Rajeev Chawla wrote:
>
> Hi,
>
> I am writing a non-blocking server using openssl 0.9.5.
> I noticed that this version added support for a hardware
> accelerator - compaq's atalla card. I am interested
> in using this accelerator with my non-blocking server.
> However, I see a problem - it breaks the non-blocking
> paradigm. The BN_mod_exp_atalla() in bn/mod_exp.c could
> block on ASI_RSAPrivateKeyOpFn() being performed on
> the hardware accelerator. I'd appreciate any help/suggestions
> in making this non-blocking.
This isn't really a blocking issue in the usual sense, since
unaccelerated mod_exp is inherently "blocking", the Atalla completes in
less time than all but the fastest CPUs would, and in some sense the
mod_exp is deterministic and so less worrisome than the more usual
blocks.
That said, I can see that being able to regain control during the
mod_exp would be cute, but it would need a new API, natch. If you have
any proposals in that direction, we'd be interested to hear them
(support for other accelerators is in the pipeline, btw, so it needs to
be general).
OTOH, you could just thread it.
Cheers,
Ben.
--
SECURE HOSTING AT THE BUNKER: http://www.thebunker.net/hosting.htm
http://www.apache-ssl.org/ben.html
Coming to ApacheCon Europe? http://ApacheCon.Com/
______________________________________________________________________
OpenSSL Project http://www.openssl.org
Development Mailing List [EMAIL PROTECTED]
Automated List Manager [EMAIL PROTECTED]
