https://gcc.gnu.org/bugzilla/show_bug.cgi?id=78102
--- Comment #10 from Uroš Bizjak <ubizjak at gmail dot com> --- (In reply to Jakub Jelinek from comment #8) > How would it expand? Repeat all the vector condition/comparison lowering > code from tree-vect-generic.c on RTL? Compared to that I think it would be > simpler to handle it in ix86_expand_int_vcond - if the only problematic mode > is V2DImode, then that might be easier than just writing code that would > handle arbitrary vectors. That still means the vectorizer believes all the > comparisons are supported and might result in really terrible code (rather > than just giving up). There are predicates in optabs-tree.c: expand_vec_cmp_expr_p and expand_vec_cond_expr_p) that currently only check if the insn code is available. But, you are right. These predicates don't check if the comparison code can be handled by the optab...
