From: Theo Buehler Subject: Re: opensc vs llvm22 on i386 (DES_LONG) To: Klemens Nanni , ports Date: Tue, 9 Jun 2026 05:22:19 +0200 On Mon, Jun 08, 2026 at 10:45:23PM +0100, Stuart Henderson wrote: > /usr/include/openssl/des.h: > 64 #ifndef DES_LONG > 65 /* XXX - typedef to unsigned int everywhere. */ I hope I won't forget to fix this in the next ABI break... It's been in my huge queue for far too long. > 66 #ifdef __i386__ > 67 #define DES_LONG unsigned long > 68 #else > 69 #define DES_LONG unsigned int > 70 #endif > 71 #endif > > is this an appropriate fix for now? I think it is. Arguably all 'unsigned int' in here should be DES_LONG, but there are too many, so too much churn and for the others this wouldn't actually change anything. ok While it caught many actual bugs and rightfully suggested that some software visit the attic, I don't think there's UB involved in the code as it is without patch (no coffee yet, though). This new error sure is a bit on the picky side...