summaryrefslogtreecommitdiff
path: root/doc/tex/handshake.tex
diff options
context:
space:
mode:
Diffstat (limited to 'doc/tex/handshake.tex')
-rw-r--r--doc/tex/handshake.tex40
1 files changed, 25 insertions, 15 deletions
diff --git a/doc/tex/handshake.tex b/doc/tex/handshake.tex
index 9e43ac7648..c10f63009f 100644
--- a/doc/tex/handshake.tex
+++ b/doc/tex/handshake.tex
@@ -39,6 +39,7 @@ feature of the {\bf TLS} protocol which allows a client to connect to a server,
after a successful handshake, without the expensive calculations (by using the previously
established keys). \gnutls{} supports this feature, and the
example \hyperref{resume client}{resume client (see Section }{)}{resume-example} illustrates a typical use of it (This is a modification of the simple client example).
+\par
Servers only need to use the
\hyperref{gnutls\_db\_set\_name()}{gnutls\_db\_set\_name() (see Section }{)}{gnutls_db_set_name} function if they want to use the gdbm
backend to store sessions.
@@ -53,12 +54,16 @@ The resuming capability (mostly in the server side) is one of the problems of a
implementations. The problem is that all threads must share information in
order to be able to resume sessions. The gnutls approach is, in case of a
client, to leave all the burden of resuming to the client (ie. copy and keep the
-nesessary parameters). See \hyperref{gnutls\_session\_get\_data()}
-{gnutls\_session\_get\_data() on section }{}{gnutls_session_get_data},
-\hyperref{gnutls\_session\_get\_id()}
-{gnutls\_session\_get\_id() on section }{}{gnutls_session_get_id} and
-\hyperref{gnutls\_session\_set\_data()}
-{gnutls\_session\_set\_data() on section }{}{gnutls_session_set_data}.
+nesessary parameters). See the functions:
+\begin{itemize}
+\item \hyperref{gnutls\_session\_get\_data()}
+{gnutls\_session\_get\_data() on section }{}{gnutls_session_get_data}
+\item \hyperref{gnutls\_session\_get\_id()}
+{gnutls\_session\_get\_id() on section }{}{gnutls_session_get_id}
+\item \hyperref{gnutls\_session\_set\_data()}
+{gnutls\_session\_set\_data() on section }{}{gnutls_session_set_data}
+\end{itemize}
+
\par
The server side is different.
Here the server only specifies a DB file, using
@@ -71,15 +76,20 @@ only one thread can write at a time. The current behaviour of gnutls is
not to block to wait for the DB to be ready for writing, but continue the
process normally (and do not save the parameters).
\par
- \gnutls{} also provides callback functions such as:
-\hyperref{gnutls\_db\_set\_remove\_function()}{gnutls\_db\_set\_remove\_function() (see Section }{)}
-{gnutls_db_set_remove_function},
-\hyperref{gnutls\_db\_set\_store\_function()}{gnutls\_db\_set\_store\_function() (see Section }{)}
-{gnutls_db_set_store_function}, \\
-\hyperref{gnutls\_db\_set\_retrieve\_function()}{gnutls\_db\_set\_retrieve\_function() (see Section }{)
-}{gnutls_db_set_retrieve_function} and
-\hyperref{gnutls\_db\_set\_ptr()}{gnutls\_db\_set\_ptr() (see Section }{)}
-{gnutls_db_set_ptr}.
+The default behaviour is not efficient in servers where many connections
+per second arrive. Thus
+ \gnutls{} provides the following callback functions:
+\begin{itemize}
+\item \hyperref{gnutls\_db\_set\_remove\_function()}{gnutls\_db\_set\_remove\_function() (see Section }{)}
+{gnutls_db_set_remove_function}
+\item \hyperref{gnutls\_db\_set\_store\_function()}{gnutls\_db\_set\_store\_function() (see Section }{)}
+{gnutls_db_set_store_function}
+\item \hyperref{gnutls\_db\_set\_retrieve\_function()}{gnutls\_db\_set\_retrieve\_function() (see Section }{)
+}{gnutls_db_set_retrieve_function}
+\item \hyperref{gnutls\_db\_set\_ptr()}{gnutls\_db\_set\_ptr() (see Section }{)}
+{gnutls_db_set_ptr}
+\end{itemize}
+
These callback functions are required in order to use a session
storage method, other than the default gdbm backend.
\par