https://gcc.gnu.org/bugzilla/show_bug.cgi?id=127297
--- Comment #2 from Joshua Berne <berne at notadragon dot com> --- I think that could work for this bug, but when working on fixing https://gcc.gnu.org/bugzilla/show_bug.cgi?id=126897 i ended up moving the diagnosis of this issue entirely somewhere else, and that resulted in this particular bug going away. Basically, odr-use is hard to identify when we complete an id-expression, so it seemed better to do the check by walking the completed tree once we are finished updating/substituting a postcondition assertion.
