https://gcc.gnu.org/bugzilla/show_bug.cgi?id=121789
--- Comment #7 from Avi Kivity <avi at scylladb dot com> --- > Or phrased differently, the choices are > the assume the bug never happens and let > users suffer with the consequences, or > trap/abort on the bug, or silently defuse > the bug and make this function a no-op. Well, when you don't optimize to memcpy(), you take the first option. I believe it's better to escalate the bug to a trap than to downgrade it to a no-op but I can understand other opinions.
