[Bug tree-optimization/126008] [17 Regression] Segmentation Fault in output at -O2 on x86_64-linux-gnu since r17-1586

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=126008

--- Comment #10 from Richard Biener <rguenth at gcc dot gnu.org> ---
OK, so apart from IVOPTs computing bullshit base objects, it computes "I have
see multiple", but tree-ssa-address.cc doesn't even look at that case.  Also
rewrite_use_address only considers the candidate IV base object, but
get_computation_aff relates both candidate and use.  In the PR125730 case
the use IV has a integer_zero_node base object (multiple base objects).

I have pieces of a fix...