<HTML>
<HEAD>
<TITLE>Technical Overview</TITLE>
</HEAD>
<BODY>
<!--
$Id: overview.html,v 1.22 2013-11-22 20:51:39 ca Exp $
-->
<H1>Technical Overview</H1>
<H2>Contents</H2>
<UL>
<LI><A HREF="#Initialization">Initialization</A>
<LI><A HREF="#ControlFlow">Control Flow</A>
<LI><A HREF="#Multithreading">Multithreading</A>
<LI><A HREF="#ResourceManagement">Resource Management</A>
<LI><A HREF="#SignalHandling">Signal Handling</A>
</UL>
<H2><A NAME="Initialization">Initialization</A></H2>
In addition to its own initialization,
libmilter expects a filter to initialize several parameters
before calling <A HREF="smfi_main.html">smfi_main</A>:
<UL>
<LI>The callbacks the filter wishes to be called, and the types of
message modification it intends to perform (required, see
<A HREF="smfi_register.html">smfi_register</A>).
<LI>The socket address to be used when communicating with the MTA
(required, see <A HREF="smfi_setconn.html">smfi_setconn</A>).
<LI>The number of seconds to wait for MTA connections before
timing out (optional, see
<A HREF="smfi_settimeout.html">smfi_settimeout</A>).
</UL>
<P>
If the filter fails to initialize libmilter,
or if one or more of the parameters it has passed are invalid,
a subsequent call to smfi_main will fail.
<H2><A NAME="ControlFlow">Control Flow</A></H2>
<P>
The following pseudocode describes the filtering process from the
perspective of a set of <CODE>N</CODE> MTA's,
each corresponding to a connection.
Callbacks are shown beside the processing stages in which they are invoked;
if no callbacks are defined for a particular stage,
that stage may be bypassed.
Though it is not shown,
processing may be aborted at any time during a message,
in which case the
<A HREF="xxfi_abort.html">xxfi_abort</A> callback is invoked and control
returns to <CODE>MESSAGE</CODE>.
<P>
<PRE>
For each of N connections
{
For each filter
process connection (<A HREF="xxfi_connect.html">xxfi_connect</A>)
For each filter
process helo/ehlo (<A HREF="xxfi_helo.html">xxfi_helo</A>)
MESSAGE:For each message in this connection (sequentially)
{
For each filter
process sender (<A HREF="xxfi_envfrom.html">xxfi_envfrom</A>)
For each recipient
{
For each filter
process recipient (<A HREF="xxfi_envrcpt.html">xxfi_envrcpt</A>)
}
For each filter
{
process DATA (<A HREF="xxfi_data.html">xxfi_data</A>)
For each header
process header (<A HREF="xxfi_header.html">xxfi_header</A>)
process end of headers (<A HREF="xxfi_eoh.html">xxfi_eoh</A>)
For each body block
process this body block (<A HREF="xxfi_body.html">xxfi_body</A>)
process end of message (<A HREF="xxfi_eom.html">xxfi_eom</A>)
}
}
For each filter
process end of connection (<A HREF="xxfi_close.html">xxfi_close</A>)
}
</PRE>
<P>Note: Filters are contacted in order defined in config file.</P>
<P>
To write a filter, a vendor supplies callbacks to process relevant
parts of a message transaction.
The library then controls all sequencing, threading,
and protocol exchange with the MTA.
<A HREF="#figure-3">Figure 3</A> outlines control flow for a filter
process, showing where different callbacks are invoked.
</P>
<DIV ALIGN="center"><A NAME="figure-3"></A>
<TABLE border=1 cellspacing=0 cellpadding=2 width="70%">
<TR bgcolor="#dddddd"><TH>SMTP Commands</TH><TH>Milter Callbacks</TH></TR>
<TR><TD>(open SMTP connection)</TD><TD>xxfi_connect</TD></TR>
<TR><TD>HELO ...</TD><TD>xxfi_helo</TD></TR>
<TR><TD>MAIL From: ...</TD><TD>xxfi_envfrom</TD></TR>
<TR><TD>RCPT To: ...</TD><TD>xxfi_envrcpt</TD></TR>
<TR><TD>[more RCPTs]</TD><TD>[xxfi_envrcpt]</TD></TR>
<TR><TD>DATA</TD><TD>xxfi_data</TD></TR>
<TR><TD>Header: ...</TD><TD>xxfi_header</TD></TR>
<TR><TD>[more headers]</TD><TD>[xxfi_header]</TD></TR>
<TR><TD>&nbsp;</TD><TD>xxfi_eoh</TD></TR>
<TR><TD>body... </TD><TD>xxfi_body</TD></TR>
<TR><TD>[more body...]</TD><TD>[xxfi_body]</TD></TR>
<TR><TD>.</TD><TD>xxfi_eom</TD></TR>
<TR><TD>QUIT</TD><TD>xxfi_close</TD></TR>
<TR><TD>(close SMTP connection)</TD><TD>&nbsp;</TD></TR>
</TABLE>
<B>Figure 3: Milter callbacks related to an SMTP transaction.</B>
</DIV>
<P>
Note that although only a single message is shown above, multiple
messages may be sent in a single connection.
Note also that a message or connection may be aborted by
either the remote host or the MTA
at any point during the SMTP transaction.
If this occurs during a message (between the MAIL command and the final "."),
the filter's
<A HREF="xxfi_abort.html">xxfi_abort</A> routine will be called.
<A HREF="xxfi_close.html">xxfi_close</A> is called any time the
connection closes.
<H2><A NAME="Multithreading">Multithreading</A></H2>
<P>
A single filter process may handle any number of connections
simultaneously.
All filtering callbacks must therefore be reentrant,
and use some appropriate external synchronization methods to access
global data.
Furthermore, since there is not a one-to-one correspondence
between threads and connections
(N connections mapped onto M threads, M &lt;= N),
connection-specific data must be accessed
through the handles provided by the Milter library.
The programmer cannot rely on library-supplied thread-specific data blocks
(e.g., <CODE>pthread_getspecific(3)</CODE>) to store connection-specific data.
See the API documentation for
<A HREF="smfi_setpriv.html">smfi_setpriv</A> and
<A HREF="smfi_getpriv.html">smfi_getpriv</A> for details.
<H2><A NAME="ResourceManagement">Resource Management</A></H2>
Since filters are likely to be long-lived,
and to handle many connections,
proper deallocation of per-connection resources is important.
The lifetime of a connection is bracketed by calls to the
callbacks <A HREF="xxfi_connect.html">xxfi_connect</A> and
<A HREF="xxfi_close.html">xxfi_close</A>.
Therefore connection-specific
resources (accessed via <A HREF="smfi_getpriv.html">smfi_getpriv</A>
and <A HREF="smfi_setpriv.html">smfi_setpriv</A>) may be allocated in
<A HREF="xxfi_connect.html">xxfi_connect</A>,
and should be freed in
<A HREF="xxfi_close.html">xxfi_close</A>.
For further information see
the <A HREF="api.html#conn-msg">discussion</A> of message- versus
connection-oriented routines.
In particular,
note that there is only one connection-specific data pointer per connection.
<P>
Each message is bracketed by calls to
<A HREF="xxfi_envfrom.html">xxfi_envfrom</A> and
<A HREF="xxfi_eom.html">xxfi_eom</A> (or
<A HREF="xxfi_abort.html">xxfi_abort</A>),
implying that message-specific resources can be allocated
and reclaimed in these routines.
Since the messages in a connection are processed sequentially by each filter,
there will be only one active message associated with a given
connection and filter (and connection-private data block).
These resources must still be accessed through
<A HREF="smfi_getpriv.html">smfi_getpriv</A> and
<A HREF="smfi_setpriv.html">smfi_setpriv</A>,
and must be reclaimed in
<A HREF="xxfi_abort.html">xxfi_abort</A>.
<H2><A NAME="SignalHandling">Signal Handling</A></H2>
libmilter takes care of signal handling,
the filters are not influenced directly by signals.
There are basically two types of signal handlers:
<OL>
<LI><TT>Stop</TT>: no new connections from the MTA will be accepted,
but existing connections are allowed to continue.
<LI><TT>Abort</TT>: all filters will be stopped as soon as the next
communication with the MTA happens.
</OL>
Filters are not terminated asynchronously
(except by signals that can't be caught).
In the case of <TT>Abort</TT> the
<A HREF="xxfi_abort.html">xxfi_abort</A> callback is invoked.
<HR size="1">
<FONT size="-1">
Copyright (c) 2000, 2001, 2003, 2006 Proofpoint, Inc. and its suppliers.
All rights reserved.
<BR>
By using this file, you agree to the terms and conditions set
forth in the LICENSE.
</FONT>
</BODY>
</HTML>