https://gcc.gnu.org/bugzilla/show_bug.cgi?id=113688
--- Comment #18 from Eric Botcazou <ebotcazou at gcc dot gnu.org> --- > It fixes a regression (but maybe I should have kept this in the title until > the release). Still, if it causes too much trouble, we can revert it. But > let me understand the issue first. What is the situation for 15? On the mainline the VIEW_CONVERT_EXPR is retained so the test passes.