https://gcc.gnu.org/bugzilla/show_bug.cgi?id=77459
--- Comment #7 from Jonathan Wakely <redi at gcc dot gnu.org> --- (In reply to François Dumont from comment #6) > Created attachment 39984 [details] > Stop using __builtin_snprintf until __has_builtin is supported. > > Could you try this patch ? Until we have __has_builtin support I propose to > simulate __builtin_snprintf with __builtin_sprint. This patch is wrong in several ways. GCC always supports __builtin_snprintf, so even if we supported __has_builtin (which is "if" not "until") the test would not be right. What happens when it's not supported by the target is a linker error, not a missing builtin. The patch adds complexity and pessimizes targets that do support snprintf.