use of memset and optimization

Sam Hartman hartmans at MIT.EDU
Fri Nov 8 15:41:00 EST 2002


So, I seem to have started this discussion off in a bad direction with
my comment about gcc.  I agree that a solution that is guaranteed to
work provided that such a solution exists is preferable to something
that happens to work.  My comment about gcc was mostly curiosity.

Would it be sufficient to have either an implementation of memset or a
function that called memset taking a pointer to volatile data as
input?


The other option seems to be to declare all key material volatile.
That might have negative performance consequences that would be nice
to avoid.




More information about the krbdev mailing list