https://gcc.gnu.org/bugzilla/show_bug.cgi?id=78251
--- Comment #5 from Iain Sandoe <iains at gcc dot gnu.org> --- is this a GCC bug, or really a gettext /iconv one? (the comments at the top of the file suggest that these two m4 modules have been imported) A quick scan of the code suggests that the writers intended the CPPFLAGS to be saved. However, variable names like "am_save_CPPFLAGS" can overwrite previously saved values in the case of nested configuration tests; it's a very common choice of save name and I have seen this in GCC config problems before. If you change the am_save_CPPFLAGS in the iconv m4 to am_iconv_save_CPPFLAGS (and regenerate all configures that include it) does that resolve the issue? It looks like the gettext save value is already intended to be specific (gt_save_CPPFLAGS)