Any type with a suffix _t is reserved as per POSIX but nobody really follows this so I can't blame you. Use of glibc specifics, from what I saw last time it wa

Any type with a suffix _t is reserved as per POSIX but nobody really follows this so I can't blame you. Use of glibc specifics, from what I saw last time it was things like rawmemchr in place of strlen which if you look at the generated asm rawmemchr is worse and glibc specific.I would use perf and see what the bottlenecks were I doubt its strlen or even the non portable pthread np api. The use of __linux__ when you meant __GLIBC__, you shouldn't assume Linux == glibc either. I think you removed all that though. I'll look over it again tonight.