No. Glibc should not give a tinker's cuss what's in <asm/posix_types.h>.
The fix is to Not Do That in glibc, which is what I did in current snaps.
actually, i'm running a pretty recent glibc snapshot. the problem
comes from <sys/socket.h>. it eventually includes <asm/posix_types.h>
by way of <asm/socket.h>. i suppose the "#define _LINUX_TYPES_H"
header-prevention trick can be used, but i've always thought of that
as massively hokey.
-a