>> I did have a question though.  Is there any reason to define the
>> dynamic thread locking functions?  There is a note stating:
>> "Also, dynamic locks are currently not used internally by OpenSSL,
>>  but may do so in the future."
>> Is this still accurate?  If OpenSSL does start to use them internally,
>> will the 'old-style' still be sufficient (but maybe just not optimal),
>> or will there be undefined behavior?
>>
> 
> It is a very good idea to define them.
> 
> Currently only the Chil ENGINE makes uses of these. If they aren't defined it
> wont work.
> 
> In future other functionality may either not be available or at least work in
> a sub optimal way if they aren't defined.

One thing that wasn't clear in the documentation.... I have been sharing
an SSL_CTX across threads, rather than creating a CTX on a per-thread
basis.  I am getting crashes on this, but making the CTX per-thread
appears to correct the situation.  I understand that because of
session-caching it may need to write to the CTX, but I would have
hoped there'd be some sort of locking mechanism there since you'd
probably want to resume a session from any thread (and have that
association with the same CTX).  Seems like a perfect place for dynamic
locking functions...

-Brad
______________________________________________________________________
OpenSSL Project                                 http://www.openssl.org
Development Mailing List                       [email protected]
Automated List Manager                           [EMAIL PROTECTED]

Reply via email to