Should we remove all internal usages of bcopy/bzero?

Roland McGrath roland@frob.com
Sun Sep 16 18:18:00 GMT 2001


> bcopy: yes.  bzero: not really.  There shouldn't be many bcopy uses
> left.  And yes, there is a reason: one less PLT entry.

Another reason is that most bcopy calls don't need to handle overlap
and so can be memcpy instead of memmove and be a tiny bit cheaper.

For bzero, I wonder if it might not be worth having optimized memset
macros check __builtin_constant_p and == 0 and refer to the bzero code.



More information about the Libc-alpha mailing list