> Here's a patch for md_rand.c to use gettimeofday() instead of time().
> gettimeofday() returns returns the time in microseconds which is likely
> more unpredictable than just seconds. It's enabled when
> GETPID_IS_MEANINGLESS is defined. Maybe a new define
> HAVE_GETTIMEOFDAY is needed instead?
time() is a POSIX function while gettimeofday() is not. Also the
gettimeofday() resolution will not be *that* much better (in fact
it need not be better at all, time_t need not be an integral type).
> This patch also adds support for mixing in all the output from kstat
> on Solaris. That includes a lot of unpredictable system statistics
> like the amount of memory in use, the number of page faults and interrupts,
> number of network transmit retries, disk cache statistics, etc. It's
> several kbytes of raw data altogether. In the patch it's enabled if KSTAT
> is defined, which can only be done currently by manually adding -DKSTAT
> to the Makefile. A better way is needed.
I don't think that all this should be in the library. Use EGD
instead.
______________________________________________________________________
OpenSSL Project http://www.openssl.org
User Support Mailing List [EMAIL PROTECTED]
Automated List Manager [EMAIL PROTECTED]
