Nagaraj Bagepalli wrote:
> I am looking at the openssl code and there is a comment in ssl3_accept
> routine just before calling buffering bio which says "Buffering bio need
> to be pushed so that the output is sent in a way that TCP likes". What
> does this comment mean, is this related to the TCP Nagle algorithm?
I haven't looked at the code you mention, but that sounds likely.
> If this is related to Nagle algorithm, I assume there is no need for
> this buffering bio if Nagle algorithm is turned off in TCP. Is my
> assumption correct?
It's probably good to turn off Nagle's algorithm for SSL links, but
if you don't buffer, you may see lots of tiny IP packets when a
single larger one could have been sent.
In my own server, I grind the state machine until it's done, and
only then do I writev() the buffer to the network. (BTW, if you
have two buffers to write to the network, using writev() rather
than two write()'s gets them to go into the same packet.)
- Dan
______________________________________________________________________
OpenSSL Project http://www.openssl.org
Development Mailing List [EMAIL PROTECTED]
Automated List Manager [EMAIL PROTECTED]
