On Wed, 8 Apr 1998, Marcus Roberts wrote:
> > SSLeay appears not to work with GNU Libc 2.0 (2.0.6 to be exact).
> > Randomly dumps core while trying to write to BIO_err.
> > I tried to track down the cause, but no avail. All looks fine up to
> > the moment someone tries to write to the stream -> Zap.
>
> I'm not sure if this is related, but I had problems with random core
> dumps a while ago. The problem was the use of the signal() function in
> an "old-fashioned" in the get_RSAPrivate_Key function (where the signals
> were taken over to protect the typing of the password). It was
> interacting badly with my use of the sigaction() function to manipulate
> signals.
I have sigaction patches in place. You need to build with a -DSIGACTION
defined. I'm not sure how many boxes have this call but I'll hopefully get
feedback from people when 0.9.0 is released.
> (I tried to forward details of the problem to Eric and Tim but due to
> bouncing email I never did find out if it got through or not)
No, I got your origional message about the memory mapping being done via
sigaction so my use of signal was stuffing it up :-). I'm just rather slow to
respond to email because a) there is too much :-) b) there is not enough time
if I'm going to do my day job :-).
For the 0.9.0 there are quite a few patches and things that have not been put
in that perhaps I should have (an ldap module, a faster provable prime
generation routine, palm pilot patches etc) have not been fully incorperated
because unless I have time to play and test these myself, I'm not always keen
to end up with them in the base library. When I start playing with S/MIME a
bit more, and the X509v3 extensions, expect LDAP support to improve :-).
eric
+-------------------------------------------------------------------------+
| Administrative requests should be sent to [EMAIL PROTECTED] |
| List service provided by Open Software Associates, http://www.osa.com/ |
+-------------------------------------------------------------------------+
