https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65586
--- Comment #3 from Jakub Jelinek <jakub at gcc dot gnu.org> --- Comment on attachment 35157 --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=35157 Untested draft patch I'd say the C++ parser change is ok, but the C parser change is not, I think c_parser_skip_until_found will not do gcc_assert (parser->in_pragma); parser->in_pragma = false; at the beginning and parser->error = false; at the end. So, perhaps add (optional?) bool argument to c_parser_skip_to_pragma_eol whether it should report the error if not already at CPP_PRAGMA_EOL (the default case), or whether it should use c_parser_next_token_is/c_parser_consume_token instead?
