[Bug rtl-optimization/90311] [9/10 Regression] wrong code with -O and __builtin_add_overflow() and compare

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=90311

--- Comment #4 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
At least from other PRs, I think it is important to differentiate in what is
emitted/matched for the multi-word addition (generally that has been using CC_V
mode I think on ARM) vs. what is emitted/matched for multi-word subtraction
(which has been using CC_C mode I think).  In this PR, it seems to be a mix of
both.