the dejagnu magic that I'm missing (are there aarch64 without vector support?) This patch can always applied without a new testcase for aarch64, but it would be nice to understand what's going on. Many thanks in advance, Roger --
