File RELEASE_NOTES.2.6.1 of Package amavisd-new

                                                              June 29, 2008
amavisd-new-2.6.1 release notes


- avoid a bounce-killer's false positive when a message is multipart/mixed
  with an attached message/rfc822 (looking like a qmail or a MSN bounce)
  and having attached a message with a foreign Message-ID - by restricting
  the check to messages with an empty sender address or a 'postmaster' or
  'MAILER-DAEMON' author address;

- privileges were dropped too early when chrooting, causing chroot to fail
  (a workaround was to specify a jail directory through a command line
  option -R); reported by Helmut Schneider;

- fix unwarranted 'run_av error: Exceeded allowed time' error when using
  a virus scanned Mail::ClamAV; reported by Chaminda Indrajith;

- fix a bug in helper-progs/amavis-milter.c where atoi could be reading
  from a non-null terminated string which could result in wrong milter
  return status, or even cause a read-access violation;
  reported by Shin-ichi Nagamura;

- dsn_cutoff_level was ignored if SpamAssassin was not invoked (e.g. on
  large messages) even if recip_score_boost was nonzero, causing a DSN
  not to be suppressed for internally generated large score values;
  reported by Bernd Probst;

- add back the 'Ok, id=..., from MTA(...):' prefix to a MTA status responses
  on forwarded mail when generating own SMTP status response (it was lost
  in code transition from 2.5.4 to 2.6.0); reported by Thomas Gelf;

- replaced '-ErrFile=>*STDOUT' with '-ErrFile=>\*STDOUT' in a call to
  BerkeleyDB::Env::new in amavisd-nanny and amavisd-agent; seems it
  was failing in some setups (even though it was in accordance with
  a BerkeleyDB module documentation); reported by Leo Baltus;

- README.sql-mysql: fixed a SQL data type mismatch between (used as
  a foreign key) and msgs.sid & msgrcpt.rid; they all should be of the same
  type, either integer unsigned or bigint unsigned; a schema as published
  in README.sql-mysql could not be built because of a conflict in a data
  type; reported by Leonardo Rodrigues Magalhães and Zhang Huangbin;


- recognize an additional place-holder %P in a template used to build
  a file name in file-based quarantining, for example:

    $spam_quarantine_method = 'local:Week%P/spam/%m.gz';

  A %P is replaced by a current partition tag, which makes it easier to
  better organize a file-based quarantine by including a partition tag
  (e.g. an ISO week number) in a file name or a file path.

  For the record, here is a complete list of place-holders currently
  recognized in filename templates:
    %P  =>  $msginfo->partition_tag
    %b  =>  $msginfo->body_digest
    %m  =>  $msginfo->mail_id
    %n  =>  $msginfo->log_id
    %i  =>  iso8601 timestamp of a message reception time by amavisd
    %%  =>  %

  The following example organizes spam quarantine into weekly subdirectories:
    cd /var/virusmails
    mkdir -p W23/spam W24/spam W25/spam  ... (weeks 01..53)
    chown -R vscan:vscan W23 W24 W25     ... (weeks 01..53)
    $spam_quarantine_method = 'local:W%P/spam/%m.gz';
    $sql_partition_tag =
      sub { my($msginfo)=@_; sprintf("%02d",iso8601_week($msginfo->rx_time)) };

- add a macro %P as a synonym for a macro 'partition_tag', mainly for
  completeness with the added place-holder %P in a file name template;


- disabled a do_ascii decoder in the default @decoders list:
    #  ['asc', \&Amavis::Unpackers::do_ascii],
    #  ['uue', \&Amavis::Unpackers::do_ascii],
    #  ['hqx', \&Amavis::Unpackers::do_ascii],
    #  ['ync', \&Amavis::Unpackers::do_ascii],
  The do_ascii is invoking a module Convert::UUlib which in turn calls
  a troublesome library uulib, which has a history of security problems
  and on occasion misinterprets a text file as some encoded text, causing
  false positives (e.g. making it look like an executable);
  recent false positive on base64-decoding reported by Jeffrey Arbuckle;
  recent DoS (looping in uulib) reported by Thomas Ritterbach;

- added a rule into $map_full_type_to_short_type_re to cope with another
  example of misclassification by a file(1) utility, where a plain text
  file is considered a DOS executable:
    [qr/^DOS executable \(COM\)/ => 'asc'],  # misclassified?
  An example was provided by Leonardo Rodrigues Magalhães;

- until the issue is better understood, revert the use of 'my_require'
  and go back to the standard but less informative 'require';  some people
  were reporting problems with my_require (loading of some Perl modules can
  fail, apparently depending on a current directory where amavisd is started
  from); reports by Tuomo Soini, Max Matslofva, Bill Landry;

- use the $myproduct_name value in generated Received header field
  instead of a hard-wired 'amavisd-new'; suggested by Thomas Gelf;

- added missing required header fields to some test mail messages in a
  directory test-messages to quench down a complaint about a bad header;

- changed SQL default clauses in %sql_clause (upd_msg, sel_quar, sel_penpals)
  to always join tables using both the partition_tag and the mail_id fields,
  previously just the mail_id field was used in a join. The change has no
  particular effect (and is not really necessary) on existing 2.6.0 databases
  where a primary key is mail_id (it is just a redundant extra condition),
  but saves a day when a primary key is a composite: (partition_tag,mail_id),
  which may be a requirement of a SQL partitioning mechanism.
  Thanks to Thomas Gelf for his testing of MySQL partitioning, reporting
  deficiency in amavisd SQL schema (primary keys) which did not meet MySQL
  requirements for partitioning, and suggestions;

- an AM.PDP release request can specify an additional optional attribute:
  where a requester can supply a partition_tag value of a message to be
  released. This helps to uniquely identify a message in case where an SQL
  database did not enforce a mail_id field to be unique (as may be necessary
  with some partitioning schemes).

  If a partition_tag information is readily available to a requester, it
  is advised that the attribute is included in a request even if mail_id
  is known to be unique. This may expedite a search and provide a double
  check to a validity of a request. For backwards compatibility amavisd
  performs a query on msgs.mail_id for a partition_tag value if it is
  missing form a request, the query uses an SQL clause in a new entry
  $sql_clause{'sel_msg'}.  If exactly one record matches, then everything
  is fine, and releasing may proceed. If multiple records with the same
  mail_id exist the release request is aborted with a message asking user
  to supply a disambiguating partition_tag=xx attribute;

- a quarantine id for an SQL-quarantined message as logged in a main
  log entry is changed from:
    quarantine: aX3C4f6btXgX
    quarantine: aX3C4f6btXgX[25]
  i.e. a partition_tag in brackets is appended to mail_id.

  Correspondingly the amavisd-release is also changed to be able to parse
  'aX3C4f6btXgX[25]', splitting it into mail_id and partition_tag, and
  providing each as a separate attribute in an AM.PDP release request;

- README.sql-mysql: changed SQL datatype VARCHAR into VARBINARY for
  data fields mail_id, secret_id and quar_loc, and CHAR into BINARY for
  msgs.content and msgs.quar_type to preserve case sensitivity on string
  comparison operators; suggested by Thomas Gelf;

  The same change should eventually be done on README.sql-pg too, but as
  PostgreSQL is more picky than MySQL on matching a field data type to a
  supplied data value, the change of a data type would need to be reflected
  in SQL calls in amavisd. This will have to wait until some future version
  of amavisd-new, having to undergo more testing than I have available
  before the 2.6.1 release.

Background information on UNIQUE constraint in table SQL msgs

Amavisd does not know and need not be aware of what is a primary
key or what are UNIQUE constraints in SQL table msgs. When generating
a mail_id for a message being processed, amavisd tries to INSERT
a record with a randomly generated mail_id into table msgs (using
SQL clause in $sql_clause{'ins_msg'}). If the operation fails,
another mail_id is generated and attempt repeated, until it eventually
succeeds. Thus it depends entirely on SQL's decision whether a
particular record is allowed or would break some UNIQUE constraint.
So, by only changing a declaration on table msgs (PRIMARY KEY or
adding a CONSTRAINT), it changes what keys amavisd will be allowed
to insert and what kind of duplicates would be allowed.

Classically the msgs.mail_id is a PRIMARY KEY and as such it is unique.
This was a requirement for versions of amavisd up to and including 2.6.0.
Starting with 2.6.1 the JOINs have been tightened to include a
partition_tag besides mail_id in a relation, which makes it possible
to loosen a unique requirement on msgs.mail_id and only require a
pair (partition_tag,mail_id) to be unique. In other words, this way
the mail_id is only needed to be unique within each partition tag value.

This change allows a partitioning scheme to meet requirements on
MySQL partitioning. For non-partitioned databases the change shouldn't
make any difference, and one is free to choose between having mail_id
unique across the entire table or just within each partition_tag value.

Changing a primary key to (partition_tag,mail_id) brings consequences
to quarantining, in particular to releasing from a SQL quarantine,
where it no longer suffices to specify mail_id=xxx in AM.PDP request,
but may be necessary to specify also a partition_tag=xx to distinguish
between SQL-quarantined messages which happen to have the same mail_id.

                                                             April 23, 2008
amavisd-new-2.6.0 release notes


- integrated DKIM signing and verification; see section
  A QUICK START TO DKIM SIGNING by the end of this release note;
- loading of policy banks based on valid DKIM-signed author's address
  can be used for reliable whitelisting, for bypassing banned checks, etc.
- bounce killer feature: uses a pen pals SQL lookup to check inbound DSN;
- SQL logging and quarantining tables have a new field 'partition_tag';
- captures SpamAssassin logging, more flexibility specifying SA log areas;
- collects and logs SpamAssassin timing breakdown report (requires SA 3.3);
- releasing from a quarantine can push a released message to an attachment;
- new experimental code for abuse reporting using formats: ARF/attach/plain;
- TLS support on the SMTP client and server side;
- connection caching by a SMTP client;
- amavisd-nanny and amavisd-agent now re-open a database on amavisd restarts;
- amavisd-nanny and amavisd-agent new command line option: -c count;
- updated to support source port number in queries;
- amavisd can send queries either to or directly to p0f;


- when using SQL for logging (e.g. for a pen pals feature) or for
  quarantining, SQL tables tables maddr, msgs, msgrcpt and quarantine need
  to be extended by a new field 'partition_tag';  see below for details;

- when SQL logging (pen pals) or SQL lookups are used, one can choose a
  binary or a character data type for fields,,
  and; now may be a good opportunity to change a data type
  to binary (string of bytes);  see below for details;

- when using SQL for logging, a default for $sql_clause{'upd_msg'}
  has changed, so if a configuration file replaces this SQL clause
  by a non-default setting, it needs to be updated;

- perl module Mail::DKIM is now required when DKIM verification or signing
  is enabled or when spam checking by SpamAssassin is used and a DKIM plugin
  is enabled; a required version of this module is 0.31 (or later);

- because privileges are now dropped sooner, pid and lock files as
  generated by Net::Server can no longer be located in a directory which
  is not writable by UID under which amavisd is running (e.g. /var/run).
  A location of these files is controlled by $pid_file and $lock_file
  settings, and by default are placed in $MYHOME, which still satisfies
  the new requirement;

- white and blacklisting now takes into account both the SMTP envelope
  sender address, as well as the author address from a header section
  (address(es) in a 'From:' header field). Note that whitelisting
  based only on a sender-specified address is mostly useless nowadays.
  For a reliable whitelisting see @author_to_policy_bank_maps below,
  as well as a set of whitelisting possibilities in SpamAssassin (based
  on DKIM, SPF, or on Received header fields);

- if using custom hooks, some of the internal functions have changed,
  in particular the semantics of a method orig_header_fields - use new
  functions get_header_field() or get_header_field_body() instead;
  see updated sample code amavisd-custom.conf, and see entries labeled
  'internal' below;

- a configuration variable $append_header_fields_to_bottom is now obsolete;
  the variable is still declared for compatibility with old configuration
  files, but its value is ignored: new header fields are always prepended,
  i.e. added to the top of a header section;

- semantics of a command line option 'debug-sa' has changed due to a merge
  of SpamAssassin logging with a mainstream amavisd logging mechanism.
  A command 'amavisd debug-sa' is now equivalent to 'amavisd -d all' with
  an implied redirection of all logging to stderr. Previously it only rerouted
  SpamAssassin logging to stderr but did not affect normal amavisd logging,
  which still followed the usual $DO_SYSLOG and $LOGFILE settings.

  Also, a SpamAssassin log level 'info' is now turned on by default (as was
  previously achievable by a command line option '-d info'), and shows merged
  with a normal amavisd logging at level 1 or higher.

  The following table shows mapping of SpamAssassin log levels to amavisd
  log levels, and for completeness also shows mapping of amavisd log levels
  to syslog priorities (which has not changed since previous version):

    SA     amavisd  syslog
    -----  -------  -----------
             -3     LOG_CRIT
             -2     LOG_ERR
    error    -1     LOG_WARNING
    warn      0     LOG_NOTICE
    info      1     LOG_INFO
              2     LOG_INFO
    dbg       3     LOG_DEBUG
              4     LOG_DEBUG
              5     LOG_DEBUG

- an additional requirement for loading a policy bank 'MYUSERS' is that
  'originating' flag must be on, which typically means that mail must
  be coming from internal networks or from authenticated roaming users
  to be able to load a policy bank 'MYUSERS';


- run_av: limit the number of filenames given as arguments to a command
  line scanner to stay within a safe (POSIX) program argument space limit,
  run a command line scanner multiple times if necessary. This required
  a larger change in the program (run_av, ask_av) which is why the fix
  was listed for a long time on a TODO list and not implemented so far.
  The problem affected command line virus scanners which are unable to
  traverse a directory by themselves and need a list of filenames as
  arguments (such as KasperskyLab's aveclient and kavscanner, MkS_Vir mks,
  and CyberSoft VFind). Actual problem reported by Danny Richter;


- DKIM signing and verification - see below: A QUICK START TO DKIM SIGNING.
  Not to forget upgrading Mail::DKIM to 0.31 (or later) and adding the
  following to amavisd.conf;
    $enable_dkim_verification = 1;
    $enable_dkim_signing = 1;

- SQL tables tables maddr, msgs, msgrcpt and quarantine are extended by
  a new field 'partition_tag'. When amavisd creates new records in these
  tables, a current value of a configuration variable $sql_partition_tag
  (or its value from policy banks) is written into 'partition_tag' fields.
  An undefined value translates to 0. The 'partition_tag' field is usually
  declared in a schema as an integer, but in principle could be any data
  type, such as a string.

  A value of 'partition_tag' field may be used to speed up purging of
  old records by using partitioned tables (MySQL 5.1 +, PostgreSQL 8.1 +).
  A sensible value is a week number of a year, or some other slowly changing
  value, allowing to quickly drop old table partitions without wasting
  time on deleting individual records. Records in all tables carrying the
  'partition_tag' field are self-contained within each value of a field.
  In other words, foreign keys never reference a record in a subordinate
  table with a value of a 'partition_tag' field different from the referencing
  record. Consequently, mail addresses in table maddr are also self-contained
  within a partition tag, implying that the same mail address may appear in
  more than one maddr partition (using different 'id's), and that tables
  msgs and msgrcpt are guaranteed to reference a within their own
  partition tag. Too fine a granularity of partition tags (e.g. changing a
  value daily) wastes space in table maddr by storing multiple copies of
  the same mail address.

  The $sql_partition_tag may be a scalar (usually an integer or a string),
  or a reference to a subroutine, which will be called with an object of
  type Amavis::In::Message as argument (giving access to information about
  a message being processed), and its result will be used as a partition
  tag value. Possible/typical usage (in amavisd.conf):

    $sql_partition_tag =
      sub { my($msginfo)=@_; iso8601_week($msginfo->rx_time) };

  yields an ISO 8601 week number (1..53) corresponding to a mail reception
  timestamp in a local time zone. 

  Another possible use of 'partition_tag' field is to let a policy bank set
  its specific value (a fixed value or a subroutine) for $sql_partition_tag.
  This would allow for example labeling of SQL records for mail originating
  from inside with a different partition_tag value, compared to entries for
  incoming mail, and consequently let them be stored in a separate partition
  if desired.

  Amavisd process itself does not use the 'partition_tag' field for its
  own purposes, all records regardless of their 'partition_tag' value
  are available for example to pen pals lookups, as before. The field is
  provided only as a convenience to SQL database maintenance, and can be
  ignored by smaller sites where current practice of database maintenance
  is fast enough. If SQL partitioning is not in use (or not intended to
  be used in a near future), it is more economical to use a fixed value
  (such as 0, which is a default) for the $sql_partition_tag. Using week
  numbers as partition tags adds about 50 % to the number of records in
  table maddr, the exact number depends on retention period and a ratio
  of regular vs. infrequent mail addresses observed.

  To convert tables of an existing database, please use ALTER command.
  Here is a conversion example (MySQL or PostgreSQL, probably others):

    ALTER TABLE maddr      ADD partition_tag integer DEFAULT 0;
    ALTER TABLE msgs       ADD partition_tag integer DEFAULT 0;
    ALTER TABLE msgrcpt    ADD partition_tag integer DEFAULT 0;
    ALTER TABLE quarantine ADD partition_tag integer DEFAULT 0;

  As the is no longer guaranteed to be unique, but a pair
  of (maddr.partition_tag, is unique, the constraint and
  an associated index needs to be changed:

  => PostgreSQL:
    DROP CONSTRAINT maddr_email_key,
    ADD  CONSTRAINT maddr_email_key UNIQUE (partition_tag,email);

  => MySQL:
    DROP KEY email,
    ADD  UNIQUE KEY part_email (partition_tag,email);

  Should a need arise to revert to amavisd-new-2.5.4 while keeping the new
  partition_tag field, the 'SELECT id FROM maddr ...' may become slow due to
  dropped index on a field, which is replaced by an index on a
  pair (maddr.partition_tag, The following change to amavisd
  2.5.4 solves the problem:

@@ -901,2 +901,2 @@
     'sel_adr' =>
-      'SELECT id FROM maddr WHERE email=?',
+      'SELECT id FROM maddr WHERE partition_tag=0 AND email=?',

  The use of partitioned tables to speed up purging of old records was
  suggested by Robert Pelletier.

- when SQL logging (pen pals) or SQL lookups are used, one can choose a
  binary or a character data type for fields,,
  and; now may be a good opportunity to change a data type
  to binary (string of arbitrary bytes, no character set associated).

  Background: values of these fields come from SMTP envelope or from a
  mail header section of processed mail. Even though RFC 2821 and RFC 2822
  restrict these addresses to 7-bit ASCII, there is nothing preventing
  a malicious or misguided sender from supplying any 8-bit byte values.
  If SQL fields are declared as VARCHAR or CHAR, a character set is
  associated with data and its rules apply, e.g. control characters may
  not be permitted, or UTF-8 byte sequences are validated, or a restriction
  to codes below 128 apply. Depending on strictness of an SQL server on
  validating data, a violation of character set rules may lead to aborting
  an SQL operation and failing of mail processing. Even though new standards
  for e-mail addresses are being negotiated allowing for UTF-8 encoding, an
  actual e-mail address may still supply arbitrary bytes, which may violate
  UTF-8 byte sequence rules.

  A new configuration variable $sql_allow_8bit_address now controls how
  amavisd passes e-mail addresses to SQL.

  If a value is true, then it is expected that SQL tables will accept
  strings of arbitrary bytes for these fields, without associating a
  character set with data. No data sanitation is done by amavisd. An
  appropriate SQL data type is 'VARBINARY' or with PostgreSQL a 'BYTEA'.

  If a value of $sql_allow_8bit_address is false (which is a default for
  compatibility) then amavisd performs sanitation before passing data to SQL:
  control characters and characters with codes above 127 are converted to '?',
  which brings strings within ASCII character set restrictions. A suitable
  SQL data type is VARCHAR or CHAR. Note that some information is lost in
  this case.
  The following clauses can convert pre-2.6.0 tables into the now preferred
  and more universal form:

    ALTER TABLE users    CHANGE email email varbinary(255);
    ALTER TABLE mailaddr CHANGE email email varbinary(255);
    ALTER TABLE maddr    CHANGE email email varbinary(255);

    ALTER TABLE users    ALTER email TYPE bytea USING decode(email,'escape');
    ALTER TABLE mailaddr ALTER email TYPE bytea USING decode(email,'escape');
    ALTER TABLE maddr    ALTER email TYPE bytea USING decode(email,'escape');

  If a binary data type is chosen for these three fields, the setting
  $sql_allow_8bit_address MUST be set to true to let the amavisd program
  use the appropriate data type in SQL commands, otherwise PostgreSQL will
  complain with:
    'types bytea and character varying cannot be matched'
  when amavisd tries to execute SQL commands. MySQL is more forgiving and
  does not complain about a data type mismatch, so one may get away with a
  mismatch, although it is appropriate to eventually make it right.

  If a change of a data type of these fields is chosen while using some
  third-party management interface to SQL data set (e.g. MailZu), make sure
  the management interface supports the changed data type. This is primarily
  a concern with PostgreSQL which is more strict in requiring a match
  between field data types in tables and data in SQL clauses.

  The need for a change was pointed out by Xavier Romero, reporting that
  PostgreSQL SQL lookups with pre-2.6.0 versions of amavisd can fail when
  8-bit data appears in SMTP envelope addresses:

    lookup_sql: sql exec: err=7, 22021, DBD::Pg::st execute failed: ERROR:
      invalid byte sequence for encoding "UTF8"

- bounce killer feature: uses a pen pals SQL lookup to check inbound DSN,
  attempting to match it with a previous outbound message. If a Message-ID
  found in an attachment of the inbound DSN matches a Message-ID of a
  message previously sent from our system by a current recipient of the DSN,
  the DSN message is spared, otherwise it receives $bounce_killer_score
  spam score points (0 by default, i.e. disabled) and can be blocked as
  spam (although technically it is just a misdirected bounce, not spam).

  A received delivery status notifications is parsed looking for attached
  header section of an original message in an attempt to find a Message-ID.
  A standard DSN structure (RFC 3462, RFC 3464) is recognized, as well as
  a few nonstandard but common formats. Other automatic reports and bounces
  with unknown structure and no attached header section are ignored for
  this purpose (are subject to other regular checks). Unfortunately there
  are still many nonstandard mailers around (12+ years after DSN format
  standardization) and many ad-hoc filtering solutions which do not supply
  the essential information.

  If a Message-ID can be found in an SQL log database matching a previous
  message sent by a local user (which is now a recipient of a DSN),
  using a normal pen pals lookup (no extra SQL operations are necessary),
  or if domain part of the Message-ID is one of local domains, then the
  DSN message is considered a genuine bounce, is unaffected by this check
  and passes normally (subject to other checks).

  On the other hand, if the attached DSN header does supply a Message-ID
  but but it does not meet the above pen pals matching criteria, then it is
  assumed that the message is a backscatter to a faked address belonging
  to our local domains, and $bounce_killer_score spam score points are
  added, so the message can be treated as spam (subject to spam kill level
  and other spam settings).

  The only user-configurable setting is $bounce_killer_score (also member
  of policy banks), its default value is 0. To activate the bounce killer
  feature set the $bounce_killer_score to a positive number, e.g. 100.
  A pre-requisite is a working SQL logging database (pen pals).

  A couple of SNMP-like counters are added to facilitate assessing
  effectiveness of the feature (e.g. viewed by amavisd-agent utility):
    InMsgsBounce                   21310    333/h     9.9 % (InMsgs)
    InMsgsBounceKilled             19967    312/h    93.7 % (InMsgsBounce)
    InMsgsBounceRescuedByDomain        7      0/h     0.0 % (InMsgsBounce)
    InMsgsBounceRescuedByOriginating 242      4/h     1.1 % (InMsgsBounce)
    InMsgsBounceRescuedByPenPals      67      1/h     0.3 % (InMsgsBounce)
    InMsgsBounceUnverifiable        1027     16/h     4.8 % (InMsgsBounce)

  More information on operations can be obtained from a log, search for:
    bounce killed
    bounce rescued by penpals
    bounce rescued by domain
    bounce unverifiable

  The feature was suggested by Scott F. Crosby.

  See also, and
  a SpamAssassin man page Mail::SpamAssassin::Plugin::VBounce
  for additional ideas on fighting joe-jobbed backscatter mail.

- a new configuration variable @author_to_policy_bank_maps (also a member
  of policy banks) is a list of lookup tables (typically only a hash-type
  lookup table is used), which maps author addresses(es) (each address in
  a 'From:' header field - typically only one) in a mail header section
  to one or more policy bank names (a comma-separated list of names).

  A match can only occur if a valid DKIM author signature or a valid
  DKIM third-party signature is found, so in as much as one can trust the
  signing domain, loading of arbitrary policy banks can be safe, offering
  a flexibility of whitelisting against spam (absolute or just contributing
  score points), bypassing of checks (banned, virus, bad-header), using
  less restrictive banned rules for certain senders, by-sender routing,
  turning quarantining/archiving on/off, and other tricks offered by the
  existing policy bank loading mechanisms.

  When a message has a valid DKIM (or DomainKeys) author signature (i.e.
  when a 'From:' address matches a signing identity according to DKIM
  (RFC 4871) or DomainKeys (RFC 4870) rules), a lookup key is an unchanged
  author address and the usual lookup rules apply (README.lookups - hash

  When a valid third-party signature is found, a lookup key is extended
  by a '/@' and a lowercased signing domain, as shown in the example below.

  The semantics is very similar to a whitelist_from_dkim feature in
  SpamAssassin, but is more flexible as is allows any dynamic amavisd
  setting to be changed depending on author address, not just skipping
  of spam checks.

  A few examples of a SpamAssassin's whitelist_from_dkim (as in
  along with equivalent amavisd @author_to_policy_bank_maps entries follow.

  To whitelist any From address with a domain when a message
  has a valid author signature (i.e. a signature by the same domain):
    SA:  whitelist_from_dkim  *
    am:  '' => 'WHITELIST',
  which is equivalent to a lengthy but redundant:
    SA:  whitelist_from_dkim  *
    am:  '' => 'WHITELIST',

  Similar to above, but applies to subdomains of carrying
  a valid author signature (i.e. signature BY THE SAME SUBDOMAIN):
    SA:  whitelist_from_dkim  *@*
    am:  '' => 'WHITELIST',
  Note that in amavisd hash lookups a '' implies a parent
  domain '' too, while in SpamAssassin and in Postfix maps
  a parent domain needs its own entry if desired.

  To whitelist From addresses from subdomains of which carry
  a valid third-party signature of its parent domain:
    SA:  whitelist_from_dkim  *@*
    am: '' => 'WHITELIST',

  To whitelist any From address as long as a message has a valid DKIM
  or DomainKeys signature by, i.e. a third-party signature.
  Typical for mailing lists or discussion groups which sign postings.
    SA:  whitelist_from_dkim  *@*
    am:  './' => 'WHITELIST',

  Here is a complete example to be included in amavisd.conf:

  @author_to_policy_bank_maps = ( {
    ''               => 'WHITELIST',
    ''             => 'WHITELIST',
    ''                 => 'WHITELIST',
    ''                 => 'WHITELIST',
    ''                 => 'WHITELIST',
    ''                 => 'WHITELIST',
    ''           => 'WHITELIST',
    ''             => 'WHITELIST',  # author signatures
    './'           => 'WHITELIST',  # 3rd-party sign. by
    '' => 'WHITELIST',
    ''              => 'WHITELIST',
    ''               => 'WHITELIST',
    ''                => 'WHITELIST',
    ''               => 'WHITELIST',
    ''       => 'WHITELIST',
    ''        => 'WHITELIST',
    '' => 'WHITELIST',
    ''              => 'MILD_WHITELIST',
    ''          => 'MILD_WHITELIST',
    './'     => 'MILD_WHITELIST',
    './'      => 'MILD_WHITELIST',
    './'    => 'MILD_WHITELIST',
    './'      => 'MILD_WHITELIST',
    ''           => 'MILD_WHITELIST',
    ''              => 'MILD_WHITELIST',
    '' => 'MILD_WHITELIST',
  } );

  $policy_bank{'MILD_WHITELIST'} = {
    score_sender_maps => [ { '.' => [-1.8] } ],
  $policy_bank{'WHITELIST'} = {
    bypass_spam_checks_maps => [1],
    spam_lovers_maps => [1],
  $policy_bank{'NOVIRUSCHECK'} = {
    bypass_decode_parts => 1,
    bypass_virus_checks_maps => [1],
    virus_lovers_maps => [1],
  $policy_bank{'NOBANNEDCHECK'} = {
    bypass_banned_checks_maps => [1],
    banned_files_lovers_maps  => [1],

- smtp client connection caching is a new feature which allows smtp client
  code in amavisd to keep a SMTP session to MTA open after forwarding a
  message or a notification, so that a next mail message that needs to be
  sent by this child process can avoid re-establishing a session and the
  initial greeting/EHLO (and TLS) handshake.

  A current value of a global settings $smtp_connection_cache_enable
  controls whether a session will be retained after forwarding a message
  or not. Its default initial value is true.

  A global setting $smtp_connection_cache_on_demand controls whether amavisd
  is allowed to dynamically change the $smtp_connection_cache_enable setting
  according to its estimate of the message frequency. The heuristics is
  currently very simple: if time interval between a previous task completion
  by this child process and the arrival of a current message is 5 seconds
  or less, the $smtp_connection_cache_enable is turned on (which will affect
  the next message); if the interval is 15 seconds or more, it is turned off.
  The default value of the $smtp_connection_cache_on_demand is true, thus
  enabling the adaptive behaviour.

  On a busy server the connection caching can save some processing time.
  Savings are substantial if client-side TLS is enabled, otherwise just a
  few milliseconds are saved. On an idle server the feature may unnecessarily
  keep sessions to MTA open (until MTA times them out), so one can disable
  the feature by setting both controls to false (to 0 or undef).

  To monitor the connection caching effectiveness, some SNMP-like counters
  were added, so amavisd-agent may display something like:

    OutConnNew                      2764    319/h    98.2 % (OutMsgs)
    OutConnQuit                     2521    291/h    89.5 % (OutMsgs)
    OutConnReuseFail                   7      1/h     0.2 % (OutMsgs)
    OutConnReuseRecent                21      2/h     0.7 % (OutMsgs)
    OutConnReuseRefreshed             31      4/h     1.1 % (OutMsgs)
    OutConnTransact                 2816    325/h   100.0 % (OutMsgs)

- client-side TLS support is added, i.e. on forwarding a passed mail back
  to MTA. Currently only encryption is supported, no client certificates
  are offered. A $tls_security_level_out is a per-policy-bank setting which
  controls client-side TLS, its value is either undefined (default), or a

    undef ... client-side TLS is disabled (a default setting);
    'may' ... TLS is used if MTA offers a STARTTLS capability (RFC 3207),
              otherwise a plain text SMTP session is established;
    'encrypt' TLS is used if MTA offers a STARTTLS capability, otherwise
              amavisd refuses to forward a message.

  The client-side TLS imposes some performance penalty on passing a message
  back to MTA, although it is still reasonably fast: a benchmark indicates
  a drop in transfer rate by about a factor of 2, from 22 MB/s (no TLS)
  to 9 MB/s (with TLS). The smtp client connection caching (see previous item)
  should preferably be left enabled (permanently or opportunistically), as
  TLS negotiation adds significantly to the initial SMTP handshake time.

- server-side TLS support is added, i.e. on accepting mail from MTA.
  Encryption is supported, server (i.e. amavisd) offers its certificate,
  but client certificates are not verified. A $tls_security_level_in is
  a per-policy-bank setting which controls server-side TLS, its value
  is either undefined (default), or a string:

    undef ...  server-side TLS is disabled, STARTTLS capability is
               not offered;
    'may' ...  STARTTLS capability is offered by amavisd, but client is
               not required to enter TLS, plain text sessions are permitted;
    'encrypt'  STARTTLS capability is offered and enforced by amavisd,
               any SMTP command other than STARTTLS, NOOP, EHLO or QUIT
               is rejected.

  If $tls_security_level_in is enabled (any value other than undef or 'none'),
  amavisd offers a certificate to a connecting client requesting TLS, so a
  path to a certificate and to its private key must be provided through two
  global settings: $smtpd_tls_cert_file and $smtpd_tls_key_file, e.g.:

    $smtpd_tls_cert_file = "$MYHOME/cert/amavisd-cert.pem";
    $smtpd_tls_key_file  = "$MYHOME/cert/amavisd-key.pem";

  The private key should be guarded as secret (not world-readable).
  A self-signed certificate is acceptable by most mailers.

  Server-side TLS imposes a significant performance penalty on accepting
  a message from MTA. A benchmark indicates a drop in transfer rate by a
  factor of 10, from about 10 MB/s (no TLS) to 1 MB/s (using TLS), so it
  should only be enabled with a good reason or for experimentation.

- enhanced a subroutine delivery_status_notification (along with
  dispatch_from_quarantine and msg_from_quarantine) to produce a message
  in one of several formats (derived from a message being processed, or
  from a quarantined message). Its new arguments can be strings as follows:
    $request_type:  dsn, release, requeue, report
    $msg_format:    dsn, arf, attach, plain, resend
    $feedback_type: abuse, fraud, miscategorized, not-spam, opt-out,
                    opt-out-list, virus, other (according to ARF draft)

  Per-policy settings $report_format and $release_format control the format
  of a generated message. Their value can be one of the following strings,
  although not all combinations make sense:

   'arf' .... an abuse report is generated according to
              draft-shafranovich-feedback-report-04: "An Extensible
              Format for Email Feedback Reports"; a plain-text part
              contains text from a template;
   'attach'.. generates a report message as plain text according to
              a template, with an original message attached;
   'plain'... generates a simple (flat) mail with an only MIME part
              containing a text from a template, followed inline by
              original message (some service providers can't handle
              abuse reports with attachments, e.g. Yahoo!);
   'resend'.. original message is forwarded unchanged, except for an
              addition of header fields Resent-From, Resent-Sender,
              Resent-To, Resent-Date and Resent-Message-ID;
   'dsn' .... (for internal use) a delivery status notification is
              generated according to rfc3462, rfc3464 and rfc3461;

  When a request_type is 'release' or 'requeue', the format of a generated
  message is governed by a per-policy setting $release_format according to
  the table above. Only the 'attach', 'plain' and 'resend' values are useful.
  A default setting is:
    $release_format = 'resend';  # with alternatives: attach, plain, resend
  A plain-text part (if used) is taken from a $notify_release_templ template
  and a sending address is obtained from %hdrfrom_notify_release_by_ccat.

  When a request_type is 'report', the format of a generated message is
  governed by a per-policy setting $report_format according to the table
  above. Only the following settings are useful: arf, attach, plain, resend.
  A default setting is:
    $report_format = 'arf';      # alternatives: arf, attach, plain, resend
  A plain-text part (if used) is obtained from a $notify_report_templ
  template, and a sending address from %hdrfrom_notify_report_by_ccat.

  It is possible to automatically generate abuse reports from custom hooks
  by calling delivery_status_notification() and mail_dispatch(). Extreme
  care must be taken to only produce legitimate abuse reports (about true
  fraud and true spam), sent only to parties that are truly responsible for
  a message being reported. Non-repudiation is a key factor here - trust
  only header fields covered by a valid DKIM signature, or generated by
  your own MX MTA (such as an IP address of the last hop), and only report
  messages received from a network which officially belongs to the party
  (according to whois). Rate-limiting should be used, and abuse reports on
  the same abuser should only be sent once in a time interval of several
  hours. A SQL database can be used to maintain a list of recently reported
  abusers, thus preventing excessive reports.

- introduced a variation of a message release from a quarantine, allowing
  a releaser to send an abuse report based on a quarantined message.
  It is implemented by:

  * enhancing a subroutine delivery_status_notification as described
    in the previous item;

  * extending AM.PDP protocol with a 'request=report' attribute
    which can be used in place of a 'request=release',

  * enhancing the 'amavisd-release' utility program to allow sending an
    attribute 'request=release' or 'request=requeue' or 'request=report'
    based on its program name. By making a soft or hard link named
    'amavisd-report' linking to 'amavisd-release', the utility will
    send a 'request=report' in place of the usual 'request=release',
      # ln -s amavisd-release amavisd-report
      # ln -s amavisd-release amavisd-requeue
      $ amavisd-report spam/k/kg2P0rP9Lpu3.gz ''

- releasing from a quarantine can push a released message to an attachment
  (Content-Type: message/rfc822), with a configurable template for a header
  section and the plain-text part; select by: $release_format='attach';
  suggested by Patrick Ben Koetter;

- detect and save a new attribute SOURCE from an XFORWARD smtp command;
  the value is also accepted as AM.PDP protocol attribute 'client_source'.
  Possible values are: 'LOCAL', 'REMOTE', or '[UNAVAILABLE]', the information
  corresponds to 'local_header_rewrite_clients' postfix setting and is
  not supposed to be used for security decisions according to Postfix
  documentation (which makes it less interesting for our purposes);

- added client and server support for a PORT attribute of an XFORWARD command,
  allowing MTA to pass a TCP port number of a remote client to a content
  filter (and back if necessary); the PORT attribute is made available
  with Postfix version 2.5 (20071004); a source port number is also
  accepted as an AM.PDP protocol attribute 'client_port';

- updated now supports a source port number information in
  queries while preserving backwards compatibility with previous versions
  of amavisd-new. Version 2.6.0 of amavisd requires a new version of (supplied in the 2.6.0 distribution) if operating system
  fingerprinting is enabled. A source port number information in a query
  allows to locate a matching entry in its cache faster and
  also more accurately when multiple connections are present from clients
  behind NAT using the same IP address. The source port number is made
  available to a content filter since Postfix version 2.5 (20071004);

- besides the ability to send queries to, amavisd now also
  supports sending queries directly to a p0f program over a Unix socket
  using a p0f query protocol. There is a bug in p0f-2.0.8 (and probably in
  earlier versions) which makes it send back incorrect results at times, i.e.
  results belonging to some other unrelated session, so a patch to p0f-2.0.8
  MUST be applied in order to use a direct querying mechanism - author has
  been notified. The patch is supplied: p0f-patch.

  There are currently no advantages (and some disadvantages) in choosing
  direct queries to p0f, compared to sending queries to,
  so this new method is not currently recommended. Disadvantages are:
  * p0f uses a linear search over its list of recent sessions, whereas uses a fast hash lookup method;
  * p0f keeps a relatively small list of recent sessions which is limited
    by the number of slots (size can be specified on a command line, but
    is limited by a linear search time), whereas expires
    old entries according to time since entered and is thus independent
    of a current mail rate;
  * a direct p0f query protocol uses packed binary data and its on-the-wire
    representation may depend on a compiler used, so it may be incompatible
    with queries sent by amavisd, whereas the queries and
    replies use a more environment-independent textual representation.

  To let amavisd sent queries directly to p0f, specify a p0f socket path:
    $os_fingerprint_method = 'p0f:/var/amavis/home/p0f.sock';
  and specify an IP address and a port number on which MTA is listening:
    $os_fingerprint_dst_ip_and_port = '[]:25';
  because p0f requires this information in a query and the information
  is not made available to a content filter via XFORWARD command
  (the does not need this information).

  To send queries to (traditional and recommended), use:
    $os_fingerprint_method = 'p0f:';
  as before. The $os_fingerprint_dst_ip_and_port in this case is not needed
  and is ignored.

- usually a sending address in spam messages is faked and it is desirable
  to suppress most if not all bounces by keeping $sa_dsn_cutoff_level low,
  but sometimes it may be possible to be more certain of the validity of
  a sending address, and when such mail is considered spam, it may still be
  desirable to send a non-delivery notification, knowing that a notification
  will most likely be addressed to a genuine sender.

  Two new settings are provided for this purpose:
    @spam_crediblefrom_dsn_cutoff_level_bysender_maps and
  (with their default being $sa_crediblefrom_dsn_cutoff_level),
  complementing the existing @spam_dsn_cutoff_level_bysender_maps and

  It is expected that $sa_crediblefrom_dsn_cutoff_level would be set somewhat
  higher than $sa_dsn_cutoff_level, allowing for more bounces to be generated
  for spam from likely-to-be-genuine senders (possibly false positives).

  The choice between taking a cutoff value from one or the other pair of
  settings depends on an attribute $msginfo->sender_credible - when it is
  true (e.g. some nonempty string) the *spam_crediblefrom_* settings will
  be used instead of the baseline @spam_dsn_cutoff_level_*maps.

  An initial value of a sender_credible attribute as provided by amavisd
  is true if either the 'originating' flag is true (e.g. mail from inside),
  or if dkim_envsender_sig attribute is true, e.g. a domain of a valid
  DKIM signature matches envelope sender address, otherwise it is false.
  A user-provided custom hook code is free to change the value of
  sender_credible attribute. An exact value does not matter (it is only
  interpreted as a boolean), but serves for logging purposes. Heuristics
  may be based on some tests provided by SpamAssassin, on DKIM signatures,
  on p0f results, on policy banks, etc.

  Here is one complete example of a custom hook, which turns on the
  sender_credible attribute based on some criteria. Note that some of
  the referenced SpamAssassin tests may not yet be available in the
  last officially released version of SpamAssassin.

  added to amavisd.conf:

  /etc/amavisd-custom.conf :

    package Amavis::Custom;
    use strict;
    sub new { my($class,$conn,$msginfo) = @_; bless {}, $class }
    sub after_send {
      my($self,$conn,$msginfo) = @_;
      if ($msginfo->sender ne '') {
        my(@cred); local($1);
        my($tests) = $msginfo->supplementary_info('TESTS');
        $tests = ''  if !defined($tests) || $tests eq 'none';
        push(@cred,'orig')   if $msginfo->originating;
        push(@cred,$1)       if $tests =~ /\b(RCVD_IN_DNSWL_HI)\b/;
        push(@cred,$1)       if $tests =~ /\b(RCVD_IN_DNSWL_MED)\b/;
        push(@cred,$1)       if $tests =~ /\b(RP_MATCHES_RCVD)\b/;
        my($os_fingerprint) = $msginfo->client_os_fingerprint;
        if ($os_fingerprint !~ /^Windows XP(?![^(]*\b2000 SP)/) {
          push(@cred,'dkim') if $msginfo->dkim_envsender_sig;
          push(@cred,$1)     if $tests =~ /\b(SPF_PASS)\b/;
        $msginfo->sender_credible(join(",",@cred))  if @cred;
    1;  # insure a defined return

- a new setting $reputation_factor (also a member of policy banks) with
  a value between 0 and 1 (default 0.2), controlling an amount of 'bending'
  of a calculated spam score towards a fixed score assigned to a signer
  identity (i.e. its 'reputation') through @signer_reputation_maps;
  the formula is: adjusted_spam_score = f*reputation + (1-f)*spam_score;
  which has the same semantics as auto_whitelist_factor in SpamAssassin AWL;

- a new setting @signer_reputation_maps (also a member of policy banks) may
  contain a list of lookup tables (typically just one hash lookup table),
  mapping a signing identity to a score, which is typically a long term
  average spam score of all messages signed by this signing identity.
  Based on a lookup result and a formula given above ($reputation_factor),
  the resulting value (positive or negative) is added to the spam score.

  Here is an example setting:

@signer_reputation_maps = (
{ ''                  => -10.95,
  ''                  =>  -9.57,
  ''               =>  -8.59,
  ''                 =>  -8.03,
  ''                  =>  -3.59,
  ''                  =>  -3.38,
  ''          =>  -4.57,
  ''           =>  -3.20,
  ''               =>  -6.66,
  ''          =>  -3.70,
  ''       =>  -0.67,
  ''  =>   1.35,

  ''                 =>  -7.89,
  ''            =>  -7.51,
  ''                   =>  -4.44,
  ''            =>  -5.44,
  ''                =>  -4.03,
  ''             =>  -3.03,
  ''             =>  -2.99,
  ''            =>  -2.80,
  ''                    =>  -2.54,
  ''            =>  -2.06,
  ''          =>  -2.01,
  ''                  =>  -1.69,

  ''          =>  -9.85,
  ''               =>  -9.31,
  ''             =>  -9.18,
  ''            =>  -9.05,
  ''           =>  -9.04,
  ''                  =>  -8.96,
  ''               =>  -8.84,
  ''              =>  -8.81,
  ''               =>  -8.81,
  ''            =>  -8.75,
  ''               =>  -7.99,
  ''              =>  -7.58,
  ''         =>  -6.93,
  ''           =>  -5.39,
  ''                 =>  -4.93,
  ''              =>  -4.91,
  ''           =>  -2.16,
  ''              =>   1.06,

  ''        =>  -9.11,
  ''          =>  -8.64,
  ''           =>  -8.32,
  ''         =>  -8.15,
  ''             =>  -5.14,
  ''                =>  -4.95,
  ''       =>  -3.77,
  ''                 =>  -2.16,
  ''               =>   0.09,
  ''              =>  -0.63,
  ''                =>  -1.50,
  ''        =>  -0.34,
  ''=> -0.74,
  ''              =>   1.11,
  ''    =>   1.98,
  ''      =>   2.26,
  ''=>  3.02,
  ''=> 3.73,
  ''            =>   0.51,
  ''     =>   2.04,
  ''                =>   3.51,
  ''             =>   4.12,
  ''                  =>   4.81,
  ''           =>   2.46,
  ''           =>   3.48,
  ''        =>   4.58,
  ''   =>   2.89,

  ''                =>  -0.53,
  ''                 =>  -1.48,
  ''                 =>  -1.33,
  ''              =>   0.85,
  ''                 =>   1.13,
  ''                 =>   1.22,
  ''                 =>   2.04,
  ''                 =>   4.48,
  ''                 =>   6.35,
  ''                 =>   7.09,
  ''                 =>   7.10,
  ''              =>   5.34,
  ''             =>   4.34,
  ''             =>   5.48,
  ''              =>   9.25,
  ''             =>   9.30,
  ''             =>  13.97,
  ''                 =>  21.94,
  ''            =>  -1.57,
  ''          =>  -1.76,
# ''        =>   0.10,
# ''          =>   5.99,
  ''           =>  19.96,
  ''       =>  17.41,
  ''                =>  -3.52,
  ''         =>  -3.10,
  ''           =>   3.42,
  ''               =>   1.47,
  ''              =>  -8.91,
  ''                 =>  -2.38,
  ''            =>  -2.27,
  ''           =>  -1.96,
  ''              =>  -0.96,
  ''               =>   2.05,
  ''                =>   2.40,
  ''             =>  -9.80,
  ''            =>  -4.70,
  ''       =>   3.38,
  ''                   =>  -4.57,
  ''             =>  -4.48,
  ''        =>  -4.32,
  ''                 =>  -3.50,
  ''=> -2.58,
  '' =>  -1.97,
  ''                => -11.67,
  ''               =>  -1.94,
  ''         =>  -1.92,
  ''           =>  -1.91,
  ''              =>  -0.81,
  ''                =>  -0.10,
  ''  =>   3.58,
  ''               =>  10.53,
  ''     =>  12.19,
  ''   =>  12.41,
  ''    =>  12.71,
  ''         =>  12.88,
  ''    =>  13.03,
  ''           =>  13.82,
  ''            =>  15.59,
  ''      =>  19.46,
  ''         =>  19.49,
  ''            =>  19.71,
  ''          =>  19.75,
  ''            =>  20.01,
  ''            =>  20.02,
  ''        =>  20.19,
  ''            =>  20.27,
  ''          =>  20.32,
  ''          =>  20.84,
  ''               =>  20.95,
  ''          =>  20.96,
  ''             =>  20.96,
  ''            =>  21.57,
  ''            =>  21.68,
  ''            =>  21.71,
  ''  =>  23.16,
  ''     =>  23.69,
  ''      =>  24.00,
  ''  =>  24.69,
  ''=> 26.39,
  ''          =>  26.65,
  ''              =>  28.75,
  ''             =>  29.08,
  ''        =>  29.24,
  ''        =>  13.06,
  ''     =>  13.35,
  ''        =>  13.40,
  ''   =>  13.49,
  ''      =>  14.13,
  ''      =>  14.17,
  ''    =>  14.33,
  ''       =>  14.58,
  ''     =>  14.61,
  ''    =>  14.72,
  ''      =>  15.23,
  ''       =>  15.96,
  ''   =>  15.98,
  ''      =>  16.12,
  ''        =>  16.83,
  ''    =>  17.14,
  ''         =>  17.30,
  ''       =>  18.46,
  ''         =>  23.36,
  ''       =>  24.50,
  ''      =>  24.62,
  ''    =>  24.63,
  ''      =>  25.71,
  ''     =>  25.98,
  ''   =>  26.14,
  ''        =>  26.85,
  ''          =>  23.39,
  ''         =>  25.82,
  ''           =>  26.02,
  ''          =>  26.41,
  ''        =>  21.96,
  ''          =>  23.27,
  ''            =>  27.41,
  ''        =>  29.91,
  ''       =>  22.71,
  ''   =>  24.77,
  ''      =>  26.84,
  ''     =>  26.87,
  ''      =>  30.04,
  ''         =>  31.48,
  ''        =>  32.46,
  ''       =>  34.11,
  ''      =>  31.12,
  ''          =>  33.05,
  ''        =>  34.95,

- drop privileges sooner if possible - right after reading config file and
  before forking;

- allow inserting X-Quarantine-ID header field into passed (and quarantined)
  mail for local recipients only; remote recipients should not be made aware
  that we may have a copy in a quarantine; reported by Robert Fitzpatrick;

- check for multiple occurrences of RFC 2045 and RFC 2183 MIME header fields
  (in addition to checks on RFC 2822 header fields):
    MIME-Version Content-Type Content-Transfer-Encoding
    Content-ID Content-Description Content-Disposition
  and the RFC 3834 header field: Auto-Submitted

- capture SpamAssassin logging and integrate it into the usual amavisd log;
  suggested by Jeff Moss;

- when parsing SpamAssassin log areas/facilities, recognize negations
  (a name prefixed by 'no') and remove duplicates, last entry wins, e.g.
    amavisd -d rules,noall,norules,dcc,norules,rules debug-sa
  is equivalent to:
    amavisd -d noall,dcc,rules debug-sa

- a SpamAssassin debug level 'info' is now implicitly prepend to a list of
  SpamAssassin facilities; it may be overridden by explicitly negating it,
  e.g. 'amavisd -d noinfo';

- when a command line argument 'debug-sa' is present, or if $sa_debug is
  true, a SpamAssassin debug level 'all' is prepend to a list of SpamAssassin
  facilities, which would bring SpamAssassin log level to 'dbg';  it may be
  overridden by negating it, e.g. 'amavisd -d noall,plugins,rules debug-sa';

- include milliseconds in a log timestamp when logging is directed to stderr;

- create a custom hook object sooner, so that loading a policy bank from a
  custom hook becomes useful (but not soon enough to influence partition_tag);

- added a custom hook after_send, which may suppress DSN, send reports,
  quarantine, etc;

- fetch additional information (tag) 'TIMING' from SpamAssassin, making it
  available through macro 'supplementary_info' (if a version of SpamAssassin
  in use provides it - available since 3.3.0);

- a SpamAssassin TIMING-SA report (timing breakdown by sections) is now
  collected and logged at log level 2 when available; the information is
  available since version 3.3.0 of SpamAssassin (currently in development
  and available through SVN);

- add a global configuration variable $listen_queue_size (undef by default)
  which is passed as an option 'listen' to Net::Server, which in turn passes
  it on to listen(2) as a 'backlog' parameter. The Net::Server provides
  a default value of SOMAXCONN in the absence of a valid integer in
  $listen_queue_size (e.g. 128, but on Solaris it defaults to 5, which
  is too small for some purposes). Suggested by David Schweikert.
  A workaround for a small SOMAXCONN default on Solaris is provided
  by Net::Server 0.98 (?).

- in the absence of a smtp client's IP address (normally received by XFORWARD
  smtp command from Postfix, or in the 'client_address' attribute of AM.PDP),
  parse the topmost one or two Received header fields and use the first
  valid IP address found there; based on a suggestion by Richard Bishop;

- new macros (useful in notifications and $log_template):
  - 'week_iso8601'  returns an ISO 8601 week number (between 1 and 53);
  - 'partition_tag' returns a current value of a $sql_partition_tag variable;
  - 'dkim'          reports various DKIM verification results;
  - 'report_format' gives one of the: dsn, arf, attach, plain, resend
  - 'feedback_type' is expected to give one of the ARF (draft) strings:
                    abuse, fraud, miscategorized, not-spam, opt-out,
                    opt-out-list, virus, other
  - 'rfc2822_from'  an e-mail address from a From header field;
  - 'rfc2822_sender' an e-mail address from a Sender header field, or empty;
  - 'tls_in'        returns TLS ciphers in use by a SMTP session if mail came
                    to amavisd through a TLS-encrypted session, otherwise empty
  - 'limit'         takes two arguments: a string size limit and some string,
                    returning the string from the second argument unchanged
                    if its size is below the limit, or cropped to the size
                    limit, with '[...]' appended;
  For details see README.customize.

- new configuration setting $allow_fixing_long_header_lines, also member
  of policy banks, defaults to true - provides control over truncation of
  header section lines longer than 998 characters as limited by RFC 2822.
  The $allow_fixing_improper_header must also be true for fixing to take
  place. Previously it was only possible to turn off all header fixes,
  but not specifically just the long header truncation;

- strip X-Spam-* headers and other prepended header fields when releasing
  a quarantined message;

- turn on message part attributes 'C' (crypted) and 'U' (undecipherable)
  if a MIME Content-Type of that body part is <anything>/encrypted;
  (note: if $defang_undecipherable is turned on, this would push a
  received PGP/GPG-encrypted MIME top-part into an attachment, just
  as with other password-protected archives);

- fetch additional information (tags) DCCB and DCCR from a SpamAssassin
  DCC plugin, making them available through a macro 'supplementary_info';

- amavisd-nanny and amavisd-agent now reopen a database if/when the underlying
  database file is re-created (i.e. when its inode changes), as is the case
  when amavisd restarts; based on a patch by Rob Foehl;

- amavisd-nanny and amavisd-agent got a new command line option: -c count,
  which can restrict the number of iterations for display;
  usability deficiency pointed out by John Evans;

- amavisd-nanny and amavisd-agent will use a value from an environment
  variable AMAVISD_DB_HOME (if it exists) as a database home directory,
  otherwise fall back to a built-in default '/var/amavis/db';
  suggested by Leo Baltus;


- added an AV entry for a new version of a ESET File Security (for Linux and
  FreeBSD) command line scanner (ESET Software ESETS Command Line Interface
  v 2.71.12); the nod32cli utility has been replaced by esets_cli;
  update provided by Willi Gruber;

- keep time of a message reception (or creation) in rx_time as a floating
  point value as provided by Time::HiRes::time in order to avoid discrepancy
  in SNMP-like counters showing elapsed times; suggested by David Schweikert;

- store additional attributes in per-message and per-recipient objects,
  reducing a need for repeated lookups;

- to save time the: tag, tag2, tag3 and kill levels are not looked up when
  a recipient is bypassing spam checks;

- no longer reverse-resolve an IP address obtained from a Received header
  field on infected mail: it does not work for IPv6, it can be stuck for
  prolonged periods waiting for a response from a non-responsive DNS server,
  and is only used for logging/notifications (macros %e and %o) on infected
  mail, so it is not too bad to drop it;

- obey the always_bcc_by_ccat even when releasing from a quarantine;
  thanks to Tomas Horacek;

- for consistency with other quarantine methods: store recipient address
  (e.g. a quarantine mailbox e-mail address) as a quarantine location
  even if a quarantine method is 'smtp:' or 'lmtp:';

- a quarantined message now receives one additional header field: an
  X-Envelope-To-Blocked. An X-Envelope-To still holds a complete list
  of envelope recipients, but the X-Envelope-To-Blocked only lists its
  subset (in the same order), where only those recipients are listed
  which did not receive a message (e.g. being blocked by virus/spam/
  banning... rules). This facilitates a release of a multi-recipient
  message from a quarantine in case where some recipients had a message
  delivered (e.g. spam lovers) and some had it blocked;

- a release request now takes its default recipients list from a header
  field X-Envelope-To-Blocked, no longer from X-Envelope-To. This avoids
  releasing a message to recipients which have already received it in
  the first place, e.g. spam lovers. For backwards compatibility, if
  X-Envelope-To-Blocked header field is not found in a quarantined
  message, the recipients list defaults to X-Envelope-To as before.
  A release request can still provide its explicit list of recipients
  to override a default, like before. Loosely based on suggestions
  by Christer, by Paolo Schiro and others;

- when mail is received from a helper program such as a milter, update
  a true mail size ($msginfo->msg_size) according to size definition
  in RFC 1870, adjusting for CRLF line terminators;

- updated SMTP enhanced status response codes to an AUTH command
  according to RFC 4954 (SMTP Service Extension for Authentication);

- quote_rfc2821_local: within qcontent all non-qtext characters should be
  represented as quoted-pairs (RFC 2822); in addition to " and \ (which were
  already handled), this includes NUL, CR, LF, HT, SP and 8-bit characters;

- macros T, C and B now bring e-mail addresses in quoted form and in angle
  brackets, notification templates no longer add angle brackets around
  %T and %C;

- when defanging mail body no longer insert our own Sender header
  field on a pretense that it helps with DKIM resigning - according
  to ADSP/SSP drafts the DKIM/ADSP does not care for the Sender
  header field (unlike a historical DomainKeys);

- always provide X-Amavis-PolicyBank header field in a copy of a mail as
  submitted to SpamAssassin, even if a policy bank path is empty - this
  allows for simpler SpamAssassin rules to avoid being tricked by a
  presence of such header field inserted by third parties;

- add missing installs of amavisd-nanny and amavisd-agent utilities
  to amavisd-new.spec; suggested by Eddy Beliveau;

- internal: original header section is no longer kept in two copies (in
  orig_header and in orig_header_fields). Instead, orig_header now holds
  an array of header *fields* (previously: array of header *lines*), and
  orig_header_fields now stores a by- header-field-name hash of indices
  into orig_header (previously it held copies of header field bodies).
  To facilitate access to individual header fields and allow for top-down
  and bottom-up search for a n-th occurrence of a header field, two new
  access methods are provided: get_header_field and get_header_field_body;
  these are optimized for a quick access to the *last* header field
  (previously the *first* one was kept easily accessible);

- internal: when specific header fields are looked up in an original header
  section (such as From, To, Subject, Message-ID, ...) now the *last* (the
  most to the bottom) occurrence of a header field (instead of the first)
  is used for better compatibility with DKIM which searches for header
  fields bottom-up;

- internal: when a message is received, a current setting of a boolean
  attribute $originating (global or from current policy banks) is now copied
  to an Amavis::In::Message object and becomes a property of a message;
  this allows for other message objects (like notifications, quarantine)
  to have their own individual setting of this attribute, for example
  notifications are always flagged as originating and as such are eligible
  for DKIM signing;

- internal: provide an Amavis::MIME::Body::OnOpenFh package which acts as
  a MIME::Body -type object (read-only) with an underlying representation
  in an existing (permanently open) temporary mail file, avoiding a need to
  open it by file name on a separate file handle.  It is useful (simpler,
  faster) when defanging (pushing original mail to an attachment), or when
  generating notifications or reports which contain an original mail;

- internal: new subroutines init_child and rundown_child in the interface
  module Amavis::SpamControl::SpamAssassin, which call SpamAssassin plugin
  methods "spamd_child_init" and "spamd_child_post_connection_close";
  this is required for correct operation of some SpamAssassin plugins
  such as the Mail::SpamAssassin::Plugin::DBI; thanks to Michael Scheidell
  for pointing out that plugin;

- internal: renamed a subroutine string_to_mime_entity into build_mime_entity
  and generalize its functionality to make it useful for generating reports;

- internal: generalized subroutine delivery_status_notification which can
  now also prepare message in a format of a feedback report (ARF) or just
  as a plain included (inline) original mail;

- internal: a subroutine passed to edit_header() may return undef, in which
  case a header field will be deleted instead of being replaced/edited;

- internal: sub lookup_sql incompatible change to the order of arguments;

- internal: a small speedup in receiving mail contents over SMTP or LMTP;

- internal: abandon package Amavis::Lock, do the four calls to flock directly;

- internal: move subroutines dealing with processes from module Amavis::Util
  into a new module Amavis::ProcControl;

- internal: reduce buffer sizes in some copying loops from 64kB back to 32kB,
  larger buffers do not perform any better;

- documentation: amavisd.conf-default: add a list of legacy dynamic
  configuration variables which can be used in policy banks;
  thanks to Gary V;

- terminology: use a term 'header section' (in comments, log entries and
  templates) instead of a 'header' to go along with the 2822upd draft;


Starting with 2.6.0, verification of DKIM signatures (and historical
DomainKeys signatures) is provided directly by amavisd (not only by a
SpamAssassin plugin DKIM). A required version of a perl module Mail::DKIM
is 0.31 or later. Signature verification is sufficiently fast so there
is no need for concern about extra processing load. To turn on DKIM (and
historical DomainKeys) signature verification, please add the following
line to amavisd.conf (if not already there):

  $enable_dkim_verification = 1;


- Whitelisting of banned checks or spam on messages carrying valid DKIM
  or DomainKeys signatures from trustworthy signers is possible through
  the @author_to_policy_bank_maps list of lookup tables. The mechanism
  uses loading of policy banks based on author's e-mail address (addresses
  in a 'From:' header field) and a signing domain, so a full flexibility
  of per-policy-bank settings is available. See description of a new
  configuration variable @author_to_policy_bank_maps earlier in this
  release notes.

- To each message passed to local recipients amavisd inserts a header field
  Authentication-Results (according to draft-kucherawy-sender-auth-header)
  for each signature found in a message, reporting a verification result.
  These header fields can reliably tell a recipient or his MUA what domains
  claimed responsibility for a message, or can be used for troubleshooting
  DKIM signing, verification and tracking mail transformations.

- Can adjust spam score based on some metrics on a signing domain's reputation
  for valid signatures found in a message. A useful reputation metric is
  an average long term spam score for past messages signed by a domain,
  which can currently be provided manually by @signer_reputation_maps in a
  configuration file (see example earlier in this release notes). A spam
  score is shifted towards this reputation score by a configurable factor
  $reputation_factor (value between 0 and 1, default is 0.2) using a formula:
    adjusted_spam_score = f * reputation + (1-f) * spam_score
  Semantics of a $reputation_factor is equivalent to auto_whitelist_factor in
  a SpamAssassin's AWL plugin, which shifts spam score towards a long term
  spam score average of a sender.

- Notifications and bounces show a "(dkim:AUTHOR)" next to a From address,
  and a "(dkim:SENDER)" next to a Sender address if these header fields
  were signed and their domain corresponds to a signer's domain identity.

- A valid DKIM or DomainKeys signature turns on a 'sender_credible' attribute
  which serves to choose one of the two DSN cutoff levels, so that delivery
  status notifications can be restricted to or preferred for likely-to-be-valid
  sending addresses, and bounces to possibly fake addresses can be minimized.
  More information on the 'sender_credible' attribute can be found earlier
  in this release notes.

Currently the ADSP (Author Domain Signing Practices, formerly SSP) draft
is not implemented, neither by amavisd, nor by SpamAssassin's plugin DKIM.
Work on a draft is still in progress, and until it settles one needs to
resort to SpamAssassin rules to block fake mail with no valid signature
from domains which are known to be signing all their mail, such as PayPal,
eBay,, and others. In essence, the ADSP information
(usually inferred, or actually published (quite rare today)) from such
domains needs to be encoded into SpamAssassin rules.


1. Generate one or more keys to be used for signing and enable signing code
   by adding the following line to amavisd.conf (if not already there):

     $enable_dkim_signing = 1;  # loads DKIM signing code

   Signing keys must be made available to amavisd, each private key in a
   separate file in PEM format. Customarily such keys would be generated
   and kept in a dedicated directory such as /var/db/dkim or /var/lib/dkim,
   preferably owned by root.

   Private keys can be generated by a 'openssl genrsa' command (see RFC 4871
   Appendix C, also in README_FILES directory), or by an amavisd equivalent.
   Commonly one key per signing domain or one key per signing host is used,
   but other choices are possible. If such keys were already prepared for
   some other DKIM-signing solution, they can be reused by amavisd.

     # amavisd genrsa /var/db/dkim/a.key.pem
     # amavisd genrsa /var/db/dkim/b.key.pem 786
     # amavisd genrsa /var/db/dkim/sel-example-com.key.pem
     # amavisd genrsa /var/db/dkim/g-guest-ex-com.key.pem
     # amavisd genrsa /var/db/dkim/notif-mail.key.pem 512

   Amavisd already ensures the generated files are only readable by owner,
   a manual procedure may require explicitly setting of file permissions.
   Private keys must be protected from unauthorized access, only the
   signing software such as amavisd should have access. Amavisd loads
   these files on startup before dropping privileges, so if amavisd is
   started as root it is not necessary that these key files are readable
   by uid under which amavisd is running.

2. Add commands to amavisd.conf to load private keys, associate them with
   signing domains and selectors, and describe constraints (tags) to be
   published with public keys. For example:

# Load all available private keys and supply their public key RR constraints.
# Arguments are a domain, a selector, a key (a file name of a private key in
# PEM format), followed by optional attributes/constraints (tags, represented
# here as Perl hash key/value pairs) which are allowed by RFC 4871 in a public
# key resource record (v, g, h, k, n, s, t), of which only g, h, k, s and t
# are considered to be constraints limiting the choice of a signing key.
# A command 'amavisd showkeys' can be used for displaying corresponding
# public keys in a format directly suitable for inclusion into DNS zone files.
#        signing domain  selector     private key              options
#        -------------   --------     ----------------------   ----------
dkim_key('', 'abc',       '/var/db/dkim/a.key.pem');
dkim_key('', 'yyy',       '/var/db/dkim/b.key.pem', t=>'s');
dkim_key('', 'zzz',       '/var/db/dkim/b.key.pem', h=>'sha256');
dkim_key('', 'sel-2008',  '/var/db/dkim/sel-example-com.key.pem',
         t=>'s:y', g=>'*', k=>'rsa', h=>'sha256:sha1', s=>'email',
         n=>'testing; 1, 2');
dkim_key('', 'g',     '/var/db/dkim/g-guest-ex-com.key.pem');
dkim_key('',  'notif', '/var/db/dkim/notif-mail.key.pem');

   A selector paired with a domain name uniquely identifies a key, both for
   a signer as well as for a recipient. There may be multiple keys for each
   domain as long as each one has its own selector.

   A selector with a domain name will be used by a receiving mailer in
   assembling a DNS query (selector._domainkey.signingdomain) to fetch a
   public key from a signing domain's DNS server when verifying signature

   A selector with a domain name will also be used by a signing amavisd
   when choosing a key applicable to signing, meeting constraints on its
   public key (tags, RFC 4871 section 3.6) as given by optional arguments.
   The optional arguments serve as documentation, may help amavisd choose
   between multiple choices, and supply additional information for step 3.

   For a list of options (tags) see RFC 4871 section 3.6.  Amavisd does not
   check the syntax of tag values, except for performing qp-section encoding
   of a tag 'n'. Note the Perl syntax of key/value pairs, e.g. t => 's:y'
   will end up as "t=s:y", and n => 'testing; 1, 2' will end up encoded
   as "n=testing=3B 1, 2".

3. Prepare and publish public keys.

   Public keys can be extracted from generated key files (which contain
   both a private and a public key). To publish public keys they need to
   be edited into a format suitable for inclusion in a DNS server's zone
   file for each signing domain, either by following a procedure in RFC 4871
   Appendix C, or if step 2 was completed, by asking amavisd to do so:
     # amavisd showkeys

   or more selectively, e.g.:
     # amavisd showkeys  .org

   This step is not needed if public keys were already prepared and
   published earlier for some other DKIM-signing solution.

4. Edit zone files in master DNS server(s) for each signing domain, adding
   the just prepared TXT resource records, not forgetting to bump up the
   serial number in a SOA record. Optionally add a TXT record with ADSP
   information (formerly SSP) if a default Author Domain Signing Practices
   is not appropriate. Then reload zone(s) or restart DNS server(s).

5. Test published public keys.

   Similar to 'showkeys', a 'testkeys' command walks through available
   signing keys, generates a test message signed with each key, and
   validates it by fetching a corresponding public key from a DNS server.
     # amavisd testkeys
   or more selectively, e.g.:
     # amavisd testkeys  .org

6. Restart amavisd, watch the log at log level 2, searching for " dkim: ".

   Note that signing could be started (amavisd reload) right after
   completing step 2, but mail recipients would not be able to verify
   validity of signatures until public keys are made available by a
   signing domain through its DNS. Recipients are supposed to treat mail
   with signatures which fail verification exactly the same as mail
   with no signatures, so there is usually no harm done with a premature
   start of signing, but there is no benefit either.

7. Optional: to override default values for signature tags, one may specify
   by-sender signature tags through @dkim_signature_options_bysender_maps,

# @dkim_signature_options_bysender_maps maps author/sender addresses or
# domains to signature tags/requirements; possible signature tags according
# to RFC 4871 are: (v), a, (b), (bh), c, d, (h), i, l, q, s, (t), x, z;
# of which the following are determined implicitly: v, b, bh, h, t
# (tag h is controlled by %signed_header_fields);  currently ignored tags
# are l and z;  instead of an absolute expiration time (tag x) one may use
# a pseudo tag 'ttl' to specify a relative expiration time in seconds, which
# is converted to an absolute expiration time prior to signing: x = t + ttl;
# a built-in default is provided for each tag if no better match is found
@dkim_signature_options_bysender_maps = ( {
  '' => { a => 'rsa-sha1', ttl =>  7*24*3600 },
  ''   => { a => 'rsa-sha1', ttl =>  7*24*3600 },
  ''            => { a => 'rsa-sha1', ttl => 10*24*3600 },
  # explicit 'd' forces a third-party signature on foreign (hosted) domains
  'guest.example'               => { d => '' },
  ''                => { d => '' },
  # catchall defaults
  '.' => { a => 'rsa-sha256', c => 'relaxed/simple', ttl => 30*24*3600 },
  # 'd' defaults to a domain of an author/sender address,
  # 's' defaults to whatever selector is offered by a matching key 
} );

  The result of a by-sender lookup into @dkim_signature_options_bysender_maps
  is a hash (a set) of DKIM signing requirements (tags), i.e. canonicalization
  method, hashing algorithm, domain, identity, selector and expiration time.
  Resulting tags are then used to choose the most appropriate signing key from
  a set of keys as declared by calls to dkim_key. Main selection criterium
  is a match on tags d (domain) and s (selector), but other signature
  requirements must also meet the constraints of a public key (e.g. subdomain
  matching flag, granularity, hashing algorithm, key type). If a lookup does
  not find a signing key which meets requirements, no signing takes place.
  Also, only mail with 'originating' flag is eligible for signing. A lookup
  is based on either the From (header field), the Sender (header field), or
  on a mail_from address from the envelope, whichever yields a useful result
  first. This results either in an author signature (i.e. a first-party
  signature, based on a From), or a third-party signature (based on Sender
  or mail_from or forced through d tag).

  An associative array %signed_header_fields controls which header fields
  are to be signed. By default it contains a standard (RFC 4871) set of
  header field names, augmented by some additional header field names
  considered appropriate at the time of a release (RFC 4021, RFC 3834).
  In addition a 'Sender' header field is excluded because it is frequently
  replaced by a mailing list, and as the RFC 2821 mandates there can only
  be one such header field the original one is dropped, invalidating a
  signature. Also the 'To' and 'Cc' are excluded from a default set because
  sendmail mailers are known to gratuitously reformat the list, invalidating
  a signature.

  The default set of header fields to be signed can be controlled
  by setting %signed_header_fields elements to true (to sign) or
  to false (not to sign). Keys must be in lowercase, e.g.:
    $signed_header_fields{'received'} = 0;  # turn off signing of Received
    $signed_header_fields{'to'} = 1;        # turn on signing of To

                                                             March 12, 2008
amavisd-new-2.5.4 release notes


- simplify regular expressions in parse_quoted_rfc2821() to avoid perl crashing
  on a long degenerated e-mail address; reported by Sébastien Aveline;

- further simplify (split in two) regular expressions in parse_address_list()
  to avoid perl crashing on long degenerated e-mail addresses in From, To,
  and Cc header fields, also reported and sample provided by Tomi Lukkarinen;

- incorrect parsing of header fields could let a header field to be ignored
  when preparing notification templates, or when adding a spam tag to a
  Subject header field, causing a second Subject header field to be inserted;
  reported by Mike Cisar;

- untaint a policy bank name when it comes from an AM.PDP protocol request;
  symptom was a failure to insert a pen pals SQL record in a milter setup;
  reported by Peter Huetmannsberger;

- smtp client code inappropriately concluded there is no progress being made
  when forwarding a message back to MTA, and exited a rw_loop when sysread
  returned status EAGAIN despite a 'select' setting an input-ready flag;
  the problem was detected on Solaris, although it could be more general;
  thanks to Aleksandr for a detailed problem report;

- limit the number of filenames given as arguments to a file(1) utility
  to stay within a safe program argument space limit, run file(1) multiple
  times if necessary;

- change the sprintf format for conversion of 64-bit SNMP-like counter values
  into a string (replaced %020d by %020.0f) to properly convert large values
  (beyond 32 bits) into strings on versions of Perl which are not compiled
  with support for 64-bit integers (Solaris?); reported by David Schweikert;


- invoke unrar and rar without option 'av-', which is no longer valid
  since version of unrar 3.7.5, and was previously apparently just ignored;
  the rar documentation states that option 'av' is only available with
  registered versions, so it appears the 'av-' is redundant with rar as well.
  Unrar aborting with 'unknown option' error was reported by Piotr Meyer;

- a new AV entry for fpscan 6.x, (F-PROT Antivirus for UNIX command-line
  scanner) based on information from Erik Slooff, Bruno Friedmann, and Steve);

- a new AV entry for fpscand 6.x (F-PROT Antivirus for Linux/BSD/Solaris),
  based on initial research by Alexander Wirt, Henrik K, and F-Prot
  documentation, thanks also to Haukur Valgeirsson of F-Prot International;

- ask_daemon_internal now keeps persistent connection to an F-PROT fpscand
  daemon, in addition to Sophie and Trophie daemons;

- new AV entry 'bdscan' for a new version of BitDefender Antivirus Scanner
  for Linux and FreeBSD; thanks to Gary V;

- updated README.postfix, thanks to Chris Pepper and Patrick Ben Koetter;

- updated amavisd-new-docs.html;

                                                          December 12, 2007
amavisd-new-2.5.3 release notes


- fix parsing a SMTP status response from MTA when releasing from a
  quarantine, when a MTA response did not include an enhanced status
  code (RFC 3463) (such as with old versions of Postfix);  a parsing
  failure resulted in attribute "setreply=450 4.5.0 Unexpected:..."
  in an AM.PDP protocol response, even though a release was successful;
  reported by Ron Miller, John M. Kupski, investigated by Tony Caduto
  and Jeremy Fowler;

- change parsing of addresses in From, To, and Cc header fields, avoiding
  complex Perl regular expressions which could crash a process on certain
  degenerate cases of these header fields; thanks for detailed problem
  reports to Carsten Lührs and Attila Nagy;

- completely rewritten parsing of Received header field to work around a
  Perl regular expression problem which could crash a process on certain
  degenerate cases of mail header fields; problem reported by Thomas Gelf;

- harden to some extent regular expressions in parse_message_id to cope
  better with degenerate cases of header fields carrying message-id;

- sanitize 8-bit characters in In-Reply-To and References header fields
  before using them in Pen Pals SQL lookups to avoid UTF-8 errors like:
    penpals_check FAILED: sql exec: err=7, 22021, DBD::Pg::st execute failed:
    ERROR: invalid byte sequence for encoding "UTF8": 0xd864

- when turning an infection report into a spam report, avoid adding newly
  discovered virus names (i.e. fraud names) to a cached list if these names
  are already listed; previously the list would just grow on each passage
  through a cache, leading to unsightly long lists of spam tests in a
  report; based on a patch by Henrik Krohns;

- fix diagnostics when an invalid command line argument is given;


- reduce log clutter when certain Perl modules are loaded late, i.e. after
  chrooting and daemonizing, but still before a fork; now only issue one
  log entry by a parent process: "extra modules loaded after daemonizing: ";

- slightly relax mail address syntax in subroutine split_address;

- fetch additional information (tags) from SpamAssassin: TESTS, ASN,
  ASNCIDR, DKIMDOMAIN and DKIMIDENTITY, making them available through a macro
  'supplementary_info' (if a version of SpamAssassin in use provides them);

- updated DKIM section in amavisd-new-docs.html, removing the historical
  DomainKeys milter from examples;

- declared a dummy subroutine dkim_key() and new dummy configuration
  variables @dkim_signature_options_bysender_maps, %signed_header_fields,
  $reputation_factor, @signer_reputation_maps and $sql_partition_tag, members
  of policy banks, in preparation for 2.6.0 - declared now for improved
  downgrade compatibility of 2.6.0 configuration files, if need arises.

                                                              June 27, 2007
amavisd-new-2.5.2 release notes


- in a milter setup log_id was left undefined, which resulted in log lines
  without id, and a SQL constraint violation "Column 'am_id' cannot be null"
  when logging to SQL was enabled. The bug was introduced in 2.5.1;
  problem reported by Martin Svensson; 

- suppress a second quarantining attempt if the message also needs to be
  archived to the same location (same sql key or same local filename);
  reported by Wazir Shpoon;

- adjust $socketname in amavisd-release to match its default counterpart
  in amavisd (i.e. /var/amavis/amavisd.sock); reported by Stanley Appel;


- add snmp-like counters for PenPalsSavedFromKill, PenPalsSavedFromTag3
  and PenPalsSavedFromTag2, which correspond to the number of messages
  since a program (re)start in which spam level would have exceeded a
  corresponding level had there not been for (negative) score points
  contributed by pen pals lookups. Note that for any message only one
  of the three counters could increment, the one corresponding to the
  highest level crossed. To find more information about rescued mail
  messages, search the log for a string 'PenPalsSavedFrom' (available
  at log level 2 or higher). Practical value: mail saved by pen pals
  from being blocked often indicate false positives by SpamAssassin;
  examining rules which contributed significantly to the score may
  indicate which rules need adjustment;

- when preparing a SQL SELECT clause in lookup_sql, provide an additional
  placeholder %a in a clause template, which is much like the existing %k,
  but evaluates to an exact mail address (i.e. the same as the first entry
  in the %k list), which makes it suitable for SQL pattern matching;
  suggested by Daniel Duerr;

- macro supplementary_info can supply information on two additional
  SpamAssassin tags: AUTOLEARNSCORE and LANGUAGES if corresponding plugins
  are enabled in SpamAssassin; see README.customize for the complete list;

- provide two new subroutines available for calling from config files:
  include_config_files() and include_optional_config_files(), each take
  a list of filenames as arguments, and reads & evaluates them just like
  normal configuration files specified on a command line (option -c
  or a default amavisd.conf). This provides a simplified and uniform
  mechanism for 'including' additional configuration files, which formerly
  could be invoked through a perl do() function. The only difference
  between include_config_files and include_optional_config_files is that
  the former aborts if some specified file does not exist, while the later
  silently ignores specified but missing files. Both/each subroutine
  may be called multiple times, recursion is allowed (but some sanity
  limit to recursion is provided); based on a suggestion by Gary V.

  Example line in amavisd.conf:


- provide a workaround for a crashing altermime by removing its leftover
  temporary file which would otherwise cause a temporary failure:
    TempDir::check: Unexpected file
  problem reported by Dennis A. Kanevsky;

- add a mapping to 'doc' for a result 'Microsoft Installer' from a file(1)
  utility; it seems like versions 4.20 and 4.21 of file(1) (possibly earlier
  versions too) misclassify MS Word, Excel, and PowerPoint documents as
  'Microsoft Installer';  problem investigated and a workaround suggested
  by Noel Jones, Mike Cappella and Michael Scheidell;

- add a mapping to 'asc' for a result 'COM executable for DOS' from a file(1)
  utility; it seems like later versions of file(1) can misclassify a text
  in a GB2312 character set as a COM file; reported by Daniel J McDonald;

- updated AV entry for ESET NOD32 Linux Mail Server again - command line
  interface (nod32cli): added a status 3 (e.g. corrupted archive) back to
  the list of clean statuses;  the 3 was removed in 2.5.1 as the entry
  was substituted with the one from a NOD32 documentation; reported by
  Tamás Gregorics;

- updated AV entry for 'F-Secure Antivirus for Linux servers' to cope
  with version 5.5 and later; a new entry provided by Peter Bieringer;

- when a command line option -g requests changing of group ID, do so
  by calling POSIX::setgid, after also attempting to assign to perl
  variables $( and $), which may not work correctly on systems where
  group ID can be negative (like group 'nobody' being -2 on Mac OS X);
  follows a SpamAssassin problem report 3994, investigated
  by Sidney Markowitz;

- when an AUTH command parameter (RFC 2554, now RFC 4954) is supplied on a
  MAIL FROM SMTP command but AUTH support has not been previously offered
  (like when authentication is disabled by an empty @auth_mech_avail),
  no longer treat the situation as a fatal error:
    503 5.7.4 Error: authentication disabled
  but mercifully ignore the parameter and just log an informational
  message. This is a deviation from RFC 2554, but makes it friendlier
  for those insisting on running amavisd as a Postfix pre-queue smtp
  proxy; suggested by Alexander 'Leo' Bergolth;

- adjust the list of pre-loaded perl modules required by SpamAssassin;

- internal: pass a mail message to SpamAssassin as a GLOB instead of an
  array reference, saving one in-memory copy of a message during a SA call;

- internal: make it slightly easier to switch message digest from MD5 to
  a Digest::SHA family by turning a hard-wired key length into a parameter
  (admittedly it is still ugly, requiring a change in three places for
  switching); also pave a transition from Digest::SHA1 to Digest::SHA;

- documentation: updated files README.postfix and README.postfix.html
  now include a section 'Advanced Postfix and amavisd-new configuration'
  explaining a multiple cleanup service architecture; thanks to
  Patrick Ben Koetter;  retired file: README.postfix.old

- documentation: updated README.sql-pg to include a faster alternative to
  purging a SQL logging database: the alternative 'DELETE FROM maddr' on
  PostgreSQL runs faster by a factor of 1.5 to 2 from the one previously

- suggestion: when using SpamAssassin plugin Rule2XSBody
  (available in more recent versions of SA), adding an entry like:
  to the @additional_perl_modules list allows preloading of compiled rules.

  Adding the following two lines to amavisd.conf adds the directory name
  containing modules with compiled rules to Perl modules search path and
  allows Perl to find the listed module(s):

    my($sa_instdir) = '/var/lib/spamassassin/compiled/3.002001';
    unshift(@INC, $sa_instdir, $sa_instdir.'/auto');

                                                               May 31, 2007
amavisd-new-2.5.1 release notes


- setting $bypass_decode_parts to true now also disables MIME decoding
  (see below);


- provides checking the number of archive members against $MAXFILES quota
  even when just listing an archive directory, providing some additional
  protection (besides a time limit) against runaway dearchivers
  (such as a recent Zoo archiver DoS);

- please use the most recent versions of file(1) utility (currently 4.21)
  and recent versions of external dearchivers/decoders to avoid known
  security vulnerabilities in them;


- introduced a variation of a message release from a quarantine, allowing
  a releaser to choose between forwarding a message to the back-end MTA
  port as usual (avoiding re-checking of a message), or to send it to MTA
  on its incoming port (normally 25) and let the message be rescanned,
  which might be useful after adjusting spam rules or antivirus database.

  It is implemented by:

    * adding a configuration variable $requeue_method (also a member
      of policy banks), with a default value: 'smtp:[]:25'

    * extending the AM.PDP protocol with a 'request=requeue' attribute
      which can be used in place of a 'request=release',

    * enhancing the 'amavisd-release' utility program to choose between
      sending 'request=release' and 'request=requeue' based on its
      program name, i.e. by making a soft or hard link to amavisd-release
      (or its copy) named 'amavisd-requeue', the utility will send
      a 'request=requeue' in place of the usual 'request=release', e.g.:
        # ln -s amavisd-release amavisd-requeue
        $ amavisd-requeue spam/k/kg2P0rP9Lpu3.gz

    * enhancing amavisd daemon to choose between forwarding a released
      message either to $release_method or to $requeue_method destination
      based on a 'request' attribute value in an AM.PDP request;

- new AV entry: ArcaVir for Linux and Unix, see below for links;

- a new macro 'supplementary_info' gives access to some additional information
  provided by content scanners, such as a provided by SpamAssassin API
  routine get_tag. The macro takes two arguments, the first is a tag name
  (a name of some attribute which is expected to provide an associated
  value), the second argument is a sprintf format string and is optional,
  if missing a %s is assumed. Currently the only available attributes are
  AUTOLEARN, SC, SCRULE, SCTYPE, and RELAYCOUNTRY. These are nonempty only
  when an associated SpamAssassin plugin or function is enabled.


- fixed quarantining to a SQL database of messages with a null envelope
  sender address (broken in 2.5.0, causing such messages to tempfail);
  reported by Markus Edholm, Vahur Jõesalu and Michael Scheidell;

- fixed parsing of certain broken 'From' header fields, which would
  result in a temporary failure and the following logged error:
    check_init2 FAILED: parse_address_list PANIC1 53
      at /usr/local/sbin/amavisd line 3292
  reported by Michael Scheidell;

- avoid encoding nonprintable characters in X-Envelope-From and X-Envelope-To
  header fields in a quarantined message even if envelope mail addresses
  contain such invalid characters, so that a quarantine release is possible;
  (RFC 2047 allows encoding of a 'phrase' in From, To, and similar headers,
  as well as in comments, but not in the address specification);

- avoid unnecessarily RFC 2047 -encoding of 8-bit characters in those
  lines of inserted X-Spam-Report (and similar) multiline header fields
  which only contain ASCII characters; also avoid encoding of newlines;
  reported by Anant Nitya;

- sanitize 8-bit characters in SpamAssassin report before inserting it
  into an X-Spam-Report header field;

- properly recognize PostgreSQL error code 'S8006' and reconnect to
  a disconnected server right away; thanks to Brian Wong;

- call $mail_obj->finish after a SA call to allow for garbage collection
  and removal of SA temporary files;  see:

- avoid nonstandard SMTP status code 254 on discarded malware;
  on discarding turn status 554 into a 250 instead;  violation
  of a SHOULD in RFC 2822 pointed out by Alexander 'Leo' Bergolth;

- an informational log message was reported inappropriately:
    INFO: truncated ... header line(s) longer than 998 characters
  it didn't reflect reality, it was always reported together with the:
    INFO: unfolded 1 illegal all-whitespace continuation lines

- when a SMTP option BODY=8BITMIME (RFC 1652) is not given on mail
  reception, avoid turning it on while forwarding, even if mail body
  contains 8-bit characters;  following a garbage-in-garbage-out
  principle, this doesn't break anything that isn't already broken,
  but might prevent later conversion to 7-bit quoted-printable MIME
  by some downstream MTA, invalidating signatures (DKIM, S/MIME,
  PGP, ...) - at a risk that some overzealous firewall might block
  a mail transfer;

- fixed a couple of documentation typos/bugs in README.customize,
  thanks to Mike Cappella;


- modified code for checking each eval {} status: it turns out that eval
  is able to capture certain error conditions (e.g. certain I/O errors)
  but without setting the $@ variable, leaving it empty;  use new idiom
  throughout for proper error handling and more informative reporting,
  showing errno in such cases;

- setting $bypass_decode_parts to true now also disables MIME decoding,
  not just decoders/dearchivers listed in a @decoders list, and also
  implicitly retains full original message for virus checking, equivalent
  to having a regular expression /^MAIL$/ in a @keep_decoded_original_maps
  list;  prompted by Bill Landry;

- new AV entry: ArcaVir for Linux and Unix, see:
  the entry was kindly provided by Michal Seremak;

- updated AV entry for ESET NOD32 Linux Mail Server - command line
  interface (nod32cli), version 2.7, thanks to Simon;

- updated AV entry for Sophos sweep, adding options -mime and -oe ;

- avoid repeatedly reporting the same set of modules by a log entry
  'extra modules loaded:', only report it on changes to the list;
  repeated reports could be misinterpreted that modules were loaded
  with each mail task, where actually missing modules were only loaded
  once within each child process;

- avoid reporting 'BOUNCE' in a SMTP response text when a bounce (i.e.
  a nondelivery status notification) was actually suppressed, such as
  is usually the case with infected mail or when spam score exceeds
  spam_dsn_cutoff_level. Previously the SMTP response text only reflected
  the setting of a final_*_destiny, which could mislead mail administrators
  into believing that excessive unconditional backscatter was being
  generated. The new text looks like:
    250 2.5.0 Ok, id=67685-15, DISCARD(bounce.suppressed)
  instead of previous:
    250 2.5.0 Ok, id=67685-15, BOUNCE

  A general note worth reiterating: to reduce backscatter pollution
  (sending of bounces to innocent sender addresses), please either:

    * set $final_virus_destiny and $final_spam_destiny to D_DISCARD
      or to D_PASS  (_not_ to D_REJECT or D_BOUNCE),

    * carefully configure virus and spam bounce suppression by:
      . configuring @viruses_that_fake_sender_maps correctly (the default is
        fine, it suppresses all bounces to infected mail), this way one may 
        safely set $final_virus_destiny to D_BOUNCE, it is equivalent to
        D_DISCARD for all infected mail containing malware matching the

      . and: configuring @spam_dsn_cutoff_level_maps and
        @spam_dsn_cutoff_level_bysender_maps, keeping levels just slightly
        over a kill level, have a well maintained SpamAssassin with network
        tests enabled and updated rules - then one may set $final_spam_destiny
        to D_BOUNCE, which will produce bounces for mail with spam score
        between kill level and cutoff level, and suppress bounces above a
        suppress level; some domains may still consider such practice abusive,
        so do not take this choice lightly;

      . to monitor bounces generated by amavisd, one may assign some
        dedicated monitoring e-mail address to $dsn_bcc, which will then
        receive a copy of all delivery status notifications sent out
        by amavisd;

- dspam options changed with version 3.8.0, replacing option --feature
  with --tokenizer;  reported by Jim Knuth;

- modified syslog writing to check errno after calling Unix::Syslog::syslog,
  and to informatively attempt to log status when unsuccessful;
  an unsuccessful status is just informational, as syslog(3) routine does
  its own retries and leaves an unsuccessful status of a previous attempt
  in errno even if a subsequent logging attempt did succeed; unfortunately
  the system routine syslog(3) returns no value according to documentation
  (and according to its source code), so its completion status can not be
  tested; a problem of a loss of logging on a syslogd restart on OS X was
  reported by Paul Walker, but unfortunately can not be solved on the
  application side;

- uncomment some debugging printouts in and land them
  under control of a $debug variable;

                                                             April 23, 2007
amavisd-new-2.5.0 release notes


The 2.5.0 is upwards compatible with 2.4.* versions, except
for the following:

Default notification and logging templates are enhanced to take advantage
of new macros and new concepts, so it is prudent to update templates
if defaults are overridden, e.g. $log_templ, $notify_*_admin_templ, ...

A client-side AUTH (rfc2554: SMTP Service Extension for Authentication)
is currently not available. The last version with this feature working
is amavisd-new-2.5.0-pre4. The feature may be restored in a future version
if sufficient interest is demonstrated.

A workaround for a qmail bug (which complains when CR and LF are split
across a TCP segment boundary) is no longer available, as the program
has no control over IP packet splitting done by the TCP/IP stack.

When pen pals feature is in use, it is worth creating an index
on a msgs.message_id field.


- new concept: blocking contents category;

- true per-recipient defanging/sanitation of a mail body (previously
  a true per-recipient handling was available for mail header edits,
  but not for mail body modifications);

- added interface code to invoke Anomy Sanitizer or the 'altermime' program
  allows defanging or adding disclaimers by external utilities on a
  per-recipient basis;

- rewritten SMTP client code: get rid of the troublesome module Net::SMTP;
  new code now supports pipelining, client-side LMTP, IPv6, Unix sockets,
  more reliable error detection and handling, passes on ENVID parameter
  unchanged, is bare-CR-clean, tidier code (no more workarounds for rough
  corners in Net::SMTP), fewer context switches (handshake handovers) due
  to pipelining if pipelining is offered by MTA (which usually is);

- makes available pedantically parsed addresses from a mail header:
  From, Sender, To, Cc.  Addresses from mail header may be needed for
  deciding on inserting disclaimers, signing mail (DKIM), custom hooks
  (like 'vacation'-type applications), and other future applications.
  Get rid of inexact parsing by module Mail::Address, provide own parser;

- phishing fraud as returned by ClamAV is now treated as spam, no longer
  as a virus;

- compatible with SpamAssassin 3.2.0;

- enhancements to amavisd-nanny: shows more detailed states of processes;

- enhancements to amavisd-agent: shows average processing times per message;

- extended AM.PDP protocol with an attribute 'policy_bank' which may be used
  in a client's request to require loading additional policy banks;

- add support for 7-Zip archives if external utility 7z is available;

- custom hooks allow custom code to be called at few strategic places;

- penpals can now also match replies which reference previous outgoing mail
  by its Message-Id (taking into account References or In-Reply-To header

- new key 'originating' in policy banks generalizes a MYNETS policy bank;

- a documentation rewrite for setting up amavisd-new with Postfix
  by Patrick Ben Koetter (one of the two authors of The Book of Postfix).
  Previous documentation has been renamed to README.postfix.old and will be
  removed in the next version; the new documentation is README.postfix.html,
  and its automatically converted plain text version is README.postfix.
  A big thanks to Patrick for his efforts!


- if a sender is both white- and black-listed at the same time, then
  inserted X-Spam-* header fields were inconsistent, e.g. X-Spam-Level,
  X-Spam-Flag and X-Spam-Status reflected a whitelisted status (no asterisks,
  not a spam), while X-Spam-Score showed 64 points; now whitelisting prevails
  in all X-Spam-* header fields;

- relax argument parsing in amavisd-release to allow releasing of
  quarantine id containing a body hash in a name (%b in template);
  reported by Ron Rademaker;

- skip a SQL-logging database operation if an associated clause in %sql_clause
  is disabled, e.g. set to undef or '';  this allows for example to selectively
  disable SQL logging based on a policy bank; thanks to Riaan Kok;

- let LHA decoder (do_lha) recognize also other listing formats, e.g. MS-DOS,
  symlinks, not just plain Unix archives; problem reported by Ryuhei Funatsu;


- catch and log uncaught '__DIE__' and '__WARN__' Perl pseudo-signals
  which would otherwise go to stderr and not be noticed with a daemonized
  process; patch by Alexander 'Leo' Bergolth;

- insert 'X-Spam-Flag: NO' if spam level is above tag level but below tag2
  level, previously it was not inserted at all (it is still redundant, but
  some may appreciate an explicit statement nevertheless);

- drop support for Archive::Tar; main drawback of this module is: it either
  loads an entire tar into memory (horrors!), or when using extract_archive()
  it does not relativize absolute paths (which makes it possible to store
  members in any directory writable by uid), and does not provide a way to
  capture contents of members with the same name. Use pax program instead!

- abandon the use of libnet (modules Net::SMTP and Net::Cmd), replaced by
  own code to implement client-side SMTP and LMTP protocol support, with a
  full support for pipelining and IPv6. Some of my issues with Net::SMTP go
  back to year 2002, but the one that broke camel's back is a 3+ months
  status quo in not fixing a serious misfeature introduced in 1.20, which
  mangles 8-bit characters in mail, causing a series of support questions,
  and even affecting some larger service providers without them realizing
  there is a problem. Here are some relevant bug reports:

  P.S. libnet-1.21 eventually fixed the UTF8 encoding problem and
  added support for ENVID and AUTH options; other issues are still open
  (e.g. split code/text in smtp status, no pipelining support, no LMTP);

- represent score in X-Spam-Status header field as a single numeric field
  instead of the previous explicit sum of a SA score and a score boost,
  which some MUAs don't know how to interpret and label mail spaminess

- recipient notifications (in their default template) now keep the original
  To and Cc header fields, instead of placing a recipient envelope address
  into a To header field; suggested by Jorgen Lundman;

- no longer removes an X-Amavis-Alert header field (as inserted by some
  foreign MTA), it does not hurt to keep it;

- kavscanner AV entry (Kaspersky Antivirus): added a new entry to the
  search list of paths to a binary, kavscanner changed its default
  installation location again; provided by Gary V;

- added a rule into a $map_full_type_to_short_type_re list, mapping a
  file(1) type 'RIFF...animated cursor' into '.ani';  a patch by Eric;

- example rules (commented-out) to block animated cursors (ANI) and icons
  are added to configuration files amavisd.conf and amavisd.conf-sample;

- suppress bounces not only for Precedence: (bulk|list|junk) or null return
  path (as before), but also for mail with a 2822.From header field matching
  one of: *-request | *-owner | *-relay | *-bounces | owner-* | postmaster |
  mailer-daemon | mailer | uucp, and for mail containing a header field
  List-Id (RFC 2919). Note that this new additional rule practically never
  triggers in practice, true bounces and common mailing list traffic is
  already covered by previous checks;

- make available a list of detected virus names in
  an Amavis::In::Message object (virusnames), suggested by Tom Sommer;

- split SQL documentation into three files:
    README.sql        general SQL considerations and some examples
    README.sql-mysql  MySQL-specific notes and schema
    README.sql-pg     PostgreSQL-specific notes and schema (also SQLite)

- README.sql-pg now adds CHECK (x >= 0) for fields that are supposed
  to contain unsigned integers; suggested by Hanne Moa;

- repurpose/rename a contents category CC_TEMPFAIL to CC_MTA;  it is turned
  on when MTA rejects or tempfails a mail when amavisd attempts to pass it
  back after checking it; (this should not normally happen, MTA should
  be doing most of its checks (e.g. recipient validation) before passing
  mail to a content filter, not after);


- custom hooks allow custom code to be called at few strategic places:
   * during child process initialization: allows initialization of custom
     code, including establishing an additional SQL session or similar;
   * after built-in checks: allows custom code to inspect and/or modify
     results of checks;
   * before sending, allows for additional quarantining and for sending
     additional notifications;
   * at the end of message processing: allows inspecting results of checks
     and status of mail forwarding (e.g. for statistics or logging purposes).

   Mail processing sequence:

     child process initialization
    *custom hook: new()
     loop for each mail:
       receive mail
       mail checking and collecting results
      *custom hook: checks() - may inspect or modify checking results
       deciding mail fate (lookup on *_lovers, thresholds, ...)
       sending notifications (to admin and recip)
      *custom hook: before_send() - may send additional notifications,
                                    additional quarantining, may modify mail
       forwarding (unless blocked)
       sending delivery status notification (if needed)
       issue main log entry, manage statistics (timing, counters, nanny)
      *custom hook: mail_done() - may inspect results
     endloop after $max_requests or earlier

  If Amavis::Custom::new returns undef then no further custom calls are made.
  Otherwise, this method is supposed to return an object, thus enabling
  other custom hooks, which receive this value as their first argument.

  See amavisd-custom.conf for an example, and the example invocation
  of amavisd-custom.conf at the end of file amavisd.conf-sample;
  thanks to Kasscie (Yohanna Monsalvez);

- formerly penpals could only match replies to previous outgoing mail
  where envelope sender and recipient addresses are exactly reversed.
  Now, in addition to this, penpals can also match replies which reference
  previous outgoing mail by its 'Message-ID' (taking into account the
  'References' or 'In-Reply-To' header fields), even if the envelope
  sender address of the reply is null or does not match a recipient address
  of a previous outgoing mail. This covers for incoming replies to mailing
  list postings, incoming message disposition notifications (MDN, RFC 3798)
  and incoming replies from alias or role addresses. A query on a
  message-id is fast compared to matching on recipient id, and if it
  succeeds, the later one is skipped. Based on a suggestion and a patch
  by Alexander 'Leo' Bergolth;

  The %sql_clause now contains one additional SQL SELECT clause under
  'sel_penpals_msgid' key, which is used in place of 'sel_penpals' when
  a list of references in a reply is nonempty. It is worth creating
  an index on a msgs.message_id field to speed up SQL lookups:

    CREATE INDEX msgs_idx_mess_id ON msgs (message_id);

  Note that a spammer can gain a (small) advantage by including a reference
  to a recent outgoing message in his message. With private correspondence
  this information is hard to come by efficiently in a timely manner, and
  can effectively be exploited only if generated Message-IDs of outgoing
  mail is very easy to guess (e.g. generated by a very poor random number
  generator). With postings to public mailing lists this information is more
  readily available. Setting $sql_clause{'sel_penpals_msgid'} to undef or
  to empty disables matching on message-id, if it turns out the mechanism
  is being actively exploited.

- added penpals snmp-like counters, which are displayed by amavisd-agent
  like the following:

    PenPalsAttempts      10222   438/h   18.3 % (InMsgsRecipsInboundOrInt)
    PenPalsHits           2957   127/h   48.9 % (ContentCleanMsgsInboundOrInt)
    PenPalsHitsMid          88     4/h    3.0 % (PenPalsHits)
    PenPalsHitsMidNullRPath 26     1/h    0.9 % (PenPalsHits)
    PenPalsHitsMidRid      917    39/h   31.0 % (PenPalsHits)
    PenPalsHitsRid        1926    82/h   65.1 % (PenPalsHits)

  Some comments on the above figures:
  - PenPalsAttempts shows that SQL lookups are performed only on a fraction
    of all incoming and internal mail (and not at all on outgoing mail)
    because a lookup is not performed on a high score spam (score above
    $penpals_threshold_high, which was 8 in this example, typically
    the same as kill level);
  - PenPalsHits shows that more than half of incoming clean messages can be
    associated with previous outgoing mail and can benefit from pen pals
    soft-whitelisting, such message is either a reply directly referencing
    a previous mail, or a new conversation between a sender/recipient pair
    which had previous conversation on a different topic;
  - PenPalsHitsMidRid shows that 1/3 of incoming matching replies match
    a previous outgoing mail by both the Message-ID as well as the exact
    recipient address; these are true replies to previous outgoing messages;
  - PenPalsHitsRid shows that 2/3 of incoming matching replies match a
    a previous outgoing mail by recipient address, but not by a Message-ID;
    these messages usually correspond to new topics among previous
  - PenPalsHitsMid match only in reference to a previous Message-ID, but not
    by sender/recipient addresses; these are usually mailing list replies
    to a previous posting by a local user;
  - PenPalsHitsMidNullRPath are messages with null return path which are
    matching some previous outgoing message by Message-ID;  these are
    usually incoming message disposition notifications (MDN, RFC 3798), or
    certain bounces which specify In-Reply-To or References in their header;

- configuration variable %defang_by_ccat is renamed to %defang_maps_by_ccat
  and may now contain a list of by-recipient lookup tables (or a boolean
  as before for compatibility); this allows defanging/mangling to be selected
  on per-recipient basis;  compatibility is retained by making the old
  variable %defang_by_ccat an alias for %defang_maps_by_ccat;

- provided interface code to allow mangling/defanging/sanitation
  to be performed by an external utility, either by directly calling
  a Perl module Anomy Sanitizer (within the same process, avoiding
  startup cost), or by invoking a program 'altermime' (or by internal
  defanging code as before).

  Mail body defanging is only allowed for local recipients (those matching
  @local_domains_maps), i.e. for inbound and internal-to-internal mail.

  If there is more than one mangling code option available, the result
  of a %defang_maps_by_ccat can choose between them by returning one of
  the following strings, the selection can depend on mail content type
  and on by-recipient lookups if needed:
    'anomy'     chooses Anomy Sanitizer (if $enable_anomy_sanitizer is true);
    'altermime' chooses a program whose path is $altermime (if found);
    'attach'    chooses the traditional amavisd-new defanging method
                which pushes an original mail message to an attachment;
    'null'      for testing purposes - doesn't modify mail body, but
                pretends it does (in logging and mail header);
    other non-empty and non-zero value automatically choose one
                of the above options depending on what is available;
                at least the 'attach' is always available;
    an empty, zero or undef value disables mail body modifications;

  Controls: $enable_anomy_sanitizer, @anomy_sanitizer_args,
  and: $altermime, @altermime_args_defang;

  Typical use:

  # with altermime:
  $altermime = '/usr/local/bin/altermime';
  @altermime_args_defang = qw(--verbose --removeall);

  # with Anomy Sanitizer:
  $enable_anomy_sanitizer = 1;
  @anomy_sanitizer_args = qw( /usr/local/etc/sanitizer.cfg );

  $defang_spam = 1;  # old style, applies the first available mangler
                     # to all spam-loving local recipients

  # unnecessarily complicated example of selective choices:
  $defang_maps_by_ccat{+CC_BANNED} = [
    'altermime',  # use altermime for everybody (a 'constant' lookup table)
  $defang_maps_by_ccat{+CC_SPAM} = [
    { # a per-recipient hash lookup table
      ''    => 1,  # old style, auto-selects a mangler
      ''  => 'anomy',
      ''  => 'altermime',
      ''  => 'attach',
      ''        => 0,  # no mangling
    $defang_spam,  # fallback to old style setting if no match above

- a special case of mangling is adding a disclaimer, by invoking an external
  program 'altermime' (if available and enabled). This differs from mangling
  inbound mail in two details:
  * uses a separately configurable list of arguments to altermime:
    @altermime_args_disclaimer; and
  * it applies only to mail submitted from internal networks or roaming users
    (as recognized through a policy bank which sets: allow_disclaimers => 1),
    and where any of the following addresses matches local domains:
    author (2822.From) or sender (2822.Sender) or return path (2821.mail_from);

  Typically the $allow_disclaimers should be set by a policy bank which
  also sets the $originating flag.

  In addition to strings that may be returned by %defang_maps_by_ccat
  as described above, there are two more, only taken into account
  when $allow_disclaimers is true:
    'disclaimer' invokes $altermime program for outgoing mail with
                 arguments as given in @altermime_args_disclaimer;
    'nulldisclaimer' for testing purposes - doesn't modify mail body,
                 but pretends it does (in logging and mail header);

  Typical use:

    $altermime = '/usr/local/bin/altermime';
    @altermime_args_disclaimer =
      qw(--verbose --disclaimer=/etc/altermime-disclaimer.txt);
    $defang_maps_by_ccat{+CC_CATCHALL} = [ 'disclaimer' ];
    @mynetworks = qw( ... );
    $policy_bank{'MYNETS'} = {  # mail originating from our networks
      originating => 1,
      allow_disclaimers => 1,

  For the moment there is one limitation: there can only be one mangler
  in effect at a time, it is not currently possible to both defang and to
  append a disclaimer on the same message: for internal-to-internal mail
  inserting a disclaimer takes precedence.

  To make it possible to provide different disclaimer texts when hosting
  multiple domains, there is an experimental additional configuration
  variable available: the @disclaimer_options_bysender_maps.
  It is a list of lookup tables, looked up by a sender address.
  The sender address is chosen from the following list, first match wins:
    * 'Sender:' header field,  if its domain matches @local_domains_maps;
    * 'From:' header field,    if its domain matches @local_domains_maps;
    * envelope sender address, if its domain matches @local_domains_maps;
  We already know that at least one of the above will match, otherwise
  adding disclaimers would be skipped at an earlier stage. The result of
  lookups should be one simple string, which replaces a string '_OPTION_'
  anywhere in @altermime_args_disclaimer elements.

  Typical use:

    @altermime_args_disclaimer = qw(--disclaimer=/etc/_OPTION_.txt);

    @disclaimer_options_bysender_maps = (
     { '' => 'altermime-disclaimer-host1',
       ''  => 'altermime-disclaimer-boss',
       ''      => 'altermime-disclaimer-net',
        '.'                => 'altermime-disclaimer-default' },

  It is currently not possible to disable adding disclaimers through
  @disclaimer_options_bysender_maps results. This needs to be improved.
  The exact interpretation of the @disclaimer_options_bysender_maps lookup
  result may change in the future (which is why I call it 'experimental').

  Note that disclaimers are pretty much useless legally.
  If you can help it at all, please avoid the pollution. See:

- as mentioned above, the new SMTP/LMTP client code now supports a LMTP
  protocol too. This allows amavisd-new to act as a LMTP-to-LMTP content
  filter, possibly being inserted between MTA and a LMTP-based mail delivery
  agent such as Cyrus (if checking of outgoing mail is not needed). LMTP is
  selected when the first field of a $*_method (such as $forward_method,
  $notify_method, $resend_method, $release_method, $*_quarantine_method)
  is a 'lmtp:'.

  Possible uses:
    $forward_method = 'lmtp:/var/imap/socket/lmtp';  # over a Unix socket
    $forward_method = 'lmtp:[]:24';  # over IPv4
    $forward_method = 'lmtp:[::1]:24';        # over IPv6

  If a Postfix 'lmtp' service is used to feed amavisd (instead of the
  more usual content filter feed through a service named 'amavisfeed' or
  'smtp-amavis'), make sure not to forget to limit the number of concurrent
  feeds to amavisd (e.g. lmtp_destination_concurrency_limit=15) to a value
  same (or less) than $max_servers, or limit the maxproc field in
  such as: 'lmtp unix - - n - 15 lmtp' .

  Note that invoking amavisd as a LMTP delivery agent has a disadvantage
  that outgoing mail is not being checked, so infected internal hosts are
  able to pollute the world. Also the pen pals feature is no longer useful,
  as it requires the information on previous outgoing mail to be present
  in a SQL database.

- a new command line option -i, it takes one argument which can be any
  string (an instance/personality name), which is then made available to
  amavisd.conf in a variable $instance_name (intended to be read-only);
  code in amavisd does not assign any semantics to this argument and does
  not use it for any purpose, it is purely intended for administrator's
  use in amavisd.conf if desired; this simple mechanism may facilitate
  running multiple instances of amavisd using a single configuration file,
  or to choose at startup time between amavisd personalities using the
  same config file;

  A possible usage is to start a test instance of amavisd while a production
  amavisd is still running, and letting a test instance listen on its
  dedicated TCP port number. Each server instance needs its own pid and
  lock files, its own TCP port number or socket name, and its own $db_home
  (nanny, cache, agent) unless bdb usage is disabled. A working directory
  may be shared or kept separate.

  An example to put by the end of amavisd.conf:

  if ($instance_name eq 'test') {
    $log_level = 5;
    $sa_debug = 1;
    $max_servers = 1;
    $TEMPBASE = "$MYHOME/tmp-am2";
    $pid_file  = "$MYHOME/home/";
    $lock_file = "$MYHOME/home/amavisd2.lock";
    $enable_db = 0;
    $inet_socket_port = [8888];  # listen on port 8888

  Start a test instance:
    # amavisd -i test debug
  and submit a test message to it, e.g.:
    $ mini_sendmail \
        -s127.0.0.1 -p8888 <0.msg

- policy banks now contain a new key 'originating', which generalizes
  a previously hard-wired policy bank MYNETS. It is a boolean variable,
  turned on automatically in the currently loaded policy bank when a
  smtp client's IP address matches @mynetworks_maps, to retain full
  compatibility with existing setups. When a new policy bank is loaded
  over a current one, the new policy bank may also modify the 'originating'
  key - a typical use is to turn it on by a policy bank activated by mail
  submission from authenticated roaming users (SASL/AUTH), so that such
  users are treated as locals (originating mail) even though their IP address
  does not match a @mynetworks_maps list of lookup tables.

  The current value of variable 'originating' is now the only control to
  some macros or decisions, which were previously controlled implicitly
  by a @mynetworks_maps match. These are:
  * macro %l (letter ell) now directly corresponds to the current value
    of the 'originating' variable (returning a '1' or an empty string);
  * some statistics counters differentiate between 'Inbound' and 'Internal'
    mail directly based on the value of the 'originating' variable
    (applies to mail with local recipients, otherwise it is 'Outbound');
  * pen pals is skipped for senders claiming to be from a local domain, but
    'originating' is false (i.e. unauthorized foreigners faking your domain);
  * only mail with 'originating' flag will be eligible for DKIM signing
    (starting with version 2.6.0);
  * there may be other uses in the future, so it is prudent to keep
    @mynetworks_maps and @local_domains_maps configured correctly, and (when
    appropriate) turn on the 'originating' flag for mail that is supposed to
    be treated as originating from internal or authorized roaming users;

  Example use:

    $interface_policy{'10026'} = 'ORIG';

    $policy_bank{'ORIG'} = {   # mail originating from our users
      originating => 1,  # declare that mail was submitted by our smtp client
      allow_disclaimers => 1,  # enables disclaimer insertion if available
      virus_admin_maps => ["virusalert\@$mydomain"],
      spam_admin_maps  => ["virusalert\@$mydomain"],
      warnbadhsender   => 1,
      # forward to a smtpd service providing DKIM signing service
      forward_method => 'smtp:[]:10027',
      # force MTA conversion to 7-bit (e.g. before DKIM signing)
      smtpd_discard_ehlo_keywords => ['8BITMIME'],
      bypass_banned_checks_maps => [1],  # allow sending any file names & types
      terminate_dsn_on_notify_success => 0,  # don't remove NOTIFY=SUCCESS opt.

- make it possible for a virus scanner to derate an infection report
  to a spam report, contributing to spam score and to spam report/status.
  A new configuration variable @virus_name_to_spam_score_maps
  (also member of policy banks) can turn a reported virus name
  into a spam score. Its default setting is:

  @virus_name_to_spam_score_maps =
    (new_RE( [ qr'^(Email|HTML)\.(Phishing|Spam|Scam[a-z0-9]?)\.'i => 0.1 ],
             [ qr'^(Email|Html)\.Malware\.Sanesecurity\.'        => undef ],
             [ qr'^(Email|Html)(\.[^., ]*)*\.Sanesecurity\.'     => 0.1 ],
           # [ qr'^(Email|Html)\.(Hdr|Img|ImgO|Bou|Stk|Loan|Cred|Job|Dipl|Doc)
           #       (\.[^., ]*)* \.Sanesecurity\.'x => 0.1 ],
             [ qr'^(MSRBL-Images/|MSRBL-SPAM\.)'   => 0.1 ],

  and can be replaced in amavisd.conf.
  To disable the feature assign an empty list to the configuration variable:

    @virus_name_to_spam_score_maps = ();

  When a virus scanner returns names of viruses, and all provided names are
  matched by the @virus_name_to_spam_score_maps, and no other virus scanner
  has anything more sinister to report, then a message is _not_ flagged
  as a virus, but a corresponding spam score is contributed to other
  spam results as returned by a normal spam scan by SA. All the usual
  spam rules are then followed. Phishing fraud as indicated by ClamAV
  is now by default treated as spam, and no longer as a virus.

  The log can now show entries like:

    amavis[26733]: (26733-03-2) Turning AV infection into a spam report:
      score=0.1, AV:HTML.Phishing.Auction-289=0.1

    amavis[26733]: (26733-03-2) adding SA score 38.628 to existing 0.1
      from an earlier spam check

    amavis[26733]: (26733-03-2) Blocked SPAM, ... Hits: 34.728, ...
      Tests: [AV:HTML.Phishing.Auction-289=0.1, ... L_AV_Phish=14, ...]

  The information is also available to SA rules in a form of a synthetic
  header field X-Amavis-AV-Status which will be seen by SA only (not inserted
  into passed or quarantined mail). One has a choice to adjust scores either
  in the @virus_name_to_spam_score_maps in amavisd.conf, or by providing rules
  to match on the provided header field. Doing it by SA rules has an advantage
  of letting other SA rules contribute their score points, possibly preventing
  a false positive of a ClamAV rule, or pushing score even higher for a clean
  bounce suppression. It also allows auto-whitelisting in SpamAssassin to
  account for these score points. In addition, it makes more sense when checks
  are cached and result reused later for some other message with the same
  contents in body.

  Here is one example of such SA rules (some long lines are wrapped,
  these should be unwrapped before placing them into

    header L_AV_Phish      X-Amavis-AV-Status =~ 
    header L_AV_SS_Phish   X-Amavis-AV-Status =~ 
      m{\b(Email|Html)\.Phishing(\.[^., ]*)*\.Sanesecurity\.}m
    header L_AV_SS_Scam    X-Amavis-AV-Status =~ 
      m{\b(Email|Html)\.(Scam[A-Za-z0-9]?)(\.[^., ]*)*\.Sanesecurity\.}m
    header L_AV_SS_Spam    X-Amavis-AV-Status =~
        (\.[^., ]*)*\.Sanesecurity\.}m
    header L_AV_SS_Hdr     X-Amavis-AV-Status =~ 
      m{\b(Email|Html)\.Hdr(\.[^., ]*)*\.Sanesecurity\.}m
    header L_AV_SS_Img     X-Amavis-AV-Status =~ 
      m{\b(Email|Html)\.(Img|ImgO)(\.[^., ]*)*\.Sanesecurity\.}m
    header L_AV_MSRBL_Img  X-Amavis-AV-Status =~ m{\bMSRBL-Images/}m
    header L_AV_MSRBL_Spam X-Amavis-AV-Status =~ m{\bMSRBL-SPAM\.}m

    score  L_AV_Phish      14
    score  L_AV_SS_Phish   -3
    score  L_AV_SS_Scam    8
    score  L_AV_SS_Spam    8
    score  L_AV_SS_Hdr     6
    score  L_AV_SS_Img     3.5
    score  L_AV_MSRBL_Img  3.5
    score  L_AV_MSRBL_Spam 6

- added a new concept of a 'blocking contents category', which in most cases
  corresponds to a familiar 'main contents category' (the highest ranking
  category of contents pertaining to a message, e.g. virus, blocked, spam,
  spammy, bad header ...).  The difference between the two arises when
  recipients are declared to be 'lovers' of some higher-ranking contents,
  or when a higher ranking contents category has its *_destiny set to D_PASS.

  For example: a message contains a banned part, but is also spam
  and may even have a bad header. Its contents categories are (simplified):
  CC_BANNED, CC_SPAM and CC_BADH, in this order. The main contents
  category of a message is CC_BANNED, which usually is also a reason
  for blocking a message, yielding a blocking ccat to also be CC_BANNED.

  But if some recipient is banned_files_lover (or if $final_banned_destiny
  is set to D_PASS), then the main ccat remains to be CC_BANNED, but the
  blocking ccat is CC_SPAM, i.e. the next in the list which is responsible
  for actually blocking the mail. If recipient would also be a spam lover,
  the blocking ccat might be CC_BADH (if $final_bad_header_destiny were
  not D_PASS);

  If a message is not being blocked, the 'blocking contents category'
  (i.e. a blocking_ccat attribute of a per-message or a per-recipient object)
  remains empty (undefined). For convenience some internal routines
  and some new macros fall back to showing the main contents category
  in this case.

  Almost all processing decisions, DSN, notification assembling, quarantining,
  logging etc. is now based on 'blocking contents category' when a message
  is being blocked, and on 'main contents category' (as before) when a
  message is not being blocked.

  There is a new macro 'ccat' which is useful in notification and logging
  templates, which can query the blocking contents category, as well
  as a main contents category. It provides access to information that
  was formerly available through macros ccat_maj, ccat_min, ccat_name,
  plus access to additional information. Macros ccat_maj, ccat_min and
  ccat_name are still available, but their use is deprecated, as their
  functionality has been incorporated into the new macro 'ccat'.

  Macro 'ccat' takes two optional fixed-string arguments, which are
  interpreted case-insensitively. In their absence it expands to a
  string "(maj,min)" which shows a major and a minor contents category
  number of a blocking ccat for a blocked message, and of a main contents
  category for a passed message.

  The first argument specifies which attribute of a ccat is to be provided,
  the second argument specifies whether a main or a blocking contents
  category is to be consulted:

   The first argument may be any of the following strings:
     name   ... provide a human-readable name of a ccat (%ccat_display_names)
     major  ... provide a number: a major contents category,
                values correspond to CC_* constants in the program
     minor  ... provide a number: a minor contents category, often a 0
     <empty>... empty argument (also a default) results in a string "(maj,min)"
     is_blocking   ... '1' if blocking_ccat is true (message is being blocked),
                        or an empty string when a message is being passed;
     is_nonblocking .. the opposite: '1' if blocking_ccat false, '' otherwise
     is_blocked_by_nonmain .. '1' if blocking_ccat is true
                       _and_ is different from a main contents category;

   The second argument may be any of the following strings:
     main   ... provide information on main contents category
                when asked for name/major/minor/<empty>
     blocking.. provide information on blocking contents category if it exists,
                otherwise it falls back to providing info on main ccat;
                this is also a default in the absence of this argument;

  For illustration, instead of a former call [:ccat_maj] use [:ccat|major] ,
  instead of [:ccat_min] use [:ccat|minor], and instead of [:ccat_name]
  please use [:ccat|name] .  For more examples please consult the default
  templates, glued to the end of file 'amavisd'.

- when amavisd-nanny is given an invalid command-line argument
  it now shows 'Usage: ...' as well as a legend for process states;

- amavisd-nanny enhanced and new process-state instrumentation added to
  amavisd daemon; previously only busy/idle states of child processes were
  shown in amavisd-nanny output, now a more detailed process state can be
  shown by setting a new verbosity control configuration variable
  $nanny_details_level to a higher than a default value of 1, e.g. to 2;

  The following characters in amavisd-nanny bars represent amavisd child
  process states as follows, in the shown order of events:

    A  accepted a connection
    b  begin with a protocol for accepting a request
    m  'MAIL FROM' smtp command started a new transaction in the same session
    d  transferring data from MTA to amavisd
    =  content checking just started
    D  decoding of mail parts
    V  virus scanning
    S  spam scanning
    P  pen pals database lookup and updates
    r  preparing results
    Q  quarantining and preparing/sending notifications
    F  forwarding mail to MTA
    .  content checking just finished

  A nanny bdb database has changed in an incompatible way, so older
  versions of amavisd-nanny would complain about contents of a new
  database. Backward compatibility is retained: new version of
  amavisd-nanny is able to deal with a database from older versions
  of amavisd-new. There is no need for conversion, a new database
  is created on each amavisd restart.

  Note that a history of process states is _not_ maintained in a nanny
  database, but only in a running amavisd-nanny, which is why a just-started
  amavisd-nanny can not show previous states of processes from time before
  amavisd-nanny was started - a '=' is shown instead. A display eventually
  catches up and all newly-entered states are shown correctly.

- snmp-like database can now also store 64-bit counters data type,
  amavisd-agent utility modified accordingly;

- amavisd-agent utility and amavisd daemon enhanced to provide and to display
  cumulative elapsed time by sections; currently only some of the more
  important sections have been instrumented, e.g.:

    TimeElapsedReceiving             10631 s      0.608 s/msg (InMsgs)
    TimeElapsedDecoding                970 s      0.056 s/msg (InMsgs)
    TimeElapsedVirusCheck              629 s      0.036 s/msg (InMsgs)
    TimeElapsedSpamCheck             75866 s      4.341 s/msg (InMsgs)
    TimeElapsedPenPals                2150 s      0.123 s/msg (InMsgs)
    TimeElapsedSending                2231 s      0.128 s/msg (InMsgs)
    TimeElapsedTotal                 94709 s      5.419 s/msg (InMsgs)

  Don't be surprised if the total elapsed time exceeds amavisd uptime,
  10 processes progressing slowly for 5 seconds each will accumulate
  50 seconds of reported elapsed time. The average seconds-per-message
  figure as reported in the last column makes more sense;

- amavisd-agent utility and amavisd daemon enhanced to provide and to
  display cumulative mail sizes in bytes, and additional message counters
  based on outbound/inbound/internal mail direction:
  message counts:
    InMsgs              all mail received by amavisd (as in previous versions);
    InMsgsOutbound      at least one recipient NOT in @local_domains_map;
    InMsgsInternal      at least one recipient in @local_domains_map,
                        and client IP address (submitter) in @mynetworks_maps;
    InMsgsInbound       at least one recipient in @local_domains_map,
                        and client IP address NOT in @mynetworks_maps;
  message sizes:
    InMsgsSize          total mail size in bytes as received by amavisd;
    InMsgsSizeOutbound  as above, but with at least one recipient
                        NOT in @local_domains_map;
    InMsgsSizeInternal  at least one recipient in @local_domains_map,
                        and client IP address in @mynetworks_maps;
    InMsgsSizeInbound   at least one recipient in @local_domains_map,
                        and client IP address NOT in @mynetworks_maps;
  Note that a mail with multiple recipients can be both internal and outbound.

  P.S. in later versions a hard-wired testing against a @mynetworks_maps
  list is replaced by testing value of a boolean variable $originating.

  Example output:

    InMsgsSize                       4332MB  109MB/h 100.0 % (InMsgsSize)
    InMsgsSizeInbound                3102MB   78MB/h  71.6 % (InMsgsSize)
    InMsgsSizeInternal                554MB   14MB/h  12.8 % (InMsgsSize)
    InMsgsSizeOutbound                816MB   21MB/h  18.8 % (InMsgsSize)

- new configuration variables $always_bcc and %always_bcc_by_ccat, also
  members of policy banks, allow adding one extra envelope recipient to
  each message, either regardless of contents ($always_bcc), or selectively
  based on contents category. For example:

    $always_bcc = '';

  or selectively based on contents category:

    $always_bcc_by_ccat{+CC_CLEAN} = '';
    $always_bcc_by_ccat{+CC_VIRUS} = '';

  or as a member of policy banks:

    $policy_bank{'MYNETS'} = {
      always_bcc_by_ccat => {
        CC_BADH,    '',
        CC_CLEAN,   '',
        CC_CATCHALL, undef,

- amavisd-nanny and amavisd-agent utilities now recognize an optional
  command-line option:  -w <wait> , where the specified value is time
  in seconds between re-displays. The default interval is 2 seconds
  for amavisd-nanny, and 10 seconds for amavisd-agent as before.
  The specified interval time may be fractional;

- macro 'useragent' can accept an optional argument: a string 'name' or 'body',
  restricting the information to be returned as follows: macro 'useragent'
  returns 'User-Agent: ...' or 'X-Mailer: ...' header field from a message
  (whichever is present, or empty); an optional argument specifies whether:
  an entire field is to be returned (empty or unrecognized argument),
  or just a field name (argument: 'name'), e.g. 'X-Mailer';
  or just a field body (argument 'body'),  e.g. 'Thunderbird_1.5.0.9';

- interfacing to Mail::ClamAV (a perl module to a clamav library) now
  performs processing in a subprocess to prevent bugs in external library
  from bringing down amavisd process, and to prevent virtual memory of
  an amavisd child process from expanding uncontrollably - at the expense
  of additional 20..30 ms for a fork;

- extended AM.PDP protocol with an attribute 'policy_bank' which may be used
  in a client's request to require loading additional policy banks, e.g.:
  Its value is a comma-separated list of policy bank names. Names of
  nonexistent banks are silently ignored, so are leading and trailing spaces
  and TABs around each name. The order of policy bank loading generally
  follows the order in which information about a message were obtained:
    - interface- or socket-based policy banks (when MTA connects to amavisd);
    - MYNETS (when client's IP address becomes known);
    - the list of policy bank names as specified in a
      'policy_bank' attribute of AM.PDP protocol, comma-separated;
    - MYUSERS (when sender's e-mail address becomes known);

- added a field 'Final-Log-ID' to a DSN report (RFC 3464), which will
  provide information on log_id and mail_id, e.g. '77790-10-3/uez9wtcVNTO5'
  in a standard way, much like the 'Our internal reference code for your
  message is: ...' in a DSN plain text part;

- added mapping from 'RIFF...animated cursor' to ['movie','ani']
  in $map_full_type_to_short_type_re, to facilitate blocking animated
  cursors (Microsoft Windows ANI header stack buffer overflow is being
  actively exploited); by Henrik Krohns;

- add support for 7-Zip archives if external utility program 7z
  is available (under names 7zr, 7za or 7z); suggested by Bob Marcan;

- add configurable global settings $min_servers, $min_spare_servers, and
  $max_spare_servers (all undefined by default, see Net::Server::PreFork
  documentation for their semantics), pass them to Net::Server at startup
  time (complementing the usual $max_servers setting) and allow to choose
  between Net::Server personalities Net::Server::PreForkSimple and
  Net::Server::PreFork - if $min_servers is defined the PreFork is chosen,
  otherwise the more usual PreForkSimple. The feature is mostly intended for
  use of amavisd as a pre-queue content filter, which is unsupported anyway.
  For normal post-queue use the PreForkSimple already does a good job.
  Based on a patch by Alexander 'Leo' Bergolth;

- internal: incompatibly changed order and indirection level of arguments
  to routines dealing with contents categories, and some of their names;

- new macro 'join', behaves like a Perl join function: the first argument
  is a separator string, remaining arguments are strings to be concatenated,
  with a separator string inserted at every concatenation point;

- macro 'dquote' (as used in a default log template to protect Subject)
  previously escaped double quotes with \, but missed to escape \ itself,
  making log parsing tricky;  also, as logging layer escapes \ by itself,
  the result was ugly and inconsistently parsable; new behaviour
  is to protect a double quote within a string by doubling it, so
  a [dquote|one"oops"two] now yields "one""oops""two", instead of
  "one\"oops\"two", which when logged showed as "one\\"oops\\"two";

- convenience: do not drop privileges early despite a command line option -u
  when an option -R is also specified with a non-empty (and non-slash) value,
  otherwise the requested chroot operation is not possible (root privileges
  are required for chrooting);

- version 2.4.3 introduced some substitutions of subject tag template strings:
  SCORE, REQD, YESNO and YESNOCAPS; this list is now extended with few more,
  to facilitate cross-host troubleshooting; the full list now consists of:

  _SCORE_     spam score (hits), same as macro %c
  _REQD_      tag2_level
  _YESNO_     score above tag2_level? 'Yes' or 'No'
  _YESNOCAPS_ same, but yields:       'YES' or 'NO'
  _HOSTNAME_  fqdn of this host ($myhostname),             same as macro %h
  _DATE_      rfc2822 timestamp of mail entering this amavisd,  as macro %d
  _U_         iso8601 UTC timestamp of mail entering this amavisd,    as %U
  _LOGID_     log id (am_id) as shown in the log, e.g. 58725-05-2,    as %n
  _MAILID_    mail_id as used in quarantine names, e.g. jaUETfyBMJHG, as %i

  See also README.customize for explanation of macros.

                                                           January 30, 2007
amavisd-new-2.4.5 release notes


- Recommended version of Convert::UUlib is 1.08 or higher
  to avoid processing of uninitialized data containing 'random' garbage.

  Note that a security hole in uulib which comes with Convert::UUlib 1.04
  and older is now (as of 2006-12-05) known to be exploitable:
  credits to Jean-Sébastien Guay-Leroux;

- will no longer reply to queries coming from low-numbered
  UDP ports below 1024 or from nfsd port 2049, and will ignore queries
  with nonce longer than 1024 character or containing characters outside
  of \040-\177 range to limit its usefulness as a potential reflector
  for an attacker from internal networks.


- now only binds to a loopback interface by default, instead
  of to all interfaces;  change $bind_addr in to ''
  if is running on a different host from amavisd or from
  other querying clients; suggested by Shaun T. Erickson and Mario Liehr;


- let exit when a pipe on stdin is closed (e.g. when p0f
  is killed or crashes), instead of entering a tight loop; reported by
  Justin Piszcz and Henrik Krohns;

- hard-blacklisting no longer skips quarantining when
  $spam_quarantine_cutoff_level is undefined (or is an empty string);

- restart timer after Sophie times out; previously the next attempt
  would run with no time limit; reported by Nick Leverton and
  Nicklas Bondesson;

- fix error reporting in open_on_specific_fd when POSIX::dup2 fails;
  thanks to Chris (decoder);

- fix signal handling in read_snmp_variables() and register_proc(),
  a signal could previously get lost (not re-signaled) if it occurred
  within these subroutines;

- fixed get_body_digest which incorrectly determined 7- or 8-bitness
  of mail header and body, setting body_type incorrectly (with only
  cosmetic ill-effects);

- fixed AM.PDP code to always provide a smtp-quoted form in angle brackets
  in 'delrcpt' and 'addrcpt' attributes of a response, i.e. in the same form
  as was received in 'sender' and 'recipient' attributes. The attribute value
  syntax is specified in RFC 2821 as 'Reverse-Path' (i.e. smtp-quoted form,
  enclosed in <>); previously enclosing angle brackets were missing in a
  server reply;

- documentation - amavisd.conf-default incorrectly stated that a default
  value for $prepend_header_fields_hdridx is 1;  actually the default is 0
  as correctly indicated in release notes; reported by Jo Rhett;


- qmail interfacing notice:
  MTA timeout for waiting on results from amavisd should be longer than
  $child_timeout (8 minutes by default) with some margin, setting MTA timeout
  to 15 or 20 minutes is usual. With qmail however the QMQP code in qmail
  has hard-coded timeouts set, 10 seconds for connect and 60 seconds for
  read/write. If amavisd processing takes longer than 60 seconds, the MTA
  drops connection and retries later, yet amavisd continues processing
  and eventually delivers a mail (with each MTA retry), causing repeated
  deliveries of the same message. The following patch by Eric Huss on
  the page:
  should be applied to qmail when interfacing it to a post-queue content
  filter. Problem researched by Nicklas Bondesson;

- better timeout handling in interface code to daemonized virus scanners
  like clamd, Sophie, Trophie: allow short time (10 s) for connect and
  for sending a request, then allow normal (long) time to collect results;
  keep evidence of the initial deadline on retries;

- prefer '7bit' as Content-Transfer-Encoding when attaching original message
  or its headers (message/rfc822 or text/rfc822-headers) to DSN or to a
  defanged mail, and only specify '8bit' when necessary;

- remove protecting the $ and @ characters in second argument
  of a regexp selector macro, it is unnecessary and confusing;

- macros %m, %r and header_field now return parsed and sanitized
  message IDs in header fields Message-ID, Resent-Message-ID. In-Reply-To,
  or References, void of CFWS (comments and FWS as specified by RFC 2822),
  through the use of new subroutine parse_message_id();

- when logging to SQL, the field msgs.message_id now contain just
  a message id, without CFWS and other garbage that might appear in
  a Message-ID header field; this facilitates queries, and pen pals
  matching of IDs in In-Reply-To or References header fields
  of a reply to an original Message-ID;

- updated $map_full_type_to_short_type_re to avoid mapping file(1) result
  'MS-DOS executable (built-in)' to types 'exe-ms' and 'exe'; the file(1)
  utility generously declares any text file starting with LZ to be a
  'MS-DOS executable (built-in)';  thanks to Noel Jones, Jakob Curdes
  and Clifton Royston for troubleshooting;

- add X-Spam-* header fields to quarantined mail if spam score is at or
  above tag_level. Previously message needed to be recognized as spammy
  or spam (tag2 or kill level) in order to receive spam header fields
  in quarantined copy. This also makes it more consistent with adding
  such header fields to passed mail;  suggested by Michael Gaskins;

- add X-Amavis-OS-Fingerprint header field to quarantined mail;

- header field X-Spam-Score in a passed or quarantined mail now reflects
  score boost even when SA score is unknown (e.g. when SA was not called),
  and reflects white and blacklisting by pushing score to 0 or 64, to
  make it consistent with a bar size in X-Spam-Level header field;

- resignal "timed out" after (almost) every eval {} which has no subsequent
  call to prolong_timer() to ensure we do not continue running with
  disabled timer. Exceptions are DESTROY and END handlers, and code which
  handles timer in some other way (e.g. by keeping evidence of a deadline);

- for the purpose of looking up client IP address in @mynetworks_maps,
  treat unknown/unavailable IP address as;  this allows treating
  directly submitted mail on the MTA host (not submitted through SMTP) as
  coming from IP address (i.e. "This" Network - according to RFC 1700);

  Note that this is indistinguishable from other reasons when IP address
  is not made available to amavisd, e.g. when smtp_send_xforward_command
  option in Postfix smtp service is not enabled, which is why the default
  setting of @mynetworks does not include a network to prevent
  unintentionally loading a MYNETS policy bank.

  One should add to a @mynetworks list only when XFORWARD is known
  to work and if some software on the MTA host is submitting its mail to MTA
  directly, e.g. through a sendmail command, and MYNETS policy bank loading
  is needed for proper processing of such mail;

- report a more informative message when a file(1) utility fails to produce
  useful results: joins exit status with a parsing report into one message;
  thanks to Andres, whose file(1) utility was crashing with SEGV;

- consistency: rearrange implicitly adding $X_HEADER_TAG to a hash
  %allowed_added_header_fields so that it is possible to turn off
  insertion of $X_HEADER_TAG header field by turning off associated key in
  %allowed_added_header_fields even when $X_HEADER_TAG is explicitly defined;

- let %allowed_added_header_fields also control insertion of header fields
  into quarantined message;

- amavisd-nanny now displays a title line indicating the semantics of columns;

- Courier patch: ensure the information is stored to newly introduced
  recip_addr_smtp and sender_smtp object attributes, which are needed
  to preserve pristine address forms for DSN and ORCPT use and for logging;
  a patch by Martin Orr;

- qmqpqq (qmail): ensure the information is stored to newly introduced
  recip_addr_smtp and sender_smtp object attributes;

- qmail patch now activates line-by-line sending to qmail to avoid qmail bug
  ('bare LF' reported when CR and LF are separated by a TCP packet boundary);

- tighten a regexp on matching a p0f fingerprint for Windows XP to avoid
  matching 'Windows XP SP1+, 2000 SP3';  suggested by Michael Scheidell;

- updated AV entry for CentralCommand Vexira (vascan):
  removed hard-coded option '--vdb';  by Brian Wong;

- internal: move code dealing with a SA call to a dedicated
  subroutine call_spamassassin;

- internal: provide new routines to collect scalar and structured results
  from a subprocess (collect_results, collect_results_structured) and
  take advantage of them in decoding, in AV and in dspam interface routines,
  unifying code and providing results size sanity limit and consistent
  killing of runaway external programs;

- experimental: taking advantage of the above, make it possible to run SA in
  a spawned process, requested by setting a new config variable $sa_spawned
  to true (it is off by default); benefits are that a mainstream child process
  can not be brought down by potential processing problems in SA or its
  external modules, and timeouts are handled cleanly by a calling process;
  downside is an increase of process count (worst case: doubled), with
  corresponding increase in memory footprint, plus about 20 .. 30 ms
  of additional processing time for each call to SA;

- added a tuning tip on buffer sizes to README.sql for MySQL with InnoDB,
  by Wayne Smith;

- updated URL of Sophie AV scanner;

                                                          November 20, 2006
amavisd-new-2.4.4 release notes


- PostgreSQL quarantining: data type of field quarantine.mail_text should
  be 'bytea' (instead of 'text') to allow storing arbitrary octets without
  associating them with a character set.  See below for a conversion of an
  existing database. Similarly with MySQL the data type should be 'blob'.

- Note: in a sendmail milter setup with Petr Rehor's helper program
  amavis-milter, one should set:  $prepend_header_fields_hdridx = 1;
  when dk or dkim signing milters are used in the same setup.
  See below for details.


- do_ascii: fix a bug where timer was not restored after decoding of a
  textual mail part, so a timeout for subsequent decoding operations
  on the same message was limited to 10 seconds (and to 30 seconds
  for a call to SpamAssassin), regardless of $child_timeout setting;

- don't call PerlIO::get_layers with Perl 5.8.0, the function was
  introduced with 5.8.1; reported by Joel Nimety;

- avoid deep recursion in evaluating a regular expression in header checks
  which caused very slow testing for presence of a all-whitespace lines
  in folded header fields for degenerate cases of header; the inefficient
  expression was introduced with amavisd-new-2.4.0; reported and a sample
  provided by Kai Risku;

- when spam above kill level is to be passed and spam defanging is enabled,
  SA summary was inserted twice (once for mail contents category being
  CC_SPAMMY and once for CC_SPAM), fixed. Reported by Gary V and MHahnen;

- when logging directly to a file, do create a log file if it does
  not already exist; (bug introduced with 2.4.3)

- make sure a quota limit is untainted when it is given as a command line
  parameter to external TNEF decoder; reported by MK;

- updated Courier patch to loosen up socket protection and allow
  group write access to the socket; reported by Bill Taroli;

- SQL logging: cleanly chop an UTF-8 octet sequence according to RFC 3629
  (avoid truncating character octet sequence tail) when Subject, From or
  Message-Id header field is longer than 255 characters;

- PostgreSQL: when storing mail text to a quarantine use pg_type=PG_BYTEA
  attribute on a field 'quarantine.mail_text';  previously the following
  error could be reported:

    451 4.5.0 Storing to sql db as mail_id ... failed:
      writing mail text to SQL failed: Error closing, flush:
      sql inserting text failed,
      sql exec: err=7, 22P02, DBD::Pg::st execute failed:
      ERROR: invalid input syntax for type bytea

- updated documentation in README.sql to suggest using data type 'bytea'
  instead of inappropriate data type 'text' for a field quarantine.mail_text

  To convert an existing table (when quarantining to SQL) please use:

    ALTER TABLE quarantine ALTER mail_text TYPE bytea
      USING decode(replace(mail_text,'\\','\\\\'),'escape');

  If conversion of data type for 'quarantine.mail_text' is not done,
  the following error will be reported when storing a message to a SQL
  quarantine is attempted:

    TROUBLE in check_mail: quar+notif FAILED:
      temporarily unable to quarantine:
      451 4.5.0 Storing to sql db as mail_id ... failed:
      writing mail text to SQL failed: Error closing, flush:
      sql inserting text failed, sql exec: err=7, DBD::Pg::st execute failed:
      ERROR: column "mail_text" is of type text but expression is of type bytea
      HINT:  You will need to rewrite or cast the expression

  If converting quarantine table is not desirable or possible in a short term,
  it is possible to continue use existing SQL quarantine table without
  conversion by specifying the following in amavisd.conf:

    $sql_clause{'ins_quar'} =
      "INSERT INTO quarantine (mail_id, chunk_ind, mail_text)".
      " VALUES (?,?,encode(?,'escape'))";

    $sql_clause{'sel_quar'} =
      "SELECT decode(mail_text,'escape') FROM quarantine".
      " WHERE mail_id=? ORDER BY chunk_ind";

  This will allow PostgreSQL to convert data types on-the-fly, converting
  octets (any byte) into escaped text, and vice versa when releasing from
  a quarantine;

  Problem reported by Justin Hillyard, correct data type suggested by
  Nikola Milutinovic;

- MySQL: updated documentation in README.sql to suggest using data
  type 'blob' instead of inappropriate data type 'text' for a field
  quarantine.mail_text. To convert an existing table please use:
    ALTER TABLE quarantine CHANGE mail_text mail_text blob;
  Seems like MySQL does not complain on incompatibility between provided
  data type and a data type of a field in table, but there are reports that
  MySQL may silently truncate data which it finds violating character set
  constraints, so conversion to 'blob' is highly recommended. Truncation
  of quarantined message at an 8-bit character reported by Lubor Kolar.


- limit recursion in MIME::Parser to $MAXFILES to prevent MIME parser from
  fully traversing degenerate cases of broken MIME messages which can take
  excessive amount of time and memory; reported and a sample provided
  by Joshua Goodall, solution suggested by David F. Skoll, and requires
  a parser method max_parts(), available in MIME::Parser 5.417 or later;

- check for already running daemon at startup time, preventing a user
  mistake of trying to start another instance of the daemon without
  stopping the currently running process; suggested by Jo Rhett;

- keep sender and recipient addresses in original unparsed form (in addition
  to an internal form) to be able to always provide exact original address
  in delivery status notifications, in ORCPT, and when appending extensions
  in a milter setup (AM.PDP), which requires exact matching to the original
  form (without stripping route and without fixing poorly SMTP-quoted
  address forms);

- new configuration variable %allowed_header_tests, also member of policy
  banks, allows for selectively disabling some of the header checks,
  e.g. checks for non-encoded 8-bit characters. The %allowed_header_tests
  hash contains all available header test names as its keys by default
  (with a value of true);  removing a key, or setting its value to false,
  disables a test, e.g.:
    $allowed_header_tests{'8bit'} = 0;
    $allowed_header_tests{'missing'} = 0;
  Currently available keys (i.e. test names) are:
    other mime 8bit control empty long syntax missing multiple
  each corresponding to its own minor contents category of CC_BADH;

    ccat test
    min  name      description
    ---  -------   -----------
      0  other     (catchall for everything else, normally not used)
      1  mime      Bad MIME (sub)headers or bad MIME structure
      2  8bit      Invalid non-encoded 8-bit characters in header
      3  control   Invalid control characters in header (CR or NUL)
      4  empty     Folded header field made up entirely of whitespace
      5  long      Header line longer than RFC 2822 limit of 998 characters
      6  syntax    Header field syntax error
      7  missing   Missing required header field
      8  multiple  Duplicate or multiple occurrence of a header field
    ccat min:  minor contents category under a major category CC_BADH,
               available in templates as a macro ccat_min;
    test name: corresponding test name - a key in %allowed_header_tests;
    descr.:    description of a header test or MIME subheaders/structure test;

- timing report has a couple of new entries to facilitate troubleshooting:
  header checks section, separate entry for header and body digests,
  check_mail initialization, entries 'SMTP greeting' and 'SMTP response';

- when exec in a forked process fails, call POSIX::_exit with exist status
  8 (ENOEXEC) instead of the more common 1 to make the failure more obvious;

- initialize logging earlier so that do_log may be called earlier during
  program startup; also log attempts to stop and to reload, including
  unsuccessful ones;

- avoid logging by a forked process before exec, when there is a chance the
  log file descriptor is in a range 0..2;

- sub run_command and run_command_consumer: distinguish between undefined
  and empty values of argument $stderr_to, undef now prevents reopening of
  file descriptor 2, making it possible for the caller to keep it attached
  to the current stderr; this is useful when run_command is called by the
  master process before logging has been configured;

- SQL: explicitly call DBI::bind_param to be able to specify data types
  of values passed in @args to Amavis::Out::SQL::Connection::execute;

- bump up buffer size from 16 kB to 64 kB in some cases of copying data
  from/to a pipe, mostly to reduce the amount of logging;

- av scanner update: 'FRISK F-Prot Antivirus' entry modified to recognize
  name of a 'security risk' result, thanks to Michael Renner;

- in a commented-out code providing a qmail CF/LF bug workaround, replaced
  $smtp_handle->datasend by $smtp_data_fh->print, which is more efficient
  in a line-by-line writing mode needed by qmail; thanks to Ronald Vazquez;

- in a (banning) check for double extensions allow for whitespace around the
  second filename extension (files amavisd.conf and amavisd.conf-sample);
  based on a sample provided by Patrick T. Tsang;

- setting $max_requests to 0 disables the limit, process will not be replaced
  based on the number of requests it has completed (but may still be replaced
  for other reasons); primarily intended for testing;

- bump up a default value for $max_requests from 10 to 20 to match the
  suggested/example value in amavisd.conf-sample;

- AM.PDP/milter setup: new configuration setting $prepend_header_fields_hdridx,
  also a member of policy banks, with a default value of 0. It is used as an
  argument hdridx in an AM.PDP attribute 'insheader' which in a milter setup
  is passed on as an argument hdridx to a smfi_insheader call. The value
  of $prepend_header_fields_hdridx only affects AM.PDP protocol and only if
  $append_header_fields_to_bottom is false (it is false by default). If more
  than one milter is used, all milters should be inserting their header fields
  at the same index (all prepending or appending, avoiding insertion in the
  middle of a header), otherwise the resulting order of header fields in
  a modified header becomes surprising, and in combination with signing
  milters like DKIM or DK the signature verification will most likely fail.
  The default value of 0 is normal and useful in combination with other
  content-checking milters. Signing milters like dkim-milter and dk-milter
  insert their header at index 1 (just below the new Received header fields),
  and when amavisd-new with Petr Rehor's helper program amavis-milter
  is used as a milter along with dkim-milter or dk-milter, the value of
  $prepend_header_fields_hdridx MUST BE SET TO 1, otherwise the generated
  signature will fail verification at the receiving site!

  Discussion: when sendmail calls its milters, its Received header field
  is not yet created and passed on to milters, yet it is already counted as
  one header field for the purpose of smfi_insheader hdridx interpretation.
  When a milter wants to prepend its header field(s), specifying hdridx of
  0 does prepend its header fields above the yet-to-be-inserted Received
  header field as expected, and specifying 1 inserts its header field(s)
  just below the yet-to-be-inserted Received header field. If some milters
  in a chain specify 0 and others a 1 it affects the final order of inserted
  header fields in unexpected ways. It would be natural to always prepend
  fields with an index 0, but for signing milters like dk-milter this is
  not acceptable, as it would be expected to include a not-yet-available
  Received header field in its signature. For this reason signing milters
  like dkim-milter and dk-milter insert their header fields (signature)
  at index 1, and if amavisd-milter wants to coexist in such a setup,
  it must also insert its header fields at index 1.

  The conclusion: when amavisd (with its helper program) is used in a milter
  setup along with other milters, it should use the same hdridx value as
  other milters, which in case of signing dkim-milter and dk-milter is 1.
  If there are no such milters, either a 1 or a 0 would do, although a value
  of 0 produces a more natural order of header fields, matching that of
  a post-queue content filtering setup. See threads:

                                                         September 30, 2006
amavisd-new-2.4.3 release notes

For new instructions on setting up DKIM and DomainKeys
with Postfix and amavisd-new please see:


- in a sendmail milter setup using AM.PDP protocol (e.g. by Petr Rehor's
  amavisd-milter), inserted header fields are now prepended to mail header
  by a new AM.PDP protocol attribute 'insheader', so an upgrade of the
  milter helper is needed to support the change (e.g. amavisd-milter 1.1.3),
  otherwise header field insertions will be ignored;

- due to enhanced header checks (checking for missing required fields and
  checking for multiple occurrences of header fields which are allowed to
  occur only once), new cases of invalid mail may pop up in bad headers
  category, e.g. mail posted by MUA like Eudora;

- a string %i in quarantine filename templates (such as
  $spam_quarantine_method) now uses a 'T' as a date/time separator,
  e.g. 20060814T173846 (conforming to iso8601), instead of a former '-';

- when loading a policy bank, its entries of type hash are now merged with
  existing hash, a key at a time; previously a newly loaded hash replaced
  a previous one entirely;


- fixed a bug (introduced with amavisd-new-2.4.0): when receiving mail
  from MTA through a LMTP protocol (not SMTP) and with D_BOUNCE as a
  final*destiny setting, a suppressed non-delivery notification (e.g.
  spam above cutoff_level) did not turn LMTP status into a success,
  so an undesired bounce was generated by MTA in a post-queue filtering
  setup, contributing to excessive bounce backscatter; reported by
  Michael Scheidell, thanks to Gary V for analysis;

- bug fix to amavisd-release: a regexp needs to be relaxed to allow
  quarantine names like Y/spam-Y5y7A3J5r2Ax.gz, reported by Rob Chanter;

- fix a bug in LDAP lookups which could lead to an infinite loop while
  expanding %m in the filter; reported by Petr Vokac;

- add "LOCAL_STATE_DIR => '/var/lib'" to the SA object initialization
  for versions of SA 3.1.4 or older, so that SpamAssassin would see
  additional rules provided by sa-update and placed to its default location;
  the SA 3.1.5 provides its own default so this becomes unnecessary;

- bug fix: don't reject mail when mail size restriction is in force,
  the limit is exceeded, and $final_destiny_by_ccat{+CC_OVERSIZED}
  is not D_REJECT;

- treat blacklisting as high spam score when considering suppressing
  quarantining (@spam_quarantine_cutoff_level_maps) or suppressing sending
  a DSN (@spam_dsn_cutoff_level_maps);

- calling do_quarantine() multiple times on the same message would accumulate
  header edits from each invocation, fixed;  (such situation can only happen
  with a modified program);

- when defanging mail or releasing mail from a quarantine, with a goal of
  not breaking DKIM Sender Signing Practices (SSP) and DomainKeys policy,
  do not copy existing Sender header field to a new header, and insert
  our own Sender field (configurable by %hdrfrom_notify_recip_by_ccat);

- explicitly set PerlIO layer to ":bytes" on a temporary file handle for
  email.txt (just in case); based on a problem report by Alexander Schäfer;

- in a string produced by a macro %c remove a decimal dot if score happens
  to be an integer;

- reduce $sa_mail_body_size_limit from 512 kB to 400 kB in amavisd.conf
  and amavisd.conf-sample for the time being, while the SA folks work
  (MS Outlook Express seems to be chopping long mail in approx 500 kB chunks);

- another workaround for Perl taint bug: IO::Handle::_open_mode_string
  taints the $1 when mode string to IO::File::open is '+<', use O_RDWR
  instead; thanks to Ryan Frantz;

- abort if a specified syslog facility name is unknown, instead of
  switching to LOG_DAEMON as before;

- change the code which selects defanging so that defanging is triggered
  if any applicable contents category of a message chooses defanging;
  counterintuitive behaviour reported by Tapani Tarvainen;

- fix example in amavisd.conf-sample to use +CC_SPAM instead of CC_SPAM
  as a key to a hash, e.g. $final_destiny_by_ccat{+CC_SPAM}, otherwise Perl
  would implicitly turn CC_SPAM into a string when used in such a context.

  Note that any Perl expression syntax would do, as long as the argument
  does not look like a plain variable which receives implicit quoting;
  possibilities include $xx{&CC_SPAM}, $xx{+CC_SPAM}, $xx{CC_SPAM()},
  $xx{(CC_SPAM)} and similar; a more obvious &CC_SPAM is avoided because
  it prevents subroutine call inlining optimization in Perl;

- qmail: update amavisd-new-qmqpqq.patch to be compatible with Net::Server
  version 0.91 or later; thanks to mr from DBA Lab S.p.A.;

- AM.PDP protocol: change the order of attributes returned in an reply:
  delete and edit header fields before adding new header fields;
  problem of deleting just-inserted header fields in a sendmail milter
  setup reported by Petr Rehor;

- AM.PDP protocol change - with version 2 of the protocol the following
  changes to the protocol were made:
  * "version_server=2" is provided in a server response as the
    first attribute, older versions did not provide such attribute
    (assumed version on the server side was 1);
  * delheader and chgheader now stand in a response before insheader
    and addheader, assuming that milter MTA will execute these
    in the same order;
  * new attribute: "insheader=hdridx hdr_head hdr_body"
    (where hdridx as used by amavisd will always be 0 for now), making
    it possible to prepend header fields in a sendmail milter setup
    (instead of appending them, breaking compatibility with DomainKeys);
    problem noted by Adam Gibson and Petr Rehor;
  * new attribute: "quarantine=reason" to place message on hold or to a
    quarantine maintained by MTA, and supply a reason text (e.g. client
    may call smfi_quarantine milter routine); For future use - it is
    currently (2.4.3 or earlier) never used.


- turn UTF8 warnings into fatal errors by:  use warnings FATAL=>'utf8';

- reduced default $spam_check_negative_ttl to 10 minutes (from 30 minutes)

- when defanging, enforce 998 characters line length limitation imposed
  by RFC 2822 by truncating long lines and appending a "...";

- incompatible change: a string %i in quarantine filename templates
  (such as $spam_quarantine_method) now uses a 'T' as a date/time separator,
  e.g. 20060814T173846 (conforming to iso8601), instead of a former '-';

- enhance header checks with checking for missing required fields (Date, From)
  and with checking for multiple occurrences of header fields which are
  allowed to occur only once: Date, From, Sender, Reply-To, To, Cc, Bcc,
  Message-ID, Subject, In-Reply-To, References.  Added minor content types
  to major category CC_BADH: 7=missing, 8=multiple header field;

- use enhanced status code 5.7.0 instead of 5.7.1 for blocking spam
  and viruses, and use 5.6.0 for blocking mail with invalid header;

- macro SCORE now returns a single number instead of an explicit sum
  of SA score and boosts (contributed by soft-w/b-listing or pen-pals).
  When a single number is preferred in the log (Hits: ...), use this
  macro instead of %c in a log template:
    , Hits: [:SCORE]#
  based on a problem report by Ed Lucero;

- updated Panda pavcl AV entry to match the new version called
  'Panda CommandLineSecure 9 for Linux', thanks to Andrzej Kukula;

- updated AV scanner entries for ESET NOD32 for Linux Mail servers,
  ESET NOD32 for Linux File servers, F-Secure Antivirus for Linux servers
  and clamscan, thanks to Anders Norrbring:

- updated AV entry for Kaspersky AV version 5.5,
  thanks to Harrie Overdijk, Anders Norrbring and Gary V;

- 'reload' and 'stop' command line options now report a process id
  of the previous daemon that was killed;

- allow command-line option -d to specify a list of SA debug areas
  (also called 'facilities'); some useful examples:
    amavisd -d plugin,dkim,dk,spf         debug-sa
    amavisd -d auto-whitelist,bayes,learn debug-sa
    amavisd -d dcc,razor2,pyzor,util      debug-sa
    amavisd -d dns,uri,uridnsbl           debug-sa
    amavisd -d received-header,metadata   debug-sa
    amavisd -d rules,check                debug-sa

- substitute string _SCORE_ in spam_subject tag templates with actual score,
  _REQD_ with tag2 level, and _YESNO_ with 'Yes' (_YESNOCAPS_ with 'YES') if
  score is above tag2 level, and 'No' (or 'NO') otherwise; this is a quick-fix
  measure for often demanded feature; currently the mechanism is a simple
  string substitution and not a true macro expansion, so other macros are
  not currently available - more refined solution is expected for some future

- reshuffled the order of rules (with minor adjustments) in example files
  amavisd.conf and amavisd.conf-sample to make it easier to permit certain
  files within archives;

- introduce new variable @spam_dsn_cutoff_level_bysender_maps (also member of
  policy banks), complementing an existing by-recipient list of lookup tables
  @spam_dsn_cutoff_level_maps. The new variable serves to make it possible
  to trim down spam bounces to domains sending their own bounces with non-null
  return path (envelope sender address) and without DSN NOTIFY=NEVER option,
  but also to frequently abused domains, or to those sending marginal spam.
  When spam level exceeds either the @spam_dsn_cutoff_level_bysender_maps
  or the @spam_dsn_cutoff_level_maps level, (non)delivery status notification
  is suppressed even with $final_spam_destiny set to D_BOUNCE;

- introduce new variables $resend_method and $release_method (also members
  of policy banks), both are undefined by default.  If defined and nonempty,
  $resend_method overrides forward_method on forwarding a defanged mail,
  and $release_method overrides notify_method on releasing a message from
  quarantine. The $resend_method might be useful when a modified mail
  requires local DKIM or DomainKeys re-signing;

- added global configuration variables $sql_lookups_no_at_means_domain and
  $ldap_lookups_no_at_means_domain, both false by default. They control
  whether a database mail address field with no '@' character represents
  a local username, or a domain name. By default (value false) it indicates
  a username in SQL and LDAP lookups (but represents a domain in hash and
  acl lookups), so domain names in SQL and LDAP should be specified as
  '@domain'. Setting these to true will cause 'xxx' to be interpreted as
  a domain name, just like in hash or acl lookups, which may facilitate
  interoperability with databases from other applications;

- added global configuration variable $sql_quarantine_chunksize_max,
  which determines a maximum size (in bytes) for data written to a field
  'quarantine.mail_text' when quarantining to SQL. Must not exceed size
  allowed for a data type on a given SQL server (e.g. maximum size
  for data type 'blob' in MySQL is 65535 bytes). It also determines a
  buffer size in amavisd. Too large a value may exceed process virtual
  memory limits or just waste memory, too small a value splits large
  mail into too many chunks, which may be less efficient to process;
  defaults to 16384;

- added configuration variables @archive_quarantine_to_maps and
  $archive_quarantine_method, allowing for archival quarantine of all mail
  (configurable by recipient and by policy banks) regardless of its contents
  category. This archive is independent from other quarantining, i.e. if
  spam quarantining and archival quarantining are both enabled, two copies
  will be stored to quarantine. When quarantining for archive one has two
  choices: archive_quarantine would store all mail addressed to recipient,
  whereas enabling clean quarantine as in:
    $quarantine_method_by_ccat{+CC_CLEAN} = 'local:clean-%m';
    $quarantine_to_maps_by_ccat{+CC_CLEAN} = 'clean-quarantine';
  would quarantine only clean mail, no spam, no viruses, no banned, no badh.

  Note that logging to SQL has only one field to store quarantine location,
  so in case of multiple quarantine locations only the first is remembered.
  The usual logging however reports all quarantine locations with the main
  log entry.

- added a global configuration variable @additional_perl_modules, which
  is a list of additional Perl module names or absolute file names that
  should be compiled/executed (by calling 'require') at a program startup
  time by a master parent process, before chroot-ing and before changing
  UID takes place. Its purpose is to pre-load additional non-standard
  SpamAssassin plugins and similar modules that a standard SpamAssassin
  initialization would miss, causing them to be loaded later by each
  child process, which is inefficient and may not work in a chrooted
  process. Example:
    @additional_perl_modules = qw(
      String::Approx Net::HTTP Net::HTTP::Methods
      URI URI::http URI::_generic URI::_query URI::_server
      HTTP::Date HTTP::Headers HTTP::Message HTML::HeadParser
      HTTP::Request HTTP::Response HTTP::Status
      LWP LWP::Protocol LWP::Protocol::http
      LWP::UserAgent LWP::MemberMixin LWP::Debug
  Make sure these files are owned by root and not writable by unprivileged
  users such as amavis!

- added a global config variable $enforce_smtpd_message_size_limit_64kb_min,
  true by default; when true a rfc2822 requirement that a limit on mail size
  must not be below 64 kB is enforced, so that any specified limit below 64 kB
  is treated as 64 kB; setting this variable to false disables this check,
  so mail size restrictions below 64 kB can be used and are effective;

- added a by-contents-category setting %subject_tag_maps_by_ccat, unifying
  former separate settings @spam_subject_tag_maps, @spam_subject_tag2_maps,
  @spam_subject_tag3_maps and $undecipherable_subject_tag, and making it
  possible to specify subject tags (strings to be inserted into Subject:)
  for other categories, such as viruses, banned, and bad headers. Note that
  now only one such tag is inserted - previously if passed mail was both
  spam and undecodable two tags were inserted;

- when spam level is at or above tag_level, turn on contents category CC_CLEAN
  with a minor category 1, making it easier to configure actions and settings
  (like subject_tag) through %*_by_ccat variables;

- treat empty string in tag_level the same as undef, i.e. lets X-Spam-* header
  fields always be inserted for local recipients;

- new configuration variable: $allow_fixing_improper_header (also a member
  of policy banks) is a more-general big brother of an older configuration
  variable $allow_fixing_improper_header_folding.  It controls fixing of a
  mail header in passed and released mail; it currently controls truncating
  of header lines longer than 998 characters, and is a pre-condition for
  $allow_fixing_improper_header_folding, controlling removal of all-whitespace
  continuation lines.  The $allow_fixing_improper_header defaults to true
  for backwards compatibility. Fixing header may protect poorly written
  mail readers, but may break DomainKeys/DKIM validation of messages
  with illegal header if verification is done after content filtering,
  so if this is of concern, one has a choice of turning it off;

- added macro sprintf, which invokes Perl's sprints with the usual
  arguments semantics, e.g. [:sprintf|%%s %%.1f %%%%|text|[:SCORE]]
  based on suggestion (but swapped arguments) from Joel Nimety;

- added macros min and max, which returns minimal or maximal value from
  their arguments, ignoring empty string arguments, e.g. [:min|100|[:SCORE]]

- extend a set of saved header fields in a hash $msginfo->orig_header_fields,
  which is accessible through a macro 'header_field'; currently the set of
  kept header fields consists of: From, To, Cc, Sender, Subject, Received,
  Message-Id, Resent-Message-Id, Precedence, User-Agent, X-Mailer,
  DKIM-Signature, DomainKey-Signature, Authentication-Results;

- added macro useragent, which returns a 'User-Agent: ...' or 'X-Mailer: ...'
  header field (whichever is present); note that this is an entire field,
  including a header field head, unlike macros header_field and x-mailer;

- added macros dquote and uquote to facilitate sanitation of header fields
  in logging: dquote encloses its argument in double quotes and replaces
  existing double quotes by \" (suitable to sanitize Subject header field);
  uquote replaces one or more consecutive space or tab characters by '_',
  but does not protect existing underlines, which makes it a lossy
  transformation (suitable for From or To header fields);
  provisional - exact interpretation may change;

- updated notification templates to make use of new macros;

- insert X-Amavis-OS-Fingerprint header field (if available) into a passed
  message to local recipients (not just to a message copy submitted to
  SpamAssassin for checking); suggested by Jeff Noxon;

- insert X-Amavis-PenPals header field (if information is available) into a
  passed message, showing time interval (age) since the last message sent
  in the opposite direction, i.e. from the current recipient to the sender
  of the current message;

- insert X-Amavis-PolicyBank header field into a message that is passed to
  SpamAssassin for a check (but not to a passed message as seen by recipient).
  A header field body is a slash-separated list of all policy banks loaded,
  e.g.: X-Amavis-PolicyBank: AM.PDP-SOCK/MYNETS/MYUSERS
  (but more usually just: MYNETS).  If no policy banks are loaded,
  the header field will not be inserted. This information may be used by
  SA rules to add score points based on policy bank, or to countermeasure
  or conditionalize other rules, for example:

    header L_MYNETS X-Amavis-PolicyBank =~ m{(^|/)MYNETS(/|$)}m

    # apply score directly:
    score  L_MYNETS -0.8

    # use rule to countermeasure other rules:
    score L_MYNETS_UNDISC_RECIPS   -0.841
    score UNDISC_RECIPS             0.841

    # use rule to conditionalize other rules:
    score L_OTHERS_UNDISC_RECIPS   0.841
    score UNDISC_RECIPS            0.001

- a new hash variable %allowed_added_header_fields (also member of policy
  banks) is consulted for each header field insertion, and if the result is
  false the header field is not inserted into passed mail. Can be used to
  suppress inserting header fields such as X-Virus-Scanned, X-Spam-Report,
  X-Spam-Level, X-Amavis-PenPals, X-Amavis-OS-Fingerprint, X-Amavis-Modified,
  Received, etc. Only applies to passed mail, not to mail that is being
  written into a quarantine or to a copy submitted to SA for checking.
  Keys (header field names) must be in lowercase.

  Example use - disables insertion of certain header fields:

    $allowed_added_header_fields{lc('X-Amavis-OS-Fingerprint')} = 0;
    $allowed_added_header_fields{lc('X-Amavis-PenPals')} = 0;
    $allowed_added_header_fields{lc('X-Spam-Report')} = 0;
    $allowed_added_header_fields{lc('X-Spam-Status')} = 0;
    $allowed_added_header_fields{lc('X-Virus-Scanned')} = 0;
  Note that turning off 'X-Spam-Report' through %allowed_added_header_fields
  is equivalent to having $sa_spam_report_header at false, turning off
  'Received' is equivalent to having $insert_received_line at false, and the
  last line in the above example is equivalent to setting $X_HEADER_TAG to
  undef or $X_HEADER_LINE to undef. For compatibility the $X_HEADER_TAG is
  treated somewhat specially: if explicitly set to a nonstandard value, it is
  implicitly added to the %allowed_added_header_fields in the base policy
  bank. No automatism is provided for $X_HEADER_LINE in other policy banks.

  Example use in a policy bank:

    $policy_bank{'ALT'} = {
      allowed_added_header_fields => {
        lc('X-Amavis-PenPals')        => 0,  # turn it off
        lc('X-Amavis-OS-Fingerprint') => 0,  # turn it off

- when loading a policy bank, its entries of type hash are now merged with
  existing hash, a key at a time; previously a newly loaded hash replaced
  a previous one entirely; this is relevant for policy bank entries such
  as %allowed_added_header_fields and %*_by_ccat, making it more natural
  and specifying just keys that need to be changed in a new policy bank
  instead of entire hashes;

- kill external tnef decoder if running for too long;

- abort Convert::UUlib::LoadFile or other Convert::UUlib processing in
  do_ascii() if running for too long; problem case provided by Martin Grimm;

- add Net::Server hooks post_configure_hook() and post_bind_hook(), making
  it easier to affect protection of Unix sockets created by Net::Server
  (like allowing write access for group) by uncommenting a call to umask
  in post_configure_hook(); suggested by Mike Gaskins;

- internal: wrap most top-level initializations in BEGIN blocks, so that
  compiled initialization code will be discarded after it did its job,
  saving about 100 kB of process memory footprint;

- update and add several comments - either to clarify code, or to fix typos;

- document that a method 'smtp:' can be used for quarantining to a dedicated
  mailbox, much like the older way of using 'local:anything' with a fully
  qualified email address in quarantine_to, see:

    $notify_method = 'smtp:[]:10025';
    $quarantine_method_by_ccat{+CC_SPAM} = 'local:%m';
    $quarantine_to_maps_by_ccat{+CC_SPAM} = [''];

  now more obvious and preferred:
    $quarantine_method_by_ccat{+CC_SPAM} = 'smtp:[]:10025';
    $quarantine_to_maps_by_ccat{+CC_SPAM} = [''];

- make @viruses_that_fake_sender_maps a member of policy banks to make it
  possible to bounce certain types of viruses if originating from inside;

- added configuration variable @smtpd_discard_ehlo_keywords (also member of
  policy banks), which is a case-insensitive list of EHLO keywords (AUTH, DSN,
  8BITMIME, PIPELINING, SIZE, etc.) that the SMTP/LMTP server will not send
  in an EHLO response to a remote SMTP client. It is equivalent to a Postfix
  configuration variable of the same name.

  Practical use example - DKIM/DomainKeys signing of locally-originating
  mail after it has passed a content filter:

  # Configure MTA to send to port 10026 mail originating from our users
  # (from mydomains or authenticated roaming users, it will be returned
  # to port 10027 after checking), and to send to port 10024 all the rest
  # (incoming mail), which will be returned to port 10025 after checking:

  $forward_method = 'smtp:[]:10025';  # MTA with non-signing service
  $notify_method  = 'smtp:[]:10027';  # MTA with DKIM signing service

  $inet_socket_port = [10024,10026];  # listen on two ports
  $interface_policy{'10026'} = 'ORIGINATING';  # switch policy bank on 10026

  $policy_bank{'ORIGINATING'} = {  # mail originating from our users

    # force MTA to convert mail to 7-bit before DKIM signing
    # to avoid later conversions which could destroy signature:
    smtpd_discard_ehlo_keywords => ['8BITMIME'],

    # forward to a smtpd service providing DKIM/DomainKeys signing service:
    forward_method => 'smtp:[]:10027',

    # other special treatment of locally originating mail, e.g.:
    spam_admin_maps  => ["virusalert\@$mydomain"],  # warn of spam from us
    banned_filename_maps => ['ALT-RULES'],  # more relaxed rules...
  # mail return from a content filter (non-signing)
  10025 inet n - n - - smtpd
      -o content_filter=
  # mail from our users returning from a filter (DKIM or DK signing service)
  10027 inet n - n - - smtpd
      -o content_filter=
      -o milter_macro_daemon_name=ORIGINATING
      -o smtpd_milters=inet:

  Note that the same effect (making Postfix convert outgoing mail to
  7-bits before DKIM signing) could be achieved by a Postfix setting
  smtp_discard_ehlo_keywords=8bitmime on a smtp service feeding mail
  to be signed to amavisd, but this would require setting up two such
  services, one with the option and one without.

- README.postfix: suggest options "-o local_header_rewrite_clients="
  and "-o smtpd_milters=" on smtpd at port 10025; thanks to Noel Jones;

- README.sendmail-dual: add FEATURE(`nocanonify') on MTA-TX, and
  FEATURE(`nocanonify',`canonify_hosts') on MTA-RX; thanks to Ricardo Stella;

                                                              June 27, 2006
amavisd-new-2.4.2 release notes


- new feature: "pen pals soft-whitelisting" lowers spam score of received
  replies to a message previously sent by a local user to this address;
- new feature: added command line options to override certain configuration
  settings from a config file, see below;
- documentation bug fixes, especially on the use of SQL data type TIMESTAMP;
- zoo decoder interface routine can now use utility unzoo(1) or zoo(1);


There are no incompatible changes since 2.4.1, but please notice below
the fixes to SQL and to LDAP documentation, which may affect you.


- LDAP.schema: add missing LDAP attribute amavisSpamQuarantineCutoffLevel
  to the list of allowed attributes in objectclass amavisAccount;
  pointed out by Paolo Cravero;

- README.sql PostgreSQL notes: fixed incorrect advice in README.sql which
  suggested to declare a field msgs.time_iso as TIMESTAMP WITHOUT TIME ZONE
  instead of the correct TIMESTAMP WITH TIME ZONE.
  Previous instructions were also contradictory to suggested data type on
  ALTER TABLE msgs ALTER time_iso. Using inappropriate WITHOUT TIME ZONE
  when comparing time_iso to now() (which is aware of a time zone) offsets
  results by a current time zone offset, which unexpectedly includes too
  many or too few records in maintenance purging operation. A workaround
  if WITHOUT TIME ZONE continues to be used is to:  SET TIME ZONE 'UTC';
  before purging.

  If you were misled by previous documentation choosing WITHOUT TIME ZONE
  for time_iso, and decided now to convert it to WITH TIME ZONE, the following
  clause can convert time_iso to proper universal time by manually providing
  appropriate time offset:

    ALTER TABLE msgs ALTER COLUMN time_iso
      USING time_iso [-+] INTERVAL '[offset]';

  Thanks to Brian Wong for a problem description and advice.

- README.sql MySQL notes: fixed incorrect advice in README.sql which suggested
  to declare a field msgs.time_iso as TIMESTAMP instead of the correct
  TIMESTAMP NOT NULL DEFAULT 0.  The "DEFAULT 0" is mandatory to prevent
  MySQL from overwriting mail reception timestamp with current local time
  when other fields are updated at the end of processing of a message.
  Also not to be forgotten: $timestamp_fmt_mysql *MUST* be set to 1 in
  amavisd.conf with MySQL when msgs.time_iso data type is TIMESTAMP... !

- README.sql MySQL notes: if using field msgs.time_iso to select records
  for purging (instead of msgs.time_num), and its data type is TIMESTAMP...
  (as opposed to CHAR...), one should use function utc_timestamp() in place
  of now() in the DELETE clause to make it work correctly regardless of
  time zone. Alternatively, now() can continue to be used, provided that
  SQL client time zone is set to UTC in the purging SQL script:
  SET time_zone='+00:00';  thanks to Gary V for investigation;

- README_FILES/README.sql: added short for-the-impatient sections:
  * BRIEF MySQL EXAMPLE of a log/report/quarantine database housekeeping
  * BRIEF MySQL EQUIVALENT EXAMPLE based on time_iso if its type is TIMESTAMPS
  * BRIEF PostgreSQL EXAMPLE of a log/report/quarantine database housekeeping

- a message with only a header, without empty separator line and with no body,
  lost the last line of a header on forwarding or writing to quarantine;
  observed by Elias Oltmanns, reported through Debian bug tracking;

- header validity checks inappropriately reported 'header field syntax error'
  as a 'header field too long' (BadHdrLong) instead of BadHdrSyntax;

- ensure that notification would not be sent if notification template is empty,
  solving the following problem: when recipient notifications for bad headers
  is enabled, and a message is spam with bad headers, recipient would receive
  an empty notification message (because message contents category is spam
  and recipient notification template for spam is empty);  reported by Alex;

- changed SMTP status code 550 to 554 when rejecting mail contents,
  the 550 is not envisioned in RFC 2821 as a valid reply code to a "."
  after data transfer; thanks to Victor Duchovni;

- fixed case mismatch when storing e-mail address to SQL table maddr,
  which wasted one unnecessary failed attempt on INSERT;

- ignore $timestamp_fmt_mysql if SQL database driver (DBD) is not 'mysql';

- perl taint workaround in lookup_sql() where SQL select clause
  could become tainted; problem reported by Christer Borang;

- fixed amavisd.conf-default which stated incorrect default values of
  keys 'ins_rcp' and 'ins_quar' in %sql_clause; reported by Glenn Sieb;

- limit reported boost score to three decimal places; long fractions
  observed by Gary V;

- Postfix since version 20060610 uses xtext-encoded (rfc3461) strings in
  XCLIENT and XFORWARD attribute values, previous versions used plain text
  with neutered special characters - amavisd-new now xtext-decodes value
  if it looks xtext encoded, and encodes it on sending; the change could
  affect exotic host names (e.g. with a plus in host name) from broken
  mailers or DNS; thanks to Ralf Hildebrandt for pointing out the recent
  change in Postfix;

- improve regular expressions in the $map_full_type_to_short_type_re list
  to cope better with different versions of the file(1) utility regarding
  recognition of various MS executables; based on a problem report by Misha;

- use stricter suggested regular expression in amavisd.conf for matching
  CLSID (Class ID extension); previous expression was loose and too easily
  matched file names with braces in the name; suggested by Martin Schuster
  through Debian bug tracking;

- zoo decoder interface routine (do_zoo) can now use utility unzoo(1) or the
  traditional zoo(1);  the unzoo(1) recognizes some additional parameters
  which makes it more resilient (but still not watertight) against some
  attempts to hide archive contents or to extract members to unexpected
  locations, but unfortunately does not recognize all zoo compression schemes
  ("error, LZD not yet implemented"), and the relative modes "-j ./" or "-j X"
  do not protect against all malicious cases - so it is a mixed blessing.
  The way amavisd calls zoo(1) (piping members to stdout, which can be slow)
  avoids some of the security problems with zoo (writing to arbitrary
  directories), which were probably the main reason for ClamAV project
  deciding to switch to unzoo(1);

- zoo sucks, unzoo (v4.4) sucks more: considered, but decided against changing
  zoo entry in @decoders to ['unzoo','zoo'] in amavisd.conf, as was suggested
  by Gábor Kövesdán. It would not necessarily be an improvement (see previous
  item, misses extracting members from my test cases), so feel free to choose
  between the two poor choices, I still prefer zoo(1), partly also because it
  covers cases which clamd decoding misses;

- kill external zoo or unzoo decoder if running for too long;

- internal: saving recipient addresses to SQL table maddr is now done
  earlier to make information available to pen pals code;

- explicitly test if SQL 'prepare' silently fails to return a statement
  handle, just in case;

- adjusted list of pre-loaded SA modules to cater for SA 3.1.3;


- new feature: added command line options which override some configuration
  settings from a config file (an option to override pid_file suggested by
  Paul Murphy and Gábor Kövesdán):

    -d log_level        ... overrides $log_level
    -m max_servers      ... overrides $max_servers
    -L lock_file        ... overrides $lock_file (Net::Server serialization)
    -P pid_file         ... overrides $pid_file
    -H home_dir         ... overrides $MYHOME directory
    -Q quarantine_dir   ... overrides $QUARANTINEDIR directory, empty disables
    -T tempbase_dir     ... overrides $TEMPBASE directory
    -S helpers_home_dir ... overrides $helpers_home directory (SA workplace)
    -D db_home_dir      ... overrides $db_home, empty arg turns off $enable_db
    -R chroot_dir       ... overrides $daemon_chroot_dir, empty avoids chroot
    -p listen_port_or_socket  ... overrides $inet_socket_port as well as
                        $unix_socketname, argument may be a decimal TCP port
                        number, or an absolute path name of a Unix socket;
                        may be specified multiple times: daemon can listen on
                        multiple inet sockets and/or multiple Unix sockets;
                        example: -p 10024 -p 9998 -p /var/amavis/amavisd.sock
    -V  ... shows version and exits
    -h  ... shows version and command line options, then exits

  For completeness, here are remaining options, unchanged from
  previous versions:
     -u user            ... overrides $daemon_user
     -g group           ... overrides $daemon_group
     -c config_file     ... config file name, may be specified multiple times

- new feature: "pen pals soft-whitelisting" lowers spam score of received
  replies (or followup correspondence) to a message previously sent by a
  local user to this address;

  * both the outgoing and the incoming mail must pass through amavisd
    (although outgoing mail may have checks disabled or made more permissive
    if desired);
  * SQL logging must be enabled (@storage_sql_dsn) and records should
    be kept for at least several days (some statistics (2006-11 update):
    90% of replied mail (or followups) is sent within 2 weeks since
    previous correspondence, 40% within 24 hours, 20% within 3 hours,
    10% within 30 minutes, 5% within 12 minutes);
  * @mynetworks and @local_domains_maps must reflect reality, allowing amavisd
    to distinguish between outgoing, incoming and internal-to-internal mail;
  * the information about client IP address must be available to amavisd,
    i.e. Postfix XFORWARD protocol extension must be enabled, or AM.PDP+milter;
  * configuration variable $penpals_bonus_score must be set to a positive
    value (such as 1.0, increase to perhaps 5 or 8 after seeing that it works),
    zero disables the feature and is a default;
  * $sql_clause{'sel_penpals'} must contain a SELECT clause (which by
    default it does, unless overridden by an old assignment to %sql_clause
    in amavisd.conf);
  * sender/recipient address pair must exactly match recipient/sender pair of
    previous correspondence (except for allowed case-changes in domain part),
    which means that care must be taken when canonical and/or virtual mapping
    is performed by MTA (such as mapping between internal and external address
    forms) - if external address forms of local addresses are to be seen by
    a content filter then canonical mapping (int->ext) must be done *before*
    filtering and virtual mapping (ext->int) *after*;  alternatively, if
    internal address forms are to be seen by a content filter, then canonical
    mapping should be done after filtering, and virtual mapping before;
    (P.S. later renamed to 'Advanced Postfix and amavisd-new configuration');

  How it works:
  * SQL logging stores records about all mail messages processed by amavisd,
    their sender, recipients, delivery status, mail contents type (no changes
    there, this feature was introduced with amavisd-new-2.3.0); for the
    purpose of pen pals scheme only records with local-domain senders matter;
  * when a message is received, a SQL lookup against a SQL logging database
    is performed, looking for previous messages sent in reverse direction,
    i.e. from a local user (which is now a recipient of the current mail)
    to the address that is now the sender of the message being processed;
    A SELECT clause in $sql_clause{'sel_penpals'} is used, which by default
    only considers records of previous messages that were actually
    delivered (not rejected, discarded or bounced), and were not infected.
    SQL lookup returns a timestamp of the most recent such message (if any),
    the difference (in seconds) between the current time and the timestamp
    is an 'age' as used in the following formula;
  * an exponential decay formula calculates score points to be deducted
    from the SA score:
      weight = 1 / 2^(age/penpals_halflife)
      score_boost = -penpals_bonus_score * weight
    i.e. penpals_bonus_score is multiplied by 1, 1/2, 1/4, 1/8, 1/16, ...
    at age 0, 1*halflife, 2*halflife, 3*halflife, 4*halflife ...
    weight is a continuous function of age (actually, in steps of one second);
  * main configuration variables, members of policy banks:
      $penpals_bonus_score  ... a maximal (positive) score value by which
        spam score is lowered when sender is known to have previously received
        mail from our local user from this mail system. Zero or undef disables
        pen pals lookups, and is a default.
      $penpals_halflife  ... exponential decay time constant in seconds,
        defaults to 7 days; pen pal bonus is halved for each halflife
        period since the last mail sent by a local user to the current
        message's sender;
  * auxiliary configuration variables, global settings:
      $penpals_threshold_low ... SA score below which pen pals lookups are
        not performed to save time, defaults to 1.0;  undef lets the threshold
        be ignored (useful for testing and statistics gathering);
      $penpals_threshold_high ...
        when (SA_score - $penpals_bonus_score > $penpals_threshold_high)
        pen pals lookup will not be performed to save time, as it could not
        influence blocking of spam even at maximal penpals bonus (age=0);
        usual choice for value would be kill level or tag2 level, or other
        reasonably high value; undef lets the threshold be ignored and is
        a default (useful for testing and statistics gathering);

  Caveats / notes / exceptions with "pen pals soft-whitelisting":
  * pen pals soft-whitelisting aids incoming mail, and internal-to-internal
    mail, but has no effect on outgoing mail;
  * if SQL logging was not used so far and you are considering enabling it
    for a busy site, you would appreciate PostgreSQL 8.1 compared to MySQL,
    as purging old records seems to be *much* faster than in MySQL 4.1,
    which could lock down mail processing for an hour or more during a
    weekly (or daily) purge, as opposed to minutes or seconds;
  * infected messages are exempted from pen pals checks;
  * mail with (unadjusted) SA score below $penpals_threshold_low (1 by default)
    is exempted from pen pals check to save time and lighten the load on SQL;
    similarly for high score spam which would not have a chance of being
    'saved' even by a maximal pen pals bonus score;
  * non-delivery notifications have null sender address, so can not match
    previous correspondence and can not receive a pen pal bonus;
  * unauthenticated sender address matching local domains but coming
    from outside is not trusted and is exempted from pen pals checks;
  * messages from a local user to self are exempted from pen pals check;
  * outgoing messages (i.e. to non-local recipients) are exempted from
    pen pals checks to save some time and simplify reasoning (which reverse
    mail transaction to trust?); assuming that local users rarely send
    spammy mail, outgoing mail would rarely need help from pen pals checks;
  * messages received from mailing list typically use ML bounce or admin
    address (possibly VERPed) as the sending address, so they would not be
    considered replies to postings to a mailing list from a local user
    (to be addressed in future version 2.5.0);
  * underlying assumption is that a local-domains sender address in mail
    coming from inside can be trusted not to be faked; if this is not the case,
    an internal user cooperating with a spammer can widen spam tolerance for
    another internal user (but it probably does not pay off, too much trouble
    for too little effect);
  * if a spammer knows or can guess that a local user is frequently sending
    mail to some address (e.g. a mailing list unprotected by DKIM or SPF), he
    can gain few bonus score points by using such sending address in his spam;
  * there may be multiple MTA+amavisd servers, but all must use the same
    logging SQL database;
  * forwarding is compatible with the pen pals scheme;
  * broken forwarding scheme like SRS (with SPF), where envelope sender
    address is replaced by a forwarding mailbox address is counterproductive;
    for example: a local user may also have an external mailbox at some remote
    provider with poor spam protection; forwarding from the remote to a local
    mailbox is set up and a forwarding MTA misguidedly substitutes original
    sender address with a mailbox address; spam reaching remote mailbox
    is forwarded to a local site with a sender address rewritten, making
    it look like it is coming directly from a user's remote mailbox, and
    inappropriately benefiting from pen pals bonus of user's previous
    correspondence with his remote mailbox;

  * set $penpals_bonus_score initially to a low value such as 1
    to avoid surprises;
  * set $penpals_threshold_low and $penpals_threshold_high to undef
    to perform pen pals lookups regardless of the score;
  * at log level 2 (or higher) search the log for a string "penpals: "
    (only shows on incoming mail sent by a non-local sender); the log also
    shows mail_id of the referenced message (previous communication),
    and Subject header fields of previous and current message;

  Based on a feature request by Aaron P. Martinez, thanks to Gary V for
  suggestions and prompting and to Michael Scheidell and Richard Bishop
  for feedback.

                                                                May 8, 2006
amavisd-new-2.4.1 release notes


- notification templates incompatibility with 2.4.0 (but not with versions
  2.3.3 or older): major contents category numbers are renumbered due to a
  newly inserted category CC_SPAMMY; it affects the use of macro ccat_maj
  in templates (one field added), and only affect users which provide
  non-default templates based on 2.4.0 templates; older templates (2.3.3
  or earlier) are unaffected as they do not use macro ccat_maj;


- revert a change introduced with 2.4.0, which was adding address extensions
  at CC_SPAM, i.e. when score exceeds kill level. Previously (2.3.3) address
  extensions were inserted at tag2 level. Implemented by a new mechanism:
  a new major contents category CC_SPAMMY is inserted just below the CC_SPAM,
  where CC_SPAMMY is controlled by tag2_level and CC_SPAM continues to be
  selected at kill_level. Also spam defanging (if enabled) is now activated
  at CC_SPAMMY and no longer at CC_SPAM (which was on a TODO list for some
  time); undesired change in 2.4.0 reported and changes tested by Mario Liehr;

- fixed old nuisance bug (probably present since 2.3.0) when an external
  decoder program for self-extracting archives (rar/unrar, lha, arj/unarj)
  is defined but the program does not exist, which resulted in logged
  non-fatal errors like:
    run_command: failed to exec SCALAR(0x8598550) lq ...
    run_command: failed to exec REF(0x85985c8) v -c- -p- -av- -idcdp -- ...
    run_command: failed to exec ARRAY(0x89e5f0c) l ...
      No such file or directory
  reported by Martin Baertl, Maurizio Marini, boka, and Donald Teed,
  investigated by Gary V;

- bug fix in a Courier setup: add a missing reset of per-recipient data
  to prevent previous message check affecting the next one performed by
  the same process; fix by Martin Orr, reported by Bowie Bailey;

- the amavisd-new-courier.patch now requires Net::Server version 0.90 or
  later (preferably 0.93 or later); to use older version of Net::Server
  please apply the older amavisd-new-courier-old.patch and follow
  README.courier-old - both will go away with next version of amavisd-new;

- updated amavisd-new-qmqpqq.patch patch (qmail interface) to work with
  the current code, by Martin Solciansky, testing by Nicklas Bondesson;

- fix error handling when a problem occurs during temporary directory cleanup;

- when defanging mail make a 'Subject' header field be editable by header
  edits, so that Subject tags like ***UNCHECKED*** can still apply;

- modify unquote_rfc2821_local so that it appends an '@' as a domain name
  only if localpart contains '@', so that read_array() can still be
  used to read a list of networks in CIDR notation; a change in 2.4.0
  to properly handle addresses like "aaa@bbb" (local part in quotes,
  no domain) made read_array unsuitable for reading list of networks;
  pointed out by Petr Vokac;

- add another round of local($1) declarations as a workaround for already
  familiar Perl taint bugs, popping up again on some Perl installations;
  reported by Jaap Struyk; reported symptoms were:
    Insecure dependency in chown while running with -T switch at
      /usr/lib/perl5/site_perl/5.8.7/Net/ line 488
    Insecure dependency in eval while running with -T switch at
      /usr/lib/perl5/site_perl/5.8.7/Mail/SpamAssassin/ line 91

- added config variables: @spam_subject_tag3_maps, @spam_tag3_level_maps
  (and $sa_tag3_level_deflt), which makes it possible to split contents
  category CC_SPAMMY into two sublevels (minor categories) and give
  each its own Subject tag text;
  the "CC_SPAMMY,0" contents category still corresponds to tag2 level,
  and "CC_SPAMMY,1" contents category corresponds to tag3 level (if defined).
  Only static maps are available (also members of policy banks), but
  no corresponding SQL and LDAP attributes are provided.  Example:
    @spam_tag2_level_maps = (5.5);
    @spam_tag3_level_maps = (12);
    @spam_subject_tag2_maps = ('***LIKELY*SPAM*** ');
    @spam_subject_tag3_maps = ('***BLATANT*SPAM*** ');
  based on suggestion from Benedict White;

- add LDAP attributes: amavisSpamSubjectTag, amavisSpamSubjectTag2,
  amavisSpamDsnCutoffLevel, amavisSpamQuarantineCutoffLevel to match
  equivalent SQL lookup fields; missing amavisSpamQuarantineCutoffLevel
  noticed by Paolo Cravero;

- presence of LDAP attributes is now tested with 'defined', no longer as
  Perl booleans;

- mail_via_bsmtp: storing mail in BSMTP format now saves DSN information,
  as permitted by RFC 2442;

- apply the concept of separate timers $child_timeout and $smtpd_timeout
  as used in a SMTP session to AM.PDP and AM.CL protocols;

- apply the concept of separate timers $child_timeout and $smtpd_timeout
  as used in a SMTP session to Courier patch; by Martin Orr;

- new macros: remote_mta, smtp_response, remote_mta_smtp_response
  and score_boost available to log templates and notification templates;

- enhanced regexp selector macro [~string|regexp|then|else], which can now
  capture parenthesized regexp subexpressions and make them available
  as %1, %2, ... %9 to 'then' and 'else' replacements; a copy of the first
  argument (a string) is available to replacements as %0;

- extend the semantics of the regexp selector macro, which can now take
  more than one pair of regexp+then arguments, catering for a nested
  'if then elseif then elseif then else' structure:

- enhanced iterator macro, which can now take a long macro name as its
  first argument, and imply a %x as iterator name;

- make use of the new macro remote_mta_smtp_response and add it to
  a default $log_templ, so that a Postfix queue-id of a forwarded
  message shows up like 'queued_as: DCF2A17B9E4' in the main log entry,
  facilitating search for a related log entry in a MTA log.
  In case of a mail split, all the MTA responses would now be
  shown, e.g.:  queued_as: F3DBD17B847/F3DBD17B847/F3DBD17B847
  (customizable by the use of macros in $log_templ);

- sophos_savi_internal (SAVI module): don't include errno ($!) in the
  error message, it may be misleading; reported by Matthias Ivers;

- internal - programming style: use more predictable $b=1 instead of $b++
  where variable $b is supposed to be a boolean and not a counter;

                                                              April 3, 2006
amavisd-new-2.4.0 release notes

The most important changes since 2.3.3 at a glance:

  Delivery status notifications (DSN) are now supported, both as a SMTP
  protocol extension and in notifications. Header fields like X-Amavis
  and X-Spam are now prepended to mail header for DomainKeys compatibility.
  Configuration variables can be chosen based on mail contents category,
  which is now represented explicitly. A built-in macro expander is enhanced,
  providing new macros and call types. Added support for passive operating
  system fingerprinting with the use of p0f, supplying collected information
  as a header field to SpamAssassin. Provide compatibility with Net::Server
  0.91 and later.


- incompatible change when logging or quarantining to SQL: added field
  'quar_loc' to table 'msgs' to facilitate quarantine release, and added
  FOREIGN KEY constraint for data consistency and simplified purging;
  see below for a simple database modification;

- inserted header fields like X-Amavis-* and X-Spam-* are now _prepended_ to
  mail header instead of being appended, and occupy position just above the
  inserted Received header field; this pairing with Received makes it easier
  to identify which MTA/content filter inserted them, makes it consistent
  with position of Resent-* header fields as required by RFC 2822, and avoids
  the possibility of breaking DomainKeys, DKIM, and similar mail signing
  schemes. SpamAssassin implemented the same change with 3.1.0.
  To achieve former behaviour, specify: $append_header_fields_to_bottom=1;

- trailing whitespace is no longer trimmed by default from SQL fields,
  from LDAP attribute values and from associative array righthand-sides
  (hash values) as read by read_hash(); see below if trimming is really
  still needed;

- SMTP server side: no longer allow e-mail address without enclosing
  angle brackets in MAIL FROM and RCPT TO smtp commands; such syntax is
  illegal according to RFC 2821 and RFC 821, no compliant MTA is using it,
  so the change should not effect anyone (except perhaps sloppy testers);

- changed defaults for banned & bad header administrator address to:
    $banned_admin = undef;
    $bad_header_admin = undef;
    @banned_admin_maps = (\$banned_admin, \%virus_admin,\$virus_admin);
    @bad_header_admin_maps = (\$bad_header_admin);
  In other words, if $banned_admin is left at a default value (undefined),
  banned admin falls back to %virus_admin or $virus_admin.
  If $bad_header_admin is left at a default value (undefined), bad header
  admin has no default, admin notifications for bad headers are not sent;


- support for DSN (RFC 3461) in the SMTP protocol (parameters NOTIFY and ORCPT
  in ESMTP RCPT commands, parameters RET and ENVID in ESMTP MAIL command), with
  corresponding updates to Delivery Status Notifications (RFC 3462, RFC 3464);
  (about a missing support for option ORCPT in Net::SMTP please see );

- represent mail contents category more explicitly internally, and provide
  new configuration variables:
    %final_destiny_by_ccat %lovers_maps_by_ccat %defang_by_ccat
    %quarantine_method_by_ccat   %quarantine_to_maps_by_ccat
    %notify_admin_templ_by_ccat  %notify_recips_templ_by_ccat
    %notify_sender_templ_by_ccat %warnsender_by_ccat
    %hdrfrom_notify_admin_by_ccat %mailfrom_notify_admin_by_ccat
    %hdrfrom_notify_recip_by_ccat %mailfrom_notify_recip_by_ccat
    %admin_maps_by_ccat %dsn_bcc_by_ccat
    %warnrecip_maps_by_ccat %addr_extension_maps_by_ccat
  gradually phasing out separate configuration variables for each category;
  the change is fully backwards compatible, existing variables are referenced
  through default values of the new variables, and no longer used directly;

  The chain of lookups adhere to the following evaluation sequence
  for settings with an associated *_by_ccat mechanism:
    * policy bank chooses a *_by_ccat associative array (by TCP port or
      by client's IP address (MYNETS));
    * the most relevant contents type of the message chooses an entry in
      a _by_ccat associative array; the entry can be a final settings value,
      or a ref to an array of by-recipient lookup tables (*_maps);
      mostly for compatibility reasons an entry can also be a ref to CODE,
      which allows for delayed evaluation through legacy *_maps settings
      (which may again be members of policy banks);
    * the chosen list of lookup tables is queried based on recipient address,
      producing a final setting;

  Note that currently only settings which are applicable _after_ the mail
  contents type has already been determined, have their associated _by_ccat
  associative array. Settings like @bypass_spam_checks_maps which need to be
  evaluated _before_ mail contents is assessed, do not have their associated
  _by_ccat variable;

- added ability to explicitly kill externally running decoder process or
  a command-line virus scanner process if running for too long;

- enhanced built-in macro expander now allows long macro names (previously
  limited to one character), neutral and active macro calls, dynamically
  defining macros, new regexp matching built-in macro, more robust
  and explicit bookkeeping of quoting levels, as well as speedups achieved
  by pre-tokenization; details in see README.customize;

- improved wrapping of inserted header fields, fields in DSN, and in
  generated text sections of the new notification templates;

- improved text of notification templates, taking advantage of new macros;

- compatible with Net::Server 0.90, 0.91, 0.92 and 0.93 by providing
  workarounds; thanks to Paul Seamons, the author of Net::Server, for his
  cooperation, the 0.93 solves problems introduced by a change in 0.91
  (but it remains incompatible with version of amavisd-new 2.3.3 and older);

- experimental support for passive operating system fingerprinting with
  the use of externally running utility p0f, supplying collected information
  as a header field to SpamAssassin, making possible to add rules to score
  SMTP client hosts based on educated guess about their operating system
  type and IP distance; see below for details;

- make variable $myhostname a dynamic variable, member of policy banks,
  likewise for syslog parameters facility, priority and ident; details below;

- added config options to enable quarantining (archiving) of clean mail;

- lots of cleanups and generalizations in the code;


- fix insufficient sender address sanitation when storing quarantined or
  forwarded files as BSMTP files _and_ having a %s in the corresponding
  *_method template; potential security vulnerability (with limited scope)
  in versions of amavisd-new 2.3.1, 2.3.2 and 2.3.3 discovered by Thomas

- recognize result "ms-windows metafile" (or "ms-windows metafont") from a
  file(1) utility and provide short type 'wmf' for it; added two example
  rules to amavisd.conf (and amavisd.conf-sample) to block files containing
  Windows Metafiles, based on US-CERT Alert TA05-362A;


- incompatible change when logging or quarantining to SQL is enabled
  (as mentioned above, here is a more detailed description of the change):
  * added column 'quar_loc' to table 'msgs' to store quarantine file name
    (the same string as in macro %q, normally seen in the main log entry);
    based on input from Andrew A. Neuschwander, Brian Wong and Craig Herring;
  * add constraint FOREIGN KEY ... ON DELETE CASCADE to keep database
    consistent (free of orphaned records) and simplify maintenance deletions
    and possibly speed them up; suggested by Brian Wong;

  The following clause must be executed for upgrading pre-2.4.0
  amavisd-new SQL schema to the 2.4.0 schema:

    ALTER TABLE msgs ADD quar_loc varchar(255) DEFAULT '';

  The following clause should preferably be executed to take advantage

    ALTER TABLE msgrcpt
    ALTER TABLE quarantine

  See updated suggested set of DELETE clauses at the end of README.sql.

  The following clause can optionally be used to create an index
  on field msgs.time_num to speed up deletions in MySQL;
    CREATE INDEX msgs_idx_time_num ON msgs (time_num);
  or if purging is based on field msgs.time_iso instead of msgs.time_num:
    CREATE INDEX msgs_idx_time_iso ON msgs (time_iso);

  (compatibility note with pre-releases of 2.4.0: there were added fields
  msgrcpt.time_num and quarantine.time_num in pre-release versions of 2.4.0,
  which are now dropped in favor of FOREIGN KEY constraint; these fields
  are no longer set by the program and should not be relied-on when purging
  records, they may be removed from tables);

- solve compatibility issues brought up by changes in file descriptors
  usage as introduced with Net::Server version 0.91; thanks to Ralph Seichter,
  Matt Jackson, Jim Knuth and Paul Seamons (the author of Net::Server)
  for help;

- bug fix with LDAP lookups: if a LDAP connection to the server drops
  (i.e. after being idle for some time) amavisd is unable to reconnect;
  a symptom in the log is: 'do_search: failed again'; a fix by Petr Vokac,
  and later independently fixed by Matteo Brancaleoni and Mike Hall;
  problem also reported by Paolo Cravero;

- bug fix with LDAP amavisBannedRuleNames lookups, failing to looking up a
  set of banned rules names and referencing them in the users lookup table,
  like in other lookup tables. The LDAP lookup was returning an array reference
  for the 'amavisBannedRuleNames' attribute since it was a list (multivalued)
  and the reference wasn't being dereferenced down the line. The fix is to
  make the attribute single valued, the value can be a comma-separated list
  of names. This brings it in-line with SQL lookups which also uses a single
  field of comma-separated names. The included LDAP.schema and documentation
  files are fixed accordingly. Also moved the LDAP stuff out of README.lookups
  into its own README.ldap and updated it accordingly for the banned
  rules stuff. Fixed by Michael Hall; problem reported by Jérôme Schell,
  Aury Fink Filho and Brian Wong; thanks also to Jack Stewart and Willi Gruber;

- bug fix: properly disconnect SMTP session with 421 response if it times out;
  watchdog timer needs to be nudged during DATA transfer as well; pointed out
  by Victor Duchovni;

- introduced new configuration variable $smtpd_timeout (default 8*60 seconds)
  which controls the amount of time we are willing to wait for slow/idle client
  during incoming SMTP session before disconnecting a session. Previously the
  $child_timeout was covering complete elapsed time, both our processing and
  waiting for client, now the $child_timeout only still limits our processing,
  and $smtpd_timeout only limits waiting time. With Postfix after-queue setup
  the $smtpd_timeout should be higher than Postfix setting max_idle (default
  100s). Some other setups (like a pre-queue setup) may demand substantially
  higher $smtpd_timeout values; inflexibility pointed out by Martin Schmitt;

- added ability to kill externally running decoder process or a command-line
  virus scanner process if running for too long; currently implemented for
  all command-line virus scanners and for more common and/or more troublesome
  external decoders: do_unrar, do_unarj, do_uncompress, do_pax_cpio, do_lha,
  and partly for do_arc, do_zoo;  allowed time is calculated as 2/3 of the
  remaining time (initially at $child_timeout), but at least 10 seconds;

- use the same timeout calculation as above for calls to SA, taking
  $sa_timeout instead if that value is bigger than the calculated time,
  thus making $sa_timeout pretty much redundant;

- let do_pax_cpio recognize (and ignore) a single character in place of
  a date in more exotic cases of a pax listing; reported by Ralf Hildebrandt;

- standards compliance: recognize (and discard) source route in mail address
  as required by rfc2821;

- no longer bother to convert addresses like <""> to <>,
  both forms are invalid anyway, and recent versions of Postfix treat them
  the same. It is probably a good idea to set strict_rfc821_envelopes=yes
  in to reject such non-replyable sender addresses straight away,
  otherwise we end up processing such mail with inability to bounce it when
  needed, effectively losing it;

- make address with '@' in the localpart but without a domain (such as
  <"aaa@bbb"> ) distinguishable from <aaa@bbb> by appending an empty domain
  ('@' only) to the internal (unquoted) address form;  also, we used to strip
  off empty domain on rfc2821-quoting, but this leads Postfix to interpret
  an address with an '@' in the local part like <""> as
  <> (subject to the 'resolve_dequoted_address' Postfix
  setting), which is not what the sender requested (perhaps unintentionally)
  so we no longer do that. Both measures together, along with the new address
  parsing code, solve the inconsistency problem reported by Les Ault;

- fix string_to_mime_entity() to properly split header from body even in some
  corner cases (empty header or empty body); as a bonus a tiny speedup in
  template message splitting is gained;

- mail header of a 'defanged' message should not contain broken original
  headers (with illegal characters or whitespace lines); now sanitize such
  header fields; reported by Ivers Matthias;

- do not fix illegal all-whitespace continuation header lines when writing
  to quarantine (or when submitting notifications) in order to preserve the
  original bad header; only fix the header when such message is forwarded or
  released from a quarantine; masking problem brought up by Michael Scheidell;

- when quarantining in a Unix-style mbox format, replace null return path
  in a delimiting 'From ' line with a string 'MAILER-DAEMON', like Postfix
  and sendmail local delivery agents do, otherwise some mbox-reading
  clients do not recognize the line as a message delimiter;

- when quarantining to a mbox file, ">"-escape all /^From / lines, not just
  the ones following a blank line; this is more universal and does not break
  on more sloppy mail readers (thunderbird, kmail, mutt and pine);  MUAs like
  elm and mail(1) (the later usually comes with the OS) are more robust,
  treating as a message delimiter only /^From / lines following a blank line,
  these did not mind the more compact approach used by amavisd-new so far;

- new config variable $syslog_ident makes it possible to configure syslog
  ident string, its value defaults to 'amavis'; suggested by Andrzej Kukula;

- instead of the old config variable $SYSLOG_LEVEL (default value 'mail.debug')
  there are now two config variables $syslog_facility and $syslog_priority,
  defaulting for compatibility to the before-the-dot and after-the-dot
  substrings of the variable $SYSLOG_LEVEL.  The variable $SYSLOG_LEVEL
  still exists, can still be used, but is considered obsolete;

- make variables $syslog_ident, $syslog_facility and $syslog_priority
  dynamic variables, members of policy banks. This makes it possible for each
  policy bank to use its own specific syslog settings, for example to log
  to a different file (by using a different syslog facility like 'LOCAL3'
  for mail originating from inside (policy 'MYNETS')), or to change syslog
  ident to 'outgoing-amavis' for certain policy bank, or to rise syslog
  priority for releases from a quarantine. Note that switching syslog_ident
  or syslog_facility is done only when necessary, and involves closing and
  reopening syslog connection, which involves some (quite small) cost for
  each change. Dynamically changing syslog_priority however is for free;

- make variable $myhostname a dynamic variable, member of policy banks.
  This makes it possible for each policy bank to exhibit a different identity
  in notifications, in inserted header fields and in certain log entries.
  A quick overview of where myhostname may appear: From, Resent-From,
  Resent-Sender, Message-ID, Resent-Message-ID, X-Amavis-Modified,
  Reporting-MTA, macro %h, SQL logging in field, log identity,
  and can be incorporated in variables $smtpd_greeting_banner,
  $smtpd_quit_banner and $hdrfrom_notify_* by specifying a substring
  ${myhostname}, which will be replaced by a current value of $myhostname
  just before use;

- new configuration variables for finer control on propagation of DSN options:
  $propagate_dsn_if_possible and $terminate_dsn_on_notify_success (both part
  of policy banks). One or the other may be used to hide internal mail routing
  from outsiders if desired. Although Postfix can be configured to selectively
  announce or not announce DSN smtp extension based on client address
  (e.g. announcing it for internal clients but not for the rest of the world)
  letting DSN options reach a content filter may be desirable because a
  content filter can provide a more informative delivery status notification,
  and perhaps more importantly, it can suppress sending a DSN when it suspects
  the sender address is faked (viruses, high score spam). This means the
  front-end Postfix smtpd service should not be terminating DSN chain,
  but unfortunately the second instance of smtpd service at port 10025 can
  no longer distinguish between internal and external clients, because it
  only sees an IP address of a content filter. One solution is to turn on
  the $propagate_dsn_if_possible within policy bank MYNETS, and turn it off
  globally, e.g.:

    $propagate_dsn_if_possible = 0;

    $policy_bank{'MYNETS'} = {
      propagate_dsn_if_possible => 1,

  Turning off $propagate_dsn_if_possible is exactly equivalent to the case
  where MTA on the return port (10025) does not announce support for DSN
  extension to the SMTP protocol. The only difference is that the amavisd-new
  setting can be controlled more selectively by a policy bank.

  Turning on $terminate_dsn_on_notify_success is similar, but more refined.
  It tells amavisd not to pass NOTIFY=SUCCESS option when submitting checked
  mail back to MTA, which lets amavisd behave as if talking to a non-DSN
  compliant server, so that DSN success notification will be generated
  by amavisd itself (unless suppressed for other reasons). This is similar
  to $propagate_dsn_if_possible=0, the difference is that other DSN options
  (if any) *will* be passed to MTA, so options like NOTIFY=NEVER or RET,
  ENVID or ORCPT will not be lost. Such behaviour is not strictly by the book
  (RFC 3461), but is still in its spirit :)  Here is the most useful setting:

    $terminate_dsn_on_notify_success = 1;
    $policy_bank{'MYNETS'} = { terminate_dsn_on_notify_success => 0 };

  or if you prefer this syntax, changing only one key in an existing
    $policy_bank{'MYNETS'}{terminate_dsn_on_notify_success} = 0;

- new configuration variable $dsn_bcc (also part of policy banks) and a
  corresponding %dsn_bcc_by_ccat (by-contents-category settings) allows
  to specify an additional e-mail address to receive a copy (blind-CC)
  of a delivery status notification (DSN) that is being sent to the
  envelope sender address; it allows administrator to monitor what
  DSN messages (and how many) are being sent out by amavisd-new;

  Possible uses:
    $dsn_bcc = '';  # applies to all content categories

  or more selectively (overrides $dsn_bcc):
    %dsn_bcc_by_ccat = (
      CC_VIRUS,  '',
      CC_BANNED, '',
      CC_BADH,   '',
      CC_SPAM,   undef,
  or perhaps:
    %dsn_bcc_by_ccat = (
      CC_SPAM,     undef,
      CC_CATCHALL, '',

  Note the use of comma as a delimiter (instead of '=>'), as _values_ of
  CC_* constants must be used as hash keys, not constant _names_ as strings;
  (the Perl '=>' operator implicitly quotes its left operand);

- passive operating-system fingerprinting (p0f) support lets SA gain
  information about SMTP client's operating system and estimated IP distance,
  and can reduce the number of bounces:

  * preconditions are: $os_fingerprint_method must be configured, the process must be running, and amavisd must be receiving
    client IP address information from MTA, which in a Postfix case means
    the XFORWARD protocol extension to SMTP must be enabled in the Postfix
    service feeding mail to amavisd, e.g. "-o smtp_send_xforward_command=yes",
    or in a sendmail/milter setup the more sophisticated AM.PDP protocol
    must be used;

  * find and install the p0f utility:
    or in FreeBSD ports collection as 'net-mgmt/p0f';

  * start a p0f process on the same host where MTA (MX) is running, making
    it listen only to incoming TCP sessions (to reduce its workload) to the
    IP address and TCP port (25) where MTA is accepting incoming mail from
    outside (it doesn't hurt to let it see other traffic too, it just isn't
    needed); after testing p0f alone and seeing that it works, you may start
    it up, feeding its output to program that comes with
    amavisd-new package, e.g.:

      p0f -l 'tcp dst port 25' 2>&1 | 2345 &

    on multi-homed boxes one may need to specify interface and IP address
    where MTA is listening, the filter syntax is the same as in tcpdump, e.g.:

      p0f -l -i bge0 'dst host and tcp dst port 25' 2>&1 \
        | 2345 &

  * the program reads p0f reports on stdin, keeps a cache
    for a limited time (10 minutes, configurable) of data about incoming TCP
    sessions organized by remote IP address, and listens on UDP port 2345
    (specified as its command line argument) for queries; only queries from
    allowed IP addresses are accepted and responded to, other queries are
    silently ignored - configure @inet_acl accordingly, defaults to;

  * adding the following line to amavisd.conf, matching the chosen port
    number to the one specified on the command line to the

      $os_fingerprint_method = 'p0f:';

    makes amavisd send queries to (on the supplied IP address
    and UDP port number) to collect information about remote SMTP client's OS;
    collected response is then supplied as a header field when SpamAssassin
    is invoked;  query/response is very quick and imposes no burden on amavisd
    process nor does its extend its processing time. The $os_fingerprint_method
    setting is also a member of policy banks to make it more flexible to
    disable fingerprinting for mail from site's own SMTP clients, e.g:

      $policy_bank{'MYNETS'}{os_fingerprint_method} = undef;

  * one may now add scoring rules to SA file, e.g.:

    describe __L_P0F_EXISTS  A header field X-Amavis-OS-Fingerprint does exist
    header   __L_P0F_EXISTS  exists:X-Amavis-OS-Fingerprint

    describe L_P0F_WXP    Remote system is truly a Windows XP, not Windows 2000
    header   L_P0F_WXP    X-Amavis-OS-Fingerprint =~ /\AWindows XP(?![^(]*\b2000 SP)/m
    score    L_P0F_WXP    2.3

    describe L_P0F_W      Remote system is some Windows variant, except Win. XP
    header   L_P0F_W      X-Amavis-OS-Fingerprint =~ /\AWindows(?! XP)/m
    score    L_P0F_W      1.3

    describe L_P0F_UNKN   P0f was unable to determine remote OS type
    header   L_P0F_UNKN   X-Amavis-OS-Fingerprint =~ /\AUNKNOWN/m
    score    L_P0F_UNKN   0.8

    describe L_P0F_Unix   Remote system is running Unix, not Linux
    header   L_P0F_Unix   X-Amavis-OS-Fingerprint =~ /\A((Free|Open|Net)BSD|Solaris|HP-UX|Tru64|AIX)/m
    score    L_P0F_Unix   -1.0

    describe L_P0F_Unix   Remote system is running Linux
    header   L_P0F_Linux  X-Amavis-OS-Fingerprint =~ /\ALinux/m
    score    L_P0F_Linux -0.1

    It is also possible to add score based on estimated IP distance, for
    example to slightly favorize nearer hosts (this is probably good for Europe
    or academic/university networks, and possibly less useful elsewhere):

    header L_P0F_D1234 X-Amavis-OS-Fingerprint =~ /\bdistance [1-4](?![0-9])/m
    header L_P0F_D5    X-Amavis-OS-Fingerprint =~ /\bdistance 5(?![0-9])/m
    header L_P0F_D6    X-Amavis-OS-Fingerprint =~ /\bdistance 6(?![0-9])/m
    header L_P0F_D7    X-Amavis-OS-Fingerprint =~ /\bdistance 7(?![0-9])/m
    header L_P0F_D8    X-Amavis-OS-Fingerprint =~ /\bdistance 8(?![0-9])/m
    header L_P0F_D9    X-Amavis-OS-Fingerprint =~ /\bdistance 9(?![0-9])/m
    header L_P0F_D10   X-Amavis-OS-Fingerprint =~ /\bdistance 10(?![0-9])/m
    header L_P0F_D11   X-Amavis-OS-Fingerprint =~ /\bdistance 11(?![0-9])/m
    score  L_P0F_D1234 -0.5
    score  L_P0F_D5    -0.5
    score  L_P0F_D6    -0.5
    score  L_P0F_D7    -0.5
    score  L_P0F_D8    -0.5
    score  L_P0F_D9    -0.4
    score  L_P0F_D10   -0.3
    score  L_P0F_D11   -0.3

    It can be useful to tame false positives from some other rules, e.g.

    # tame a Botnet plugin, reducing its false positives
    score BOTNET    0.1
    meta  BOTNET_W  !DKIM_VERIFIED && (L_P0F_WXP || L_P0F_W) && BOTNET
    score BOTNET_W  2.8
    score BOTNET_WU 2.0
    score BOTNET_OTHER  0.5

  * make sure the @mynetworks is configured correctly, otherwise you will be
    inappropriately penalizing mail from internal hosts running Windows!
    Other methods to turn off fingerprinting for our own SMTP client hosts
    is to put $os_fingerprint_method in policy banks, and/or to specify
    more selective packet filter on the p0f command line;

  * based on statistics, less than 0.7 % of mail coming from external
    Windows XP -based hosts is ham, yet 20 % of all spam is coming from
    external Windows XP hosts; amavisd-new suppresses bounces to external
    Windows XP hosts, reducing bounce pollution. The amavisd-agent utility
    now provides some additional statistics based on p0f information.

    Some statistics collected from our logs in February 2006:
    p0f OS guess    ham :   spam
    Windows-XP    0.7 % : 99.3 %
    Windows-2000  5.8 % : 94.2 %
    UNKNOWN      16.5 % : 83.5 %
    Linux        58.8 % : 41.2 %
    Unix         80.3 % : 19.7 %
    (Unix+Linux  66.5 % : 33.5 %)
      (ham: mail with score below 3,  spam: score above 6)

- new configuration variable $allow_fixing_improper_header_folding (also
  a member of policy banks) controls fixing improperly folded header fields
  made up entirely of whitespace (prohibited by RFC 2822) by removing
  all-whitespace continuation lines;  defaults to true for backwards
  compatibility; fixing such header fields is desirable and can protect
  poorly written mail readers, but may break DomainKeys/DKIM validation of
  messages with illegal header, so if this is of concern, one has a choice
  of turning it off (if local MUAs can't be fixed);

- make config variable $insert_received_line a member of policy banks;

- removed mail header (macro %H) from the default template of the plain text
  part of the virus or banned sender notifications; these headers are available
  in the third MIME part of the DSN, so having them twice was redundant;

- new macros: date_unix_utc, date_iso8601_utc, date_iso8601_local,
  date_rfc2822_local, x-mailer, header_field, ccat_name, ccat_maj, ccat_min,
  wrap, lc, uc, substr, index, len, incr, decr and a couple of SpamAssassin
  lookalike macros - see README.customize for details;

- new macro ccat_min expands to a minor contents category; this makes it
  possible for a notification template to distinguish (for example) between
  cases of bad message header: 1=bad MIME, 2=8-bit char, 3=NUL/CR, 4=empty,
  5=long, 6=syntax error;  See also README.customize for the list of currently
  available macros; see %ccat_display_names for minor ccat numbers currently
  in use, feel free to add new ones;

- edited text of default notification templates to make them tidier and
  more informative; also make a good use of new macros;  older templates
  are still compatible with this version of amavisd-new;

- new global variable $timestamp_fmt_mysql, defaults to false;
  specific to MySQL, when logging to SQL is enabled and field msgs.time_iso
  is declared as TIMESTAMP instead of char(16); setting $timestamp_fmt_mysql=1
  changes the timestamp format written to field msgs.time_iso to avoid ISO 8601
  standard delimiter 'T' and trailing timezone 'Z', which MySQL does not like.
  Don't turn on $timestamp_fmt_mysql when using PostgreSQL!
  Should not turn on $timestamp_fmt_mysql with MySQL unless msgs.time_iso is
  of type TIMESTAMP.

- new config variable $trim_trailing_space_in_lookup_result_fields controls
  trimming of trailing whitespace from SQL fields, from LDAP attribute values
  and associative array righthand-sides (hash values) as read by read_hash();
  disabled by default; turn it on for compatibility with pre-2.4.0 versions.

  Note that trailing spaces may still be trimmed by SQL itself (contrary to
  SQL-99 specification). Trailing spaces in MySQL version up to and including
  4.1 are removed from values when stored in a VARCHAR column; this also means
  that spaces are absent from retrieved values. Starting with MySQL V5.1,
  trailing spaces are retained when values are stored and retrieved.
  Investigated by Gary V;

- treat exit status 2 as a warning when returned by bzip2, gzip and other
  decompressors handled by do_uncompress; problem with decoding of corrupted
  bzip2 file reported by Kim Leandersson;

- when determining file short type, match "Microsoft Cabinet file"
  result from a file(1) command case-insensitively; reported and
  patch provided by ap at zip com au via Debian bug tracking system;
  also recognize "Microsoft Cabinet archive data" as .cab;

- do_unzip: set attribute 'U' (undecodable) if zip archive fails to be
  decoded; based on a patch by Oliver Geisen; -- well, on a second thought,
  perhaps not, this line is now commented out; it flagged too many bounces
  containing chopped-off ZIP attachments as ***UNCHECKED***;

- added a minimalistic decoder interface routine to call a command line
  unpacker from to decode Macintosh StuffIt archives. Not tested
  extensively, program source is not available for inspection, use at YOUR
  OWN RISK (and the risk is non-negligible!). If using non-default assignment
  to @decoders, the following entry can be added to the list of decoders:
  thanks to Oliver Geisen for the suggestion;

- keep X-Spam-Level bar empty if sender is whitelisted;

- untaint recip_score_boost when writing a log report entry to SQL;
  'Insecure dependency in parameter 7' could have been reported when
  SQL-based score_sender lookup table is used; reported by Jim Knuth;

- treat undefined spam level as 0 for the purpose of comparing it to
  tag/tag2/kill levels, e.g. when spam scanning is skipped due to large
  message size; this now allows score_sender_maps to push non-checked
  messages over a tag2/kill limit by its score boost if desired;

- header fields X-Spam-Status, X-Spam-Score, X-Spam-Level and X-Spam-Flag
  in a quarantined message now take into account also the score_sender boost
  and white- and blacklisting (using 'any' and 'max' to summarize in case
  of multiple recipients);

- in passed and quarantined mail a header field X-Spam-Status now shows
  score as an explicit sum of SA score and a by-recipient score_sender boost
  (when the boost is nonzero); the X-Spam-Score header field still shows
  a sum of both as a single number so as not to confuse MUA filters which
  may operate on that header field;

- insert X-Quarantine-ID header field into passed mail if a passed message
  was also quarantined (e.g. *_lover or final_*_destiny=D_PASS), suggested
  by Pavel Urban;

- when folding is needed for long new or edited header fields, use TAB
  instead of a space on fold points;  also: take into account the apparent
  display size of expanded TABs instead of the actual character count,
  to make header look prettier; use of TABs suggested by Debian community; 

- improved wrapping of inserted header fields and in generated text sections
  of notification templates;

- automatic wrapping of long header lines from notification templates;

- standards compliance: wrap Diagnostic-Code field in message/delivery-status
  section of delivery status notifications according to rfc3461 section 9.2;

- added config options to enable quarantining (archiving) of clean mail;
  defaults settings are:
    $clean_quarantine_method = undef;
    $clean_quarantine_to  = 'clean-quarantine';
  quarantining clean messages is disabled by default;  to enable:
    $clean_quarantine_method = 'local:clean-%m';

- reports "Blocked TEMPFAIL" instead of "Blocked CLEAN" in case of a
  temporary 4xx failure;

- in generated MIME parts (notifications, defanging) replace suggested
  file names like 'message.txt', 'dsn_status.txt' and 'header.txt' with
  names without an extension, i.e. 'message', 'dsn_status' and 'header';
  Reportedly Outlook Express 6.0 (but not Outlook) determines how to present
  a MIME part based on its name extension, instead of based on MIME type.
  Depriving it of name extensions makes it obey a MIME type. This solution
  has been successfully tested with Outlook (5 & 6), Outlook Express (2k, 2k3),
  Thunderbird, JavaMail and Squirrelmail. Investigation and suggestion by
  Ivers Matthias;

- clamscan AV entry: change test for status 1 by a test for /:.*\sFOUND$/
  to prevent system errors like failed load (e.g.: ' Shared object
  "" not found', which also produce exit status 1) from causing
  all mail to be treated as infected; suggested by Tomasz Kojm in response
  to my problem report;

- add three AV entries for avast products (FreeBSD and Linux):
  'avast! Antivirus daemon', 'Client/Server Version' and the command-line
  avastcmd; kindly provided by Frantisek Mensik, ALWIL Software;
  later refined based on testing and feedback from Bill Landry;

- updated regexp in McAfee uvscan entry to cope with spaces in virus name;
  reported by, and output samples provided by Andreas Schulze;

- updated bdc AV entry: newer versions of BitDefender don't use option --all
  any longer; thanks to Max Matslofva, Andreas, Gary V and Bill Landry;

- updated NOD32 AV entry based on NOD32 documentation and advise
  from Willi Gruber about adding status code 3 to the list of success values;

- better handle failed decoder attempts when checking an executable file
  for self-extracting archive (SFX), avoid decoder 'dry runs' where possible;

- require minimal version 0.32 of Net::LDAP, a subroutine
  Net::LDAP::Util::escape_filter_value() is needed; reported by Harry Hoffman;

- allow to specify option 'deref' in calls to Net::LDAP->search to control
  dereferencing of aliases to locate the base object for the search; the
  default remains 'find' as before. Use a key 'deref' in the default_ldap
  hash to specify a different value. The values are those documented
  in the Net::LDAP manpage in the search function under the deref section;
  (btw, dereferencing in LDAP is supported in both Postfix and Courier);
  a patch provided by John Allman;

- added a timing report entry fwd-data-cmd for time it takes MTA to respond
  with a status to a DATA command;

- read_hash: trim whitespace off the right-hand side data only if followed by
  a comment (#), otherwise keep the rhs as it is;

- include a currently chosen dataset name (dsn) in the log entry
  when quarantining to SQL, to be able to retrieve a quarantined message
  from the correct SQL server; suggested by Cami Sardinha;

- dwell in $tempdir (chdir to) instead of in $TEMPBASE most of the time;

- edit_header() may now be called multiple times for the same header field
  to provide iterative header edits - provided mostly for completeness;

- Courier code (the patch) now supports D_BOUNCE and D_DISCARD message
  destinies and adding and deleting recipients (due to differing per-recipient
  configuration or addr_extension_*_maps); it also supports DSN (RFC 3461);
  by Martin Orr;

- prepend (!!) or (!) to log messages at levels below 0 to facilitate
  log parsing and make critical messages stand out; based on a patch
  by Henrique de Moraes Holschuh (Debian port maintainer);

- silence logging of AM.PDP commands delivery_care_of, tempdir_removed_by
  and tempdir when using amavisd-milter; a patch by Petr Rehor;

- silence Perl warnings about uninitialized values;

- optimization: about 15% speedup in macro expansion due to pre-tokenization
  of template strings at startup (affects preparing main log entry and each
  notification message);

- modified old amavis.c client program to return status 0 (success) even when
  LDA command-line arguments are not specified (i.e. when amavisd daemon is
  in charge of delivery), and amavisd daemon returns status 99 (=discard);
  simplified code in amavisd/check_amcl_policy() to take advantage of it;

- internal/coding: a much needed and appreciated hefty patch from Martin Orr
  introduces class Amavisd::TempDir and collects there existing code from
  various places dealing with maintaining a temporary directory and files
  within; it also makes possible for the Courier interface code to use
  this module and avoid duplicating code;

- internal/coding: the Amavis::In::Courier package has been tidied up and split
  into multiple methods, of a hopefully manageable length. Several new features
  in amavisd-new core code, which required changes to Amavis::In::Courier to
  take advantage of, have also been dealt with (in particular MYNETS and
  MYUSERS policy banks and SNMP counters); by Martin Orr;

- require minimal version 1.43 of DBI, working last_insert_id is needed;
  (actually last_insert_id is no longer needed, but the requirement stays :)

- internal/coding: when logging to SQL avoid a need for last_insert_id()
  by doing SELECT after INSERT when adding a new e-mail address to table maddr;
  this also avoids a rollback/retry when more than one process tries to insert
  the same new address into a database;

- internal/coding: merge subroutines do_spam and do_virus into a single
  do_notify_and_quarantine, and use the same code to prepare spam-describing
  headers as in subroutine add_forwarding_header_edits_per_recip;

- internal/coding: new $msginfo data object: contents_category, holding
  a reference to a sorted (descending order) array of entries, each one
  corresponding to one contents category under which a message can be
  classified, e.g. it can be both a CC_VIRUS and a CC_BANNED. Supporting
  subroutines are: add_contents_category, main_contents_category,
  is_in_contents_category, setting_by_contents_category;
  thanks for suggestions and feedback to Gérald Macinenti;

  The contents_category list is a sorted list of strings, each of the form
  "major" or "major,minor", where major and minor are numbers, representing
  major and minor category type. Sort order is descending by numeric values,
  major first, and subordered by a minor value. When an entry "major,minor"
  is added, an entry "major" is added automatically (minor implied to be 0).
  A string "major" means the same as "major,0". See CC_* constants for major
  category types. Minor category types semantics is specific to each major
  category, higher number represent more important finding than a lower number;

- internal/coding: merge sections in sub check_mail dealing with different
  content types one-at-a-time into a single section 'decide_mail_destiny',
  taking advantage of the new information on contents category, improving
  consistency and simplifying code;

- internal/coding: new $msginfo data objects: spam_level, spam_status,
  spam_report, autolearn_status, avoiding ugly global variables;

- internal/coding: separate SpamAssassin-specific code from general anti-spam
  code - new module Amavis::SpamControl::SpamAssassin;  based on suggestion
  and patch by Felix Schwarz;

- internal/coding: use File::Spec::catfile to splice full file name from its
  components (in mail_to_local_mailbox, more needed); a patch by Felix Schwarz.
  Actually, on a second thought, comment that out and revert to previous code:
  there are so many other similar cases which were not generalized, that
  it makes no sense to generalize (through File::Spec) one percent of them
  (and drag-in yet another Perl module), and leave the rest hard-coded;

- collect the most commonly needed header fields into an associative array
  $msginfo->orig_header_fields, removing the need to call ensure_mime_entity()
  in certain cases, saving on mime decoding when it is not really needed;

- internal/coding: do_log now takes optional arguments, and if they are
  present, the message text is treated as a format string to snprintf.
  Take advantage of this in most calls to do_log. The message (format) argument
  should not be tainted (not enforced), but arguments may be. In the absence
  of additional arguments, do_log behaves as before;

- updated README.postfix to explicitly override (just in case) two newer
  Postfix options: smtpd_data_restrictions and smtpd_end_of_data_restrictions,
  thanks to Noel Jones for the suggestion;

- documentation: updated README.sendmail-dual - added custom rules to reject
  unknown users outright; provided by Matej Vela, thanks to Simone Marx;
  added a reference to the 'milter-ahead' project info; thanks to Adam Gibson;

- documentation: fixed README.exim_v4 (don't let messages with null
  return path get through unchecked), by Igor D'Astolfo;

- documentation: updated README.customize, describing new features of a
  built-in macro expander, and describing new macros;

- documentation: updated README.sql, describing new SQL log purging
  recommendations, improved PostgreSQL instructions, and the (optional) use
  of data type TIMESTAMP in field msgs.time_iso;

                                                            August 22, 2005
amavisd-new-2.3.3 release notes

Version 2.3.3 is a maintenance release over 2.3.2. Besides fixing known
problems and providing some optimizations, no new features were added.
If using SpamAssassin older than 3.1, an upgrade of either SA to 3.1,
or an upgrade of amavisd-new to 2.3.3 is recommended.

- privacy: add a safety fuse / workaround around calls to SA to detect
  SA's failure (in SA versions before 3.1) to catch a failed exec() in a
  forked process, which could produce runaway process clones. See SA bug
  report #4370. An incident of a mail copy being delivered to unrelated
  recipient reported by Joel Nimety;

- privacy: turn warning into a fatal error when a quarantine ID of a message
  requested for a quarantine release does not match the requested mail_id;

- security: require minimal version 1.35 of Compress::Zlib to avoid
  vulnerability in the zlib compression library;

- the dsn_cutoff_level should have been ignored if undefined according to
  documentation, but was not, causing DSN to be suppressed regardless of
  spam level; discovered by Gary V;

- ensure the banned check is not performed if all recipients agree
  it is not needed, even in presence of $banned_namepath_re;
  undesired behaviour (not strictly incorrect) reported by Joel Nimety;

- missing import of lookup_ip_acl in module Amavis::In::AMCL caused
  failure in sendmail milter setup when using the new AM.PDP protocol;
  reported by Mic And;

- document and explicitly define handling of syntactically invalid IP address
  in lookup_ip_acl: it matches a zero-length-mask net, a constant lookup table,
  or a hash entry with an undef key, but no other entries in IP lookup tables;
  syntactically invalid IP addresses are now logged;

- fix parsing of IPv6 address in $notify_method and $forward_method in case
  of dynamic destination override (the use of '*' in method fields);

- check during startup that $myhostname is a fully qualified domain name
  (or 'localhost', if you must), and abort if it isn't, otherwise a non-FQDN
  can end up in places where RFC 2822 does not allow it; if uname(3) does not
  provide a FQDN, then an assignment to $myhostname must be done explicitly
  in amavisd.conf;

- when quarantining to a single file in mbox format the 'From ...' line
  needs an English date, regardless of current locale; fixed by globally
  setting locale LC_TIME to "C";

- pass on the parameter BODY=8BITMIME on MAIL FROM when submitting to MTA
  when original message reception indicated it is needed (RFC 1652).
  Note that mail forwarding may now fail if the feeding MTA requests
  BODY=8BITMIME SMTP service extension (or just passes data with msb set),
  but the MTA on the output side does not allow the use of the BODY parameter
  in SMTP. In case of Postfix this may only happen when receiving service
  on port 10025 is misconfigured and does not announce ESMTP capability
  and support for the SMTP service extension 8BITMIME;

- RFC 2554 requires auth_param to be xtext-encoded addr-spec (no angle
  brackets) or "<>", not the xtext-encoded addr-spec enclosed in angle
  brackets (when specifying submitter during authentication); fixed;

- apply some sanity limit on collected bad-header samples to ensure that
  a grossly broken mail does not unnecessarily fill up memory;

- when sending recipient warnings for viruses, banned files, or bad headers,
  recipient address must not be rfc2822-quoted twice; fixed;

- fix interpretation of $defang_all to really imply all; previously it only
  affected clean messages;

- in quarantined mail the reported spam score in X-Spam-Status header field
  now includes maximum of all by-recipient score boosts (less surprising
  when soft-whitelisting through @score_sender_maps is in use); suggested
  by Mike Cappella and Gary V;

- when a policy delegation protocol attribute "request" is not "AM.PDP"
  (perhaps it is a Postfix policy delegation request) don't attempt to find
  and open a mail file;

- do_ascii and do_unarj: set environment variable TMPDIR on a command line
  temporary directory option to "$tempdir/parts" instead of $TEMPBASE
  to minimize possible pollution of top level directory;

- don't abort even if amavisd.conf returns undef as a final value,
  as long as there are no errors reading or interpreting it;

- if during 'amavisd stop' or 'amavisd reload' the old running daemon does
  not go away for one minute after sending it a SIGTERM, use a bigger
  hammer and send it a SIGKILL; suggested by Sven Riedel;

- extend LDAP lookups to allow multiple search attributes (multiple
  occurrences of %m in a query); a patch by Michael Hall (and a similar
  one by Matthias Bandemer);

- LDAP lookup on an empty envelope address (e.g. a null return path)
  adds another lookup key "<>", as it is difficult if not impossible
  to have LDAP attributes with empty string as a value; by Michael Hall;

- LDAP.schema: drop "MUST ( mail )" from objectclass 'amavisAccount';
  suggested by Michael Hall;

- updated comments and documentation, most notably the README.chroot;

- contributed file Macintosh.tar.gz updated by Dale Walsh;


- replaced 'hits=' with 'score=' in inserted X-Spam-Status header field
  (and in some internal log entries) for compatibility with a changed
  default in SpamAssassin 3.1;

- insert X-Spam-Score header field for compatibility with SA (previously
  insertion of this header field was commented-out because the information
  is redundant, as the score already appears in X-Spam-Status);


- speed up sending a mail header of full defanged (rewritten) mail over SMTP
  back to MTA by a factor of 4 by buffering header fields into large chunks
  to avoid bottleneck in Net::Cmd::datasend, which has lots of overhead for
  line-by-line writes. Previously slow writes mostly affected mail messages
  with extreme header lengths (such as results of a broken mail loop), or
  when delivering defanged messages, particularly at sites with large MTA
  mail size limits, sometimes to a point of exceeding timeout limits;
  reported by Dominik Weber and Ralf Hildebrandt;

- move subroutine lookup_ip_acl() and associated ip_to_vec() into its own
  dedicated new package Amavis::Lookup::IP; provide a constructor to pre-parse
  IP lookup tables to speed up IP lookups in lookup_ip_acl; prepare pre-parsed
  commonly used IP lookup tables (@mynetworks_maps, @publicnetworks_maps,

- optimized reading loop in SMTP DATA state, receiving data is now about
  35% faster when mail size limit is not enforced (which is a default);
  no speedup (nor slowdowns) when mail size limit _is_ enforced;

- cache results of evaluated macros during a single call to expand(),
  as macro calls often come in pairs, like:  [?%e||\[%e\] ]
  or [? %#T ||, Tests: [%T|,]];  together with the above optimization in
  pre-parsed IP lookups it shaves off 25% of time in preparing main log entry;

- set locale LC_TIME to "C" globally, avoid changing and restoring locale
  for every log write and when generating RFC2822 timestamps;

- added an optimization note in README.sql about indexes and about
  SELECT count(*) in MySQL with InnoDB; investigation by Paolo Cravero;

                                                              June 29, 2005
amavisd-new-2.3.2 release notes

INCOMPATIBILITY with 2.3.1 and earlier versions:

If running amavisd daemon in chroot please note:

  Each child process now opens its own syslog connection or a file descriptor
  to a log file, and no longer inherits a connection from its parent.
  When running in chroot jail and logging to syslog, the syslog client
  routines need syslogd socket to be present in the chroot subtree to be
  able to establish a connection with syslogd, otherwise logging output
  may be lost. Additional syslogd sockets (to be made available in the
  jail) may be requested from the syslogd daemon, see its documentation.
  This requirement is equivalent to the requirement of chrooted Postfix
  services (see Postfix documentation file BASIC_CONFIGURATION_README).

BUG FIXES since 2.3.1:

- do not enforce $MAXFILES limit during top-level MIME decoding to avoid
  tempfailing mail;  MIME parts are still counted, so a limit exceeded may
  still be reported during subsequent decoding, but this is handled more
  gracefully and does not cause preserved temporary directories to be left
  behind; reported by Marcin Lemanski; suggested by Stephane Lentz and
  Robert LeBlanc (noted in the 2.0 release notes);

- use recv() instead of read() to get results from daemonized virus scanners
  in an attempt to avoid a bogus Perl I/O status on some Linux installations
  (reported by Sander Steffann); we now get a meaningful status codes like
  ECONNRESET instead of a bogus EBADF (Bad file descriptor);

- ignore status ECONNRESET when reading results of a daemonized virus scanner
  from a socket, specific to some Linux versions; thanks to Sander Steffann
  for the initial report and extensive help in debugging the Perl problem;

- run_av and other similar code sections: replace line-by-line reads by
  block-by-block reads wherever possible to avoid inappropriate status report
  EBADF (Bad file descriptor) caused by Perl I/O bug when last line is not
  terminated by a newline. The problem was affecting reading response from
  some command line virus checkers; reported by Sander Steffann;
  Perl bug tracking: #39060: readline of a not NL-terminated last line
  results in Bad file descriptor;

- ignore status EAGAIN when reading results on a pipe from a forked process;
  the status EAGAIN seems to be an artifact of Perl I/O on some installations;
  reported by several people to cause problems on FreeBSD with Perl 5.8.7
  (but Perl 5.8.6 is fine); thanks to Bart Matterne for testing and feedback;

- allow one level of indirection when collecting %needed_protocols;
  global setting $protocol='COURIER' did not work, a workaround was needed
  with previous version, e.g.: $policy_bank{'QMQPqq'}={protocol=>'QMQPqq'};
  reported by Nicklas Bondesson and Martin Orr;

- fix a bug (introduced with 2.3.0) in Courier and QMQPqq setups, where global
  information about processed message wasn't always reset and could leak
  into processing of a subsequent message; reported by Nicklas Bondesson;

- SQL: fix arguments in calls to last_insert_id(), failing under PostgreSQL
  (MySQL didn't mind); pointed out by Henrik Krohns;

- if module SAVI is loaded, insist it is version 0.30 or later;
  incompatibility with earlier versions reported by Andrzej Kukula;

- make use of the new Net::Server 0.88 hook run_n_children_hook() to
  reload SAVI database; removes a need to apply SAVI patch to Net::Server;
  the Net::Server hook was suggested by Paul B. Henson and others,
  and incorporated into Net::Server 0.88 by Paul Seamons;

- reopen log file or syslog connection in each child process to make it use
  its own file descriptor; also minimizes transients when syslogd is restarted
  and its socket re-created, as reported by Les Ault. When running in chroot
  please make sure a syslogd socket is also available in the chroot jail,
  see README.chroot for syslogd options (and BASIC_CONFIGURATION_README
  in Postfix documentation for the Postfix equivalent);

- close log file or syslog in forked process before exec, just to play nicely;

- do_lha: fix extracting archive member filename in case of broken archive
  or empty name (avoid interpreting creation date as a file name);
  do not increment OpsDecByLha counter for empty archives, which are
  most likely not lha archives at all;

- obey $final_bad_header_destiny D_DISCARD or D_REJECT even for messages
  with bad headers from mailing lists or with a null envelope sender (DSN);
  previously such messages were passed; undesired behaviour reported
  by Cami Sardinha.

  Such messages are still let through with $final_bad_header_destiny set to
  D_BOUNCE, as otherwise they will be lost because a bounce is suppressed
  for null sender messages and for mail from mailing list. This behaviour
  is retained for backwards compatibility, but may need to be reconsidered.

- fix regexp for extracting am_id from amavis-milter helper program requests;

- if fork/exec fails, try to commit suicide in forked process with
  POSIX::_exit(1) first, before trying kill('KILL',$$) as a last resort;

- updated $log_templ example in amavisd.conf-sample to match the default;
  pointed out by Gary V;

- further reduce a couple of more frequent Perl warnings about the use of
  uninitialized values in expressions;

- pre-load additional Perl modules required by SA 3.1 plugins;

- require minimal versions of modules: Time::HiRes 1.49, Archive::Zip 1.14;

- replaced nonexistent variable @sa_spam_modifies_subj_maps by
  @spam_modifies_subj_maps in commented-out example in amavisd.conf-sample;
  noticed by Joachim Schoenberg;

LDAP CHANGES by Michael Hall:

All the LDAP changes are transparent to the user.

- rewritten some of the code similar to the restructuring of the SQL code
  in version amavisd-new-2.3.0. A new package Amavisd::LDAP::Connection was
  added which is a LDAP connection object, and the old connection-related code
  in Amavis::Lookup::LDAP has been moved to the new package. Amavisd-new will
  now try to reconnect (once) while processing a message, similar to SQL;

- added the ability to specify a '%d' (domain) token in the LDAP base DN;
  based on idea from Alexander Wittig;

- updated default LDAP port based on whether SSL/TLS is being used or not;
  based on idea from Timo Veith;

- updated the search code to query for multiple records and return the results
  sorted in 'make_query_keys' order versus doing a query for each key.
  As a result performance is enhanced, and the tweaks 'ldap_get_all', and
  'use_query_keys' (recently added) are no longer applicable or needed
  and have been removed;

- improved LDAP error reporting and misc changes to multivalued attributes;

- documentation changes (amavisd.conf-default, README.lookups);


- macro %c (commonly used in a log template) reports spam score no longer
  as a single number, but as an explicit sum of a SA score and a by-sender
  boost score (from @score_sender_maps) when boost score is nonzero;
  suggested by Ed Walker;

- enhancement to amavisd-release: if its only command line argument is '-',
  then read arguments from stdin, one release request per line, ignoring empty
  lines; input lines have the same format as command line arguments, i.e.:
     mail_file secret_id
     mail_file secret_id alt_recip1 alt_recip2 ...

- better handle cases where a persistent temporary file email.txt
  as prepared by the SMTP server module gets replaced as a result
  of some user program modification (e.g. when invoking altermime);
  problems reported by Dinesh Shah and Leonardo Rodrigues;

                                                                May 9, 2005
amavisd-new-2.3.1 release notes

INCOMPATIBILITY with 2.3.0 and older versions:

- command line option 'foreground' no longer automatically redirects logging
  to STDERR; to request logging to STDERR turn off the $DO_SYSLOG and $LOGFILE;
  the improved flexibility suggested by Matthias Andree and Ralf Hildebrandt;

BUG FIXES since 2.3.0:

- don't test errno when line-by-line reading loop is exited before eof,
  it was inappropriate and Perl on some versions of Linux does not like it,
  possibly reporting "Error reading mail header: Bad file descriptor";
  Besides fixing the loops that needed the fix, modified also all remaining
  reading loops for consistency. Reported by Ralf Hildebrandt;

- don't call $per_msg_status->get_autolearn_status with SA older than 3.0;
  reported by Ian Abbott;

- pre-load some additional SpamAssassin modules, needed by SA 3.1 (CVS);

- reading from dspam process used wrong variable, producing empty lines
  for SA checking; reported by Chris Lewicki;

- SAVI-Perl: libsavi option for turning on mime parsing is 'Mime', not 'MIME';
  libsavi is case-sensitive since version 3.93.0 and was rejecting incorrect
  option name; thanks to Paul B. Henson;

- fixed disabling of SQL wblist ($sql_select_white_black_list=undef);
  bug reported by Tom Sommer;

- do_tnef: extract data from attribute 'Attachment' in addition to data from
  a more common attribute 'AttachData'; example data provided by Goetz Rieger;

- avoid some more frequent Perl warnings on the use of uninitialized variables;

- add prototypes for decoding subroutines; prototype mismatch warnings
  reported by Michael Muenz;

- fixed prototype for add_forwarding_header_edits_common(),
  thanks to Ian Abbott;

- replace a simple-minded logic for loading input protocol modules
  by a slightly more sophisticated code which takes into consideration
  field 'protocol' in all defined policy banks; reported by Brian Wong;

- when replacing existing address extensions don't treat the whole
  localpart as an extension if the address happens to start with
  a $recipient_delimiter; pointed out by Kaj J. Niemi;


- unfold obsolete-syntax folded header fields composed entirely of
  white space (RFC 2822 section 4.2); suggested by Ian Abbott and others.
  Note that such unfolding breaks DomainKeys/DKIM "simple" canonicalization
  algorithm (but is transparent to "nofws") if the affected header is
  included in the signature;

- do_tnef_ext: add support for decoding TNEF containers by an external
  program 'tnef' if available; selectable by an entry in the @decoders list;

- mail_via_bsmtp enhancement: substitute %s in quarantine filename template
  by a defanged sender name; based on a patch by Thomas Jarosch;

- lookup_ip_acl enhancement: when a supplied lookup table is an associative
  array (a hash) and the looked-up address is an IPv4 address, allow simple
  classful subnet specification by repeatedly truncating the trailing byte
  from the looked up IPv4 address until a match is found or until further
  truncation is not possible. Note that this does not apply to IPv6 addresses.
  For more flexible CIDR subnet specifications please use lookup arrays.

- provide a routine read_array, which is modelled after read_hash, but stores
  lines read from a file into an array lookup table, instead of a hash.

                                                             April 24, 2005
amavisd-new-2.3.0 release notes


Provides more flexible configuration of decoders. Allows recipients to have
individual banning rules. Assigns a long-term unique id to each message,
reducing clashes and facilitating retrieval of information. The daemon can
store information to a SQL database for logging, reporting and quarantine
retrieval, optionally storing entire message to a SQL database. File-based
quarantine can disperse files to 62 subdirectories. Provides a quarantine
release mechanism. Reconnects to SQL if connection is broken. Can skip
quarantining high-score spam. Compatibility with IPv6-enabled Postfix
is improved.


- require minimal version 1.05 of Convert::UUlib to avoid
  a security problem in the underlying uulib:
  which is now known to be exploitable (2006-12-05), credits to
  Jean-Sébastien Guay-Leroux;

INCOMPATIBILITY with 2.2.1 and older versions:

- structure of @banned_filename_maps config variable has changed in
  incompatible way to allow per-recipient banned rules: it is now a
  two level map indexed by recipient address (similar to the structure
  of @score_sender_maps). See further down for details.

  The change will not affect existing installations which either:
    * leave @banned_filename_maps at its default value and use the traditional
      $banned_filename_re configuration variable to specify banned rules,
      which most installations do and remains the most commonly used method;
    * or, assign to @banned_filename_maps an empty list to disable it;
    * or, use the alternative mechanism $banned_namepath_re and disable
      $banned_filename_re or @banned_filename_maps.

- macro %i (used in log and other templates) now always shows mail_id
  (see below) which is often but not necessarily also the name of a
  quarantined file (like before, available as a macro %q);  previously the
  %i reflected the actual file name, which was longer/different than now;

- default values of $virus_quarantine_method, $spam_quarantine_method,
  $banned_files_quarantine_method and $bad_header_quarantine_method
  now specify shorter file names based on the newly-introduced mail_id:
  virus-%m, spam-%m.gz, banned-%m and badh-%m respectively;

- a config variable $file now defaults to a string 'file' instead of being
  undefined; this makes it unnecessary to be explicitly assigned to in the
  configuration file;

- SQL fields virus_lover, bypass_virus_checks and bypass_spam_checks
  in table 'policy' are now optional, and if missing their value is treated
  as undef (same as if the field is present but is NULL) - instead of being
  interpreted as false. This makes it consistent with the interpretation
  of other missing fields. The change is unlikely to affect existing
  installations, because these fields were considered non-optional in
  previous versions.

BUG FIXES or missing functionality:

- avoid modifying directory which is being read by readdir, it can cause
  premature termination of the directory traversal; reported by Dale Walsh;

- minimize deep recursion in regexp evaluation while parsing some degenerate
  cases of Received header subfields via/with/id/for, which could result
  in a process crash;

- turned loops in banned checks inside out to make 'allow' rules useful and
  let them behave as one may expect: when checking mail parts against lookup
  tables in @banned_filename_maps the evaluation order of checking part's
  attributes against rules has changed: on each rule in a list, all attributes
  of a part are now checked together; previously all rules were evaluated
  (inner loop) for each attribute (outer loop), which made 'allow' rules
  hardly ever useful;

- ignore dsn_cutoff_level if undefined, instead of treating it as 0;

- fix generating the positive delivery notifications (requested by
  $warn*sender settings): the DSN was missing entirely in case of Courier or
  sendmail-milter setups and the chosen template was not the most appropriate;
  Courier problem reported by Sander Holthaus;

- fix the text in notification templates 'neutral' and 'virus/banned sender'
  to claim non-delivery when the message was truly not delivered, and to
  claim delivery for delivered messages;

- when per-recipient subject tag strings are different for each user, the
  mail forwarding clustering algorithm should have split the forwarded mail
  into separate deliveries but did not do so, resulting in all recipients
  of a multi-recipient mail to get the same string inserted into Subject;
  reported by Michael Goth;

- at last: when mail is received through LMTP protocol, gracefully handle
  a temporary failure 4xx reply from MTA to a RCPT TO command and pass it
  back to a LMTP client for tempfailed recipients only, instead of returning
  450 for _all_ recipients (needed the sending routine to be aware of the
  receiving side capabilities, which was previously not available);

- stricter and more consistent error checking and better error reporting
  on Perl read, sysread and getline operations;

- use O_CREAT|O_EXCL when creating files that are not supposed to pre-exist,
  to be able to detect a potential race condition and other programming errors;

- fixed reporting of virus names for av scanner Perl-SAVI;

- with Sophos Perl-SAVI module, avoid setting 'MaxRecursionDepth'
  if $MAXLEVELS is undefined or zero, matching its semantics to other
  uses in the program; debugging and a fix by Paul B. Henson;

- fixed sloppy regexps when parsing SMTP commands;

- fixed a typo in README.lookups leading to confusion between fields
  spam_subject_tag and spam_subject_tag2, making one to believe there is
  only one such field; pointed out by Max Clark and others;

- handle special case: Postfix hates  ""@domain  but does not mind  @domain;


- new config variable @decoders (with its policy banks counterpart) makes
  it possible to enable/disable decoding of individual file content types
  from the configuration file, and to adjust the external decoders paths and
  search order, all in one place. This list now makes the following config
  variables obsolete: $arc, $gzip, $bzip2, $lzop, $uncompress, $unfreeze, $lha,
  $unarj, $unrar, $zoo, $pax, $cpio, $ar, $rpm2cpio, $cabextract, $ripole;
  although they are still observed for compatibility reasons if the @decoders
  list is left at its default value;

- banned filename/filetype rules can now be specified on a per-recipient basis

  Structure of @banned_filename_maps config variable has changed in
  incompatible way to allow per-recipient banned rules: it is now a
  two level map, similar to the structure of @score_sender_maps.

  Lookup keys used at the first level are recipient addresses, results from
  this lookup can be either a ref to a list of second-level lookup tables, or
  a string which is interpreted as a comma-separated (or whitespace-separated)
  list of _names_ of second-level lookup tables. These names are mapped to
  actual second-level lookup tables through an associative array %banned_rules.
  The indirection by names through %banned_rules is primarily intended for
  SQL and LDAP lookups, which can not return complex data structures and
  Perl code (and should not do so for security reasons), but names may just
  as well be used by static lookups.

  The resulting list of second-level lookup tables (which in most cases is
  a list containing a single element $banned_filename_re, which makes it
  compatible with existing setups) is looked up by using keys such as part
  file name and part file type, exactly like in previous versions.

  The previous default assignment:
    @banned_filename_maps = ( \$banned_filename_re );
  is now incorrect, it can be rewritten (if explicit assignment is desired) as:
    @banned_filename_maps = ( {'.' => [$banned_filename_re]} );
  or (by giving a name to a lookup table):
    @banned_filename_maps = ( {'.' => 'MYRULES'} );
    %banned_rules = ('MYRULES' => $banned_filename_re);
  which is equivalent to the default setting of both variables:
    @banned_filename_maps = ( {'.' => 'DEFAULT'} ); # proper two-level struct.
    @banned_filename_maps = ( 'DEFAULT' );     # same as previous, but shorter
    %banned_rules = ('DEFAULT' => $banned_filename_re);

  The SQL table 'policy' may now contain an optional field 'banned_rulenames',
  which is a comma-separated (or whitespace-separated) list of _names_ of
  second-level lookup tables, with semantics as described above for static

  The configuration variable @banned_filename_maps is a member of policy banks
  as before. The associative array %banned_rules is global and is not a member
  of policy banks.

  The alternative 'banned file' mechanism $banned_namepath_re hasn't changed
  and is still not merged into @banned_filename_maps, which means it can not
  be used when per-recipient banned rules are needed. Perhaps in the
  next version...

- introduce a concept of 'mail_id', which is similar to the am_id as reported
  in the log and elsewhere (e.g. 92598-11-5), but has much stronger long-term
  uniqueness property and can be used for the purpose of uniquely identifying
  a quarantined mail, or for other uses. The mail_id is a 12-character string
  consisting of characters [A-Za-z0-9+-] (like base64, except for a '/' being
  substituted by a '+'), guaranteed to start and to end with an alphanumeric
  character (i.e. not with '+' or '-'). It is derived by cryptographically
  strong method (MD5), cumulatively collecting entropy during the life of
  child processes, folding-in entropy from processed mail and other cheaply
  accessible sources, collected when an opportunity arises (e.g. file system
  file-IDs, SA results etc), without placing a burden on system sources of
  randomness (see RFC 4086). Note that MD5 has been demonstrated to have some
  weaknesses, but we are not talking about cryptographic attacks here, but
  rather about spreading messages which have no inherent intention of causing

  The mail_id carries 71.9 bits of information (subject to the quality of
  sources). For a high-end system that keeps a year's worth of mail messages
  in evidence (e.g. in quarantine) and receives 10e6 messages each day
  (20..200 TB of yearly mail contents), the probability of a mail_id
  collision happening during one year (while gradually displacing an entire
  collection with a new set of IDs) is n^2/m = 0.3 %
    (10e6 * 365)^2 / (62 * (64^(12-2)) * 62) = (10e6 * 365)^2 / 2^71.9 = 0.003
  Eventual clash is still possible and needs to be detected, but a re-tried
  mail delivery attempt is acceptable if its probability is low, as each mail
  processing rolls a dice again. On a smallish system receiving 10.000 mail
  messages daily an 8-character mail_id would suffice, but the savings are
  not worth the trouble of providing configuration flexibility.

  Paired with a mail_id there is its companion secret_id generated for each
  message, such that mail_id can be derived from secret_id and pairing checked,
  but not the other way around. The purpose of secret_id is not yet fully
  developed, but can serve as a 'ticket', granting user a right to release
  a quarantined message addressed to him.

- SQL: can store information about every processed mail to SQL; the information
  is similar to level 0 log entries, but more detailed; a SQL database can be
  used as a basis for searching for a particular mail, for preparing reports
  and to facilitate quarantine management (searching and releasing).
  Enabled by configuring the @storage_sql_dsn list which contains information
  about a SQL server and dataset name, just like the @lookup_sql_dsn does
  for the SQL lookups. If @storage_sql_dsn is the same as the @lookup_sql_dsn,
  a single connection to SQL database will be used, otherwise separate and
  independent connections are established, possibly to different SQL servers.
  Loosely based on ideas from Maia Mailguard by Robert LeBlanc and a patch
  by Brian Wong. Thanks to Brian Wong for testing and valuable feedback.
  See README.lookups for a SQL schema. See new file amavisd-sql-maintain
  (incorporated into README.sql in later versions) for ideas on database
  housekeeping (expiring old entries).

- SQL: can quarantine to a SQL database; selected by setting config variables
  $*_quarantine_method to 'sql:' The @storage_sql_dsn list of dataset names
  is used to choose SQL server and dataset name, and must be nonempty when
  $*_quarantine_method is 'sql:';  When $*_quarantine_method is set to 'sql:'
  the SQL logging must be enabled as well;

- SQL: clean the mess of needing SQL lookup objects to be aware of each other,
  by separating SQL connection information (Amavis::Out::SQL::Connection
  object) from objects holding SQL statement handles. Statement handles
  are invalidated on reconnect, and are dynamically 'prepared' as needed.

- SQL: thanks to a reorganization of SQL objects an automatic reconnect
  to a SQL server is done without temporary failing a processed message;

- SQL: new configuration variable (an associative array) %sql_clause, also
  part of policy maps, allows SQL clauses to be switched with policy banks.
  The components of %sql_clause under keys 'sel_policy' and 'sel_wblist'
  perform the duty of legacy configuration variables $sql_select_policy and
  $sql_select_white_black_list. Compatibility with older configuration files
  is maintained when %sql_clause is left at its default value;

- can add one layer of 62 subdirectories to the quarantine directory for
  more efficient file system use by uniformly distributing quarantined mail;
  enabled by setting a new config variable $quarantine_subdir_levels to 1;

- choosing mail_id as a quarantine file name now greatly reduces the likelihood
  of two quarantined messages processes by the same child process shortly one
  after another from attempting to get quarantined under the same filename,
  leading to a temporary failure ("File...already exists, refuse to overwrite")
  and leaving behind a preserved temporary directory; reported by
  Martin Svensson;

- release from quarantine functionality is now a built-in feature;
  a message release can be requested via enhanced AM.PDP protocol. There is
  a new utility program 'amavisd-release', which currently mostly serves to
  demonstrate how to request releasing of a quarantined file. Currently the
  supported quarantine types are: plain file, gzipped plain file with a name
  ending in .gz, and a SQL-based quarantine. Currently not supported is a
  release from a BSMTP-encoded plain file or from a mbox (Unix-style) mailbox
  quarantine file.

  Example use:
    $ amavisd-release spam/V/V5htXBh0y0Hr.gz H2huh4wfrSyC

  or providing a replacement list of recipients:
    $ amavisd-release spam/V/V5htXBh0y0Hr.gz H2huh4wfrSyC

  The first argument is a (relative) quarantine file name, as reported in
  the log. It must include a 12-character mail_id which is automatically
  recognized. The second argument is a secret_id, which can be fetched from
  a SQL database if @storage_sql_dsn is enabled (see README.sql),
  for example by the command:
    $ mysql amavis -e 'SELECT secret_id FROM msgs WHERE mail_id="V5htXBh0y0Hr"'
  or (preferably) by some other more advanced utility program.

  Current simple-minded heuristics in the amavisd-release program is to assume
  a message is stored in a SQL database when the file name (first argument)
  consists only of a 12-character mail_id. Please adjust the program if this
  assumption is not true, e.g. when $virus_quarantine_method='local:%m'
  instead of a default $virus_quarantine_method='local:virus-%m';

  If secret_id is not available, administrator may choose to skip checking
  of secret_id in the amavisd daemon by setting a configuration variable
  $auth_required_release to false (it defaults to true), and supplying an
  empty secret_id or not supplying it at all in the AM.PDP release request.
  The variable $auth_required_release is also part of policy banks.

  If the release client program specifies a nonempty secret_id in the request,
  the secret_id will be validated and a request will fail if not valid,
  regardless of the setting of $auth_required_release.

  Release requests contend for the same amavisd child processes as regular
  mail processing. This may cause delays in responses to release requests,
  especially when Postfix feeds mail to amavisd-new via LMTP which is more
  persistent in keeping connections open than the Postfix SMTP client service.
  To ensure one child process is always available for processing extra
  requests such as release requests, one may choose to set $max_servers
  larger (by one) than MTA's maxproc setting.

  To enable recognition and processing of AM.PDP protocol requests in amavisd
  daemon, a dedicated Unix socket or a TCP port needs to be opened, for example
  by the following assignments in amavisd.conf:

    $unix_socketname = "$MYHOME/amavisd.sock";
    $interface_policy{'SOCK'} = 'AM.PDP';
    $policy_bank{'AM.PDP'} = { protocol => 'AM.PDP' };

  or similarly for connections through a dedicated TCP port 9998, and
  restricts it to accepting IP connections from local IP address only:

    $inet_socket_port = [10024,9998];
    $interface_policy{'9998'} = 'AM.PDP';
    $policy_bank{'AM.PDP'} = {
      protocol => 'AM.PDP', # Amavis policy delegation protocol
      inet_acl => [qw( )],
    # log_level => 3,

- new configuration variable @spam_quarantine_cutoff_level_maps turns off
  quarantining if spam score is at or above this level; suggested by
  Gary Verchick, MrC and others;

- more informative logging and SMTP status generation in smtp client code;

- IPv6: allow optional brackets around IPv6 address in lookup tables and
  configuration variables, e.g. [FE80::]/10 is treated the same as FE80::/10
  Allow (and ignore) link-local scoped address in extended numeric IPv6 syntax
  (interface specification) when parsing link-level IPv6 addresses,
  e.g. fe80::1%lo0;

- IPv6: adjust the default @mynetworks to include link-local and site-local
  address ranges [FE80::]/10 and [FEC0::]/10, and add (optional) brackets
  around [::1];  (although the use of site-local address ranges seems to be
  deprecated nowadays);

- IPv6: tested sending mail via slightly modified Net::SMTP through
  an IPv6 socket to an IPv6-enabled Postfix; updated README.postfix;

  To experiment, replace the: @ISA = qw(Net::Cmd IO::Socket::INET);
  with: use IO::Socket::INET6; @ISA = qw(Net::Cmd IO::Socket::INET6);
  in Net/, to make use of the: $notify_method='smtp:[::1]:10025';
  Don't forget to add [::1] to mynetworks at MTA smtpd service on port 10025;

- bring earlier the initialization of %local_delivery_aliases and %builtins
  so that the config file can override the defaults;

- add SA autolearn status to the logged spam status (log line: "SPAM, ..."),
  as well as full TESTSCORES info to the list of SA tests (including score
  points for each SA test); contributed by John Sivak;

- new small wrapper module Amavis::IO::Zlib around Compress::Zlib allows for
  reading back compressed quarantine files for a mail release, and allows for
  writing compressed quarantine files without having to fork a gzip process.

  This makes the utility program gzip(1) optional, which may be appreciated
  when running in a chroot environment;

- modified do_gunzip to use the new Amavis::IO::Zlib module;

- added LDAP lookups for the following attributes:
  amavisVirusAdmin, amavisNewVirusAdmin, amavisSpamAdmin,
  amavisBannedAdmin, and amavisBadHeaderAdmin; by Michael Hall;
  added attribute amavisBannedRuleNames by Brian Wong and Michael Hall;

- a policy bank name 'MYUSERS' now has a special semantics: this policy bank
  gets loaded whenever the sender matches @local_domains_maps. This only makes
  sense if local sender addresses can be trusted -- for example by requiring
  authentication before letting users send with their local address;
  the feature requested and a patch provided by Steffen Hansen;

  (a note from future: starting with 2.6.0 an additional requirement for
  loading a policy bank 'MYUSERS' is that 'originating' flag is on, which
  typically means that mail is coming from internal networks or authenticated
  roaming users);

- add cumulative percents to the TIMING report to make it easier to locate
  large contributors to elapsed time;

- updated interface code to Sophos Perl-SAVI module, taking advantage of its
  new ability to reload virus data: amavisd-new will initialize the SAVI
  object in the parent, which will be inherited by the children. The children
  will detect whenever the virus data is stale and automatically exit
  (reducing the number of messages processed with out-of-date protection),
  and the parent will reload the virus data before spawning new children;
  update provided by Paul B. Henson. To have this feature fully functional
  a small patch to Net::Server is needed:
  (patch is no longer necessary since Net::Server 0.88 and amavisd-new 2.3.2)

- provide a routine dump_hash for debugging purposes: given a reference
  to an associative array (a Perl hash) it writes its contents to a log.
  Note: if called within amavisd.conf the log is still directed to STDERR.
  For example:
    @local_domains_maps = ( read_hash("$MYHOME/local_domains") );


- new documentation file README.sql;

- tightened up a sample regular expression in amavisd.conf for catching
  Class ID (CLSID) extensions in file names;

- restrict the 'double extension' banning rule to require at least one letter
  in the next-to-last dot-delimited field; this allows filenames such as
  prog.33.22.01.exe not to be blocked by this rule;

- change hash lookups code to stop a hash search when a matching key
  exists, even if the matching result (corresponding hash value) is undef;
  this reverts the change made in a release 2.1.0 to a previous behaviour which
  is consistent with regexp lookups (but leaves SQL lookups to be different,
  continuing the scan to more general entries on a NULL field value;
  SQL lookup is specific because all attributes live in the same record);
  inconsistency reported by Gary Verchick;

- as a workaround for some versions of unrar (sparc64?) which are unable to
  create a subdirectory parts/rar by themselves, do_unrar now prepares the
  subdirectory explicitly; suggested by Andrzej Kukula;

- provide a new CentralCommand Vexira 'vascan' av entry (distinct from
  the 'Antivirus', which remains H+BEDV AntiVir -based) to work with
  the new Vexira scanner, thanks to Brian Wong and Norman C Rice;

- Vexira vascan: added status codes 3 (password protected) and
  9 (unsupported format), recognize "sequence found"; thanks to Brian Wong;

- F-Prot Antivirus: enhance the pattern to capture virus names;

- renamed AV entry "H+BEDV AntiVir" to "Avira AntiVir", reflecting the
  company name change; thanks to Patrick Ben Koetter;

- change the default $sql_select_policy and $sql_select_white_black_list to
  use the LEFT JOIN ... ON instead of WHERE for the 'join' relation; should
  be functionally equivalent, but makes the join operation explicit;

- changes to LDAP Schema to make it import and play nicely with Novell NDS,
  by Michael Tracey, SONOPRESS USA, LLC April 07 2005
  (changes are included in LDAP.schema but commented out by Mark)

- remove special handling in unmangle_sender() for ancient viruses
  Magistr and Badtrans, leave decisions on suppressing DSN entirely to

- rise a limit on the number of logged matching SA tests from 10 to 50
  (still some arbitrary sanity limit); based on observation of Bojan Zdrnja;

- add a minimal version requirement 2.22 for Digest::MD5, we need
  the new 'clone' method; reported by Thomas Jarosch;

- a command line option 'debug-sa' now sets variable $sa_debug to a string
  '1,all' instead of 1, in anticipation of the next version (3.1) of SA
  which changed interpretation of its debug option; the '1,all' seems to be
  compatible both with the old and the new interpretation, despite producing
  a warning in pre-3.1 versions of SA;

- documentation note: Macintosh.tar.gz installation instructions
  for Macintosh are not recent, they apply to Mac OS X 10.2.0-10.3.9

                                                          December 22, 2004
amavisd-new-2.2.1 release notes


- add support for the pax(1) archive decoder, which can handle tar/cpio/pax
  archives (including legacy format variants). Due to limitations in cpio
  (and in Archive::Tar), for security reasons it is preferred to decode
  such archives with pax and no longer with cpio; please add a line:
    $pax = 'pax';
  to amavisd.conf and verify that the program pax is installed on the system
  (and in the jail if running in chroot); problem reported by Ron Ogle;

- perform additional tests at startup time on proper protection of the
  configuration file;

- add file name extensions wmf, emf and grp to the example list of
  banned extension, according to recent Microsoft security bulletins;
  suggested by Stephane Lentz;

- introduces 'clean but inconclusive' av scanner result to avoid a specialized
  or quick partial av scanner like jpeg checker to claim mail is clean
  when all other general purpose av scanners fail (see below);


- removed some legacy $*_ldap variables, as they are no longer needed;

  These variables were still declared but ignored in 2.2.0 for compatibility
  with older amavisd.conf files. Such variables need to be removed from
  the amavisd.conf if they are still present there from older versions,
  otherwise Perl will complain with 'Global symbol ... requires explicit
  package name";


- files_to_scan and decompose_mail are now able to remove unexpected
  directories which may have been left behind by some failed decoding
  and were causing temporary failures and mail delivery retries;
  error recovery problem after failed unarj reported by Ralf Hildebrandt,
  and a related problem with tar, reported by Ron Ogle;

- error recovery code in files_to_scan and rmdir_recursively now tries to
  change protection on directories and files, and retry if the first attempt
  to access them fails because of denied permission;

- pre-load some additional Perl modules needed by SA when running in chroot;

- add module Net::LDAP::Search to a list of pre-fetched modules;
  omission pointed out by Paul Jacobson;

- when quarantining is disabled by keeping $QUARANTINEDIR undefined,
  the log entry and administrator notification message inappropriately
  suggested that mail was quarantined, which in fact (appropriately)
  it was not. Setting $QUARANTINEDIR='' did work as expected.
  Reported by Sascha Lucas;

- avoid the use of Encode::is_utf8 due to a Perl bug (still present in
  5.8.8, Encode::is_utf8 on tainted utf8 character string produces false);
  Perl bug tracking: #32687: Encode::is_utf8 on tainted UTF8 string
  returns false;

- modify safe_encode() to guarantee the result is a string of octets,
  not a string of UTF-8 characters; it saves some unnecessary work in
  further processing and keeps MIME::Entity from UTF swamp when running
  in chroot; problem pointed out by Branko F. Gracnar;

- avoid braindead Perl default where an empty regexp implies the last
  successfully matched regexp, which (if not being very careful) brings in
  some completely unrelated last-executed regular expression;

- change kill 'TERM' into kill 'KILL' when a forked process within run_command
  and run_command_consumer gets into deep trouble, to avoid exit handlers
  being invoked in the subprocess (which could lead to two processes trying
  to clean the same set of temporary files);

- in an old sendmail setup using the amavis(.c) helper program without
  LDA arguments, avoid inappropriate warning:
    "WARN: no recips left (forgot to set $forward_method=undef using milter?)
  and return status 0 instead of 99 when message is to be blocked, as the
  helper program amavis(.c) does not recognize status 99 in this situation
  and inappropriately passed it on to sendmail; reported by The Mindflayer;

- the @bypass_header_checks_maps is now able to also bypass the bad header
  checks as provided by MIME::Parser; inconsitency reported by CRivera;

- avoid some Perl warning messages; thanks to Bill Landry;


- add configuration variable @newvirus_admin_maps (and $newvirus_admin,
  along with corresponding SQL field 'newvirus_admin') which works like
  the existing @virus_admin_maps (and $virus_admin), except that it sends
  virus administrator notification to specified e-mail address only for newly
  encountered viruses which have not yet been encountered since the amavisd
  startup. It makes use of by-virusname counters in the SNMP counters
  database. If more than one child process starts working on infected
  message containing a not-yet-accounted-for virus, there might be more
  than one 'first time' notification, this is not a malfunction. Both
  the @newvirus_admin_maps and the @virus_admin_maps may be enabled,
  each (possibly both) would receive their notifications as appropriate.

  A useful setting is to globally enable only the new virus notifications,
  and additionally enable _all_ administrator notifications for internally
  originating mail only (by the use of policy banks);

- provide separate configuration variables @banned_admin_maps and
  @bad_header_admin_maps, along with corresponding SQL fields
  'banned_admin' and 'bad_header_admin'; their function was previously
  covered by @virus_admin_maps, which now only still controls administrator
  notifications in case of viruses;

- introduces 'clean but inconclusive' av scanner result to avoid a specialized
  or quick partial av scanner like jpeg checker to claim mail is clean
  when all other general purpose av scanners fail:

  in av scanner entries (lists @av_scanners and @av_scanners_backup) give
  an extended meaning to undefined fourth argument (the 'match for clean'
  list or regexp). The interpretation of the fourth argument is now:

  4. an array ref of av scanner exit status values, or a regexp (to be
     matched against scanner output), indicating NO VIRUSES found;
     a special case is a value undef, which does not claim file to be clean
     (i.e. it never matches, similar to []), but suppresses a failure warning;
     to be used when the result is inconclusive (useful for specialized and
     quick partial scanners such as jpeg checker);

  Also modified example jpeg checker entry in amavisd.conf accordingly.

- NOD32 av scanner: changed @av_scanners entry to match the new version
  of the scanner; thanks to Nejc Skoberne;

- added @av_scanners entry for File::Scan;

- when preparing a SQL SELECT clause for white/blacklisting lookup,
  take into account a relative position of ? and %k in the
  $sql_select_white_black_list template to improve flexibility
  of specifying the clause; suggested by Matt Petteys;

- reduce the log level of some more common and harmless log messages;

- macro %p and the log entry now reports full policy bank path,
  not just the last loaded policy bank name;

- added LDAP attributes amavisWarnVirusRecip, amavisWarnBannedRecip,
  and amavisWarnBadHeaderRecip; by Joel Nimety and Michael Hall;

- renamed LDAP attribute name amavisSpamModifiesSubject to
  amavisSpamModifiesSubj in order to match the documented LDAP schema;
  noticed by Kees Bos, patch by Michael Hall;

- add support for ripOLE decoder, which attempts to extract embedded documents
  from MS OLE documents (MS Office) (,
  by Paul L Daniels)); ripOLE is still experimental/alpha code;
  To make amavisd-new find the installed program 'ripole', add the:
    $ripole = 'ripole';
  to the amavisd.conf. Suggested by David Wilson and Noel Jones;

- allow multiple occurrences of command line option:  -c config_file
  and execute the provided configuration files one after the other;
  based on a subset of functionality provided as a patch by Davor Ocelic;

- a slight improvement in classifying mpeg and some other multimedia files
  (in the default $map_full_type_to_short_type_re);

- several minor code cleanups;

- add a recommendation by Daniel J McDonald to a documentation file INSTALL:
    If different UID is preferred for an AV scanner, a solution for
    ClamAV is to add user clamav to the amavis group, and then add
    AllowSupplementaryGroups to clamd.conf;

- enclosed a simple demonstrational Perl program, which is
  functionally much like the amavis.c helper program, but talks the new
  AM.PDP protocol with the amavisd daemon. See README.protocol for the
  description of AM.PDP protocol. To be placed in amavisd.conf:
    $protocol='AM.PDP';  $unix_socketname='/var/amavis/amavisd.sock';
  Usage: sender recip1 recip2 ...  < message.txt

- documentation updates;

                                                           November 2, 2004
amavisd-new-2.2.0 release notes


- modified MIME entity traversal to include MIME container parts
  (e.g. multipart/*) as pseudo parts. Such pseudo-parts do not carry
  any body contents but do show up (with their MIME content type only)
  in the tree structure as seen by banned names checks. This makes it
  possible to specify more complex banned rules based on the placement
  of leaf nodes within or outside of mime multipart containers.

  This also re-enables the possibility to check such MIME wrappers for
  banned MIME Content-Types (most notably for the message/partial and
  message/external-body), which was lost with the change of internal
  representation of mail parts in version 2.0 (amavisd-new-20040701);

- preserve original zip archive for virus scanners to see, if the archive
  contains any zero-length members (Archive::Zip module chooses not to
  extract members with declared zero size, even if the size does not
  correspond to the actual size); based on a patch by Dirk Datzert;

- add tests to mime_traverse to verify that the files produced by MIME::Parser
  really do exist and are readable; (and sent a patch adding I/O checks
  for MIME::Tools to its maintainer; please use MIME::Tools 5.414 or later
  from CPAN to avoid possibility of full /tmp partition causing infected
  or spam mail to pass through);

- changed recommendation in INSTALL to choose the latest version of
  MIME-Tools from CPAN - the 5.415 at the time of this writing;

- do_unrar: recognize encrypted entire archives (not just their individual
  members), and flag mail as undecipherable;

- recognize file(1) report/^MS Windows\b.*\bDLL\b/ as short type 'dll';
  add 'dll' to example patterns in amavisd.conf and amavisd.conf-sample;
  add name extension '.cpl' to the list of basic banned names;


- incompatible change: the default value for $recipient_delimiter is
  now undef and no longer '+'; adding address extensions must now be
  explicitly enabled;

- minor change in the default X-Virus-Scanned: header field, see below;


- rewritten LDAP modules, by Michael Hall;

- improvements in handling of double errors (errors that occur while
  handling consequences of some previous error); be permissive on failures
  in DESTROY methods; prevent some cases for child process not being able
  to sign off from the nanny database;

- enforce $max_requests more strictly, dropping SMTP session after task
  count is exceeded by one; this is in response to the new smtp session
  caching behaviour in Postfix, which is now much more persistent in keeping
  the session open on a busy mailer; although dropping session at the server
  side is discouraged by RFC 2821, this change was recommended on the
  postfix-users mailing list;

- added a site-wide mail size limit $smtpd_message_size_limit, and a
  per-recipient mail size limit lookups @message_size_limit_maps along with
  SQL and LDAP fields, making it possible to reject mail based on its size.
  The list of lookup tables maps recipients to mail size in bytes, e.g.:

    @message_size_limit_maps = ({'' => 20_000_000,
                                 '' => 15*1024*1024,
                                 '' => 0, # uses global limit
                                 ''      => 10*1024*1024 });
    $smtpd_message_size_limit = 20*1024*1024;  # global limit if nonzero

  A value of 0 or undef disables the check and is a default. A per-recipient
  limit is bound on the high side by the $smtpd_message_size_limit, and on the
  low side by 64kB, which is a minimal allowed size limit imposed by RFC 2821.

  This limiting really belongs to MTA and is only partially supported here
  (no admin notifications, no quarantine, no final_*_destiny configurability).
  It is mostly provided here to be able to place some sanity limit on runaway
  or malicious clients, or if someone insists on using amavisd-new in a
  pre-queue filtering setup; suggested by Tuomo Soini;

- internal: add new object $msginfo->mail_text_fn to hold the file name
  of the original mail, decouple the location of file email.txt from the
  temporary directory, which was implied until now. This is presently needed
  for the Courier interface. Add optional attribute 'mail_file' to the
  AM.PDP protocol, see README.protocol;

- in banned parts descriptor strings which are used in banned name checking,
  provide a 'T=empty' short type for empty mail parts, including for empty
  MIME parts (instead of omitting T=... altogether). This can be used in
  banned rules to test for empty parts, generally or restricted to empty
  MIME parts; suggested by Ricardo Stella and Stephane Lentz;

- a banned lookup result (which is interpreted as boolean for most purposes:
  zero or empty for false, anything else for true) may give a result 'DISCARD'
  (which is true as well), which will disable DSN if it turns out the mail
  was blocked by such banned rule (akin to viruses_that_fake_sender and
  spam_dsn_cutoff_level). Here is an example rule in $banned_namepath_re:
         ^ (.*\t)? M=application/octet-stream \t(.*\t)* T=empty (\t.*)? $'xmi
      => 'DISCARD' ],
  Suggested by Ricardo Stella and Stephane Lentz;

- fix 'Insecure dependency in unlink' in sub files_to_scan which could
  happen when some decoder left non-regular files in the directory;
  reported by Andrzej Kukula;

- bug fix: only insert LDAP and SQL lookups objects into lists of maps at the
  first task of a child process, later tasks should not insert duplicates;

- new subroutine do_ar and new configuration variable $ar to handle
  standard Unix archives, including Debian binary packages;

- recognize a Unix archive and give it a short type .a;
- recognize a Unix relocatable binary and give it a short type .o;

- do not penalize SMTP status "554 5.1.1 Error (DATA): no valid recipients",
  the situation arises regularly when pipelining is in effect and all
  recipients happen to be rejected;

- protect spaces and newlines when logging broken Message-ID and
  Resent-Message-ID header fields (macros %m and %r) to facilitate log
  parsing; protect newlines in logged Subject header field (macro %j);
  parsing difficulty pointed out by Chris Lee;

  The present solution uses =XX (hex) encoding and is a quick and dirty
  fix. A cleaner solution to avoid double sanitation of special characters
  is needed, but would involve a deeper reorganization;

- updated example list of "banned extensions - long" in amavisd.conf
  and amavisd.conf-sample;

- change the default lock file name from "$MYHOME/amavisd-$$.lock" to
  "$MYHOME/amavisd.lock", to avoid inventing a new name at every restart
  and leaving old files behind; pointed out by Dale Walsh and Martin Orr;

- updated av entry for nvcc (Norman Virus Control v5 / Linux) to include
  statuses 10,11, and 2,14 to the status lists according to documentation;
  password protected or corrupted archive (status 11) was not recognized
  as non-infected status; thanks to Michael Ramke of Norman Data Defense
  Systems GmbH;

- updated DrWebD entry to recognize and ignore flag DERR_SKIPPED;

- support Mail::ClamAV 0.12 and 0.13 or later, which is incompatible with
  0.11 due to change of constant names in the underlying ClamAV library;

- added 'check-jpeg' example entry to the @av_scanners list and provide the
  associated module; it offers a fully-fledged check for jpeg
  comment field buffer overflow attempts; should serve mainly as an example
  for adding similar quick responses to new threats;

- added 'check-jpeg-simple' example entry to the @av_scanners list (only in
  amavisd.conf-sample); it offers a quick check (and not very exact one:
  checks only the first 32kB) for jpeg comment field buffer overflow attempts;
  should serve mainly as an example for adding similar quick responses
  to new threats;

- relax too restrictive sanity check on temporary directory name
  when accepting requests from a helper program or via AM.PDP protocol
  (e.g. with sendmail milter setup); reported by Babu Kanagala;

- relax allowed set of characters when receiving XFORWARD attribute values,
  it turns out that characters like '=' and '+' are allowed;

- when using "bsmtp:" delivery method suppress X-Envelope-From and
  X-Envelope-To header fields, as the addresses are already available
  in the envelope;

- when using the "bsmtp:" quarantine method the *_quarantine_to was
  completely ignored, which made it impossible to turn off quarantining
  selectively for certain users by specifying an empty or undef value.
  Since 2.2.0 an empty *_quarantine_to turns off quarantine for a recipient
  regardless of the quarantine method. A nonempty string in *_quarantine_to
  (the exact value is still ignored) must now be used even with "bsmtp:"
  to enable quarantining. Inconsistency discovered by Sean Doherty;

- suppress leading $QUARANTINEDIR string from the value of macro %q,
  thus hiding the absolute file path from notifications;

- add configuration variable $local_client_bind_address (and equivalent
  policy bank key), to allow for explicitly binding local socket address
  to a specific interface in SMTP client; suggested by Wouter de Jong;

- keep whatever (if any) file results from gunzip and family (do_uncompress)
  even if the decompressor's exit status is nonzero; reason: gzip returns
  status 2 when decompressing file with trailing garbage; reported by
  Tobias Reckhard;

- collect declared original file name from gzipped (and friends) files
  if reported by file(1), making them available to banned name checking;

- avoid unnecessarily checking white/black lists if spam checks will not
  be used (e.g. infected mail);

- use qquote_rfc2821_local to properly quote e-mail addresses in the
  most visible log entries;

- if there is more than a single (or less then one) occurrence of %k in the
  SQL SELECT clause template, multiplicate the set of query keys accordingly,
  making possible more complicated custom SELECT clauses;

- don't forget to load amavis policy delegation protocol support code
  if AM.PDP is explicitly requested in $protocol, even if not listening
  on a Unix socket;

- add 'queue_id' attribute to the AM.PDP protocol; equivalent to a change
  in the Courier support code by Martin Orr;

- include the declared (faked) sender address in the virus recipient
  notification template, in addition to the originator IP address;

- add macro %Q and method Amavis::In::Message::queue_id, holding a
  MTA queue ID if available (in Courier and milter/AM.PDP setup);
  by Michael Musikhin (through Martin Orr);

- add macro %y to show elapsed processing time; suggested by Ed Walker;

- sanitize newlines and spaces (and some other characters) when moving
  syntactically invalid Message-ID and Resent-Message-ID to macros %m and %r
  for logging purposes; resulting wrapped main log entry reported by CRivera;

- bring up syslog priority to LOG_NOTICE when debug or debug_oneshot
  is in effect;

- make a product name, version ID and version date available as separate
  variables to avoid the need to parse $myversion for the purpose of
  customizing e.g. the setting of $X_HEADER_LINE;  based on suggestion
  by Dale Walsh; the re-introduction of a date ($myversion_date) also
  suggested by Stephane Lentz.

  Added variables: $myproduct_name, $myversion_id, $myversion_id_numeric,
  and $myversion_date, which serve to assemble the $myversion. Modified the
  default templates of $smtpd_greeting_banner and $smtpd_quit_banner to
  take advantage of the new variables. Changed $X_HEADER_LINE default
  from "by amavisd-new at $mydomain" to "$myproduct_name at $mydomain" and
  added an example of a $X_HEADER_LINE setting with version number included
  to the amavisd.conf-sample;

- added SQL fields 'virus_admin' and 'spam_admin' to lookup lists
  @virus_admin_maps and @spam_admin_maps; patch by Robert Collier;

- add a log message 'SPAM-KILL, ...' (at log level 3) for not-passed mail,
  to complement the 'SPAM-TAG' log message for passed mail;

- add Mail::SpamAssassin::Plugin::SPF to a list of modules that SA fails
  to load at init time;

- prevent sending the same SMTP response more than once, if the first
  attempt failed due to disconnected SMTP session;

- fix a double-@ formatting buglet in the log message
  "adding address extension _spam to user@@domain", reported by Vincenzo;

- add kill('TERM',$$) to the arsenal of attempts to get rid of unwanted
  forked process;

- includes rpm spec file, including the init script, contributed by
  Marius Andreiana, based on previous work by Dag Wieers;

- document the localization template directory contents
  (in file amavisd.conf-sample) when read_l10n_templates is used;
  thanks to Joël Brogniart;

- includes file Macintosh.tar.gz, which contains auto-startup scripts
  and installation instructions for Mac OS X, contributed by Dale Walsh
  of the Dale Enterprise L.L.C.

                                                          September 6, 2004
amavisd-new-2.1.2 release notes

- fixed (hard)black- and white-listing on static lookup tables
  which failed to match any sender; reported by Derck Floor;

- use $hdrfrom_notify_recip address in the From: field for recipient
  notifications, instead of $hdrfrom_notify_admin; inconsistency
  pointed out by Ekkehard Burkon;

- the 'neutral' sender notification template was joining the Subject
  and the Message-ID header fields into one longer Subject when it was
  reporting a nondelivery other than the 'invalid characters in header'.
  Likewise the first body line of this same DSN was eaten up:
    "This nondelivery report was generated by the amavisd-new program"
  (the problem was introduced in amavisd-new-20030616 and never reported);

- in amavisd-agent, amavisd-nanny, amavisd: extend the signal and error
  handling in code sections holding bdb locks from just ignoring the SIGINT,
  to controlled catching and re-signaling several signals and error conditions;
  problem reported by Tom Mulder;

- suppress duplicate names from the list of virus names in macro %V;
  by Gregor Weiss;

- fine-adjusted log level of some log messages;

- discard leading and trailing whitespace from the macro %t (Received trace);

- extend the search for IP in the Received trace from 4 to a maximum of
  6 entries;

- ignore private IPv6 addresses (RFC 3513: link-local, site-local, multicast)
  when searching through Received trace for the origin of mail;

- place mail header field X-Envelope-From in front of the X-Envelope-To in
  quarantined mail; also changed case of X-Quarantine-id into X-Quarantine-Id
  for consistency with other header fields;

- provide new macro %e which evaluates to our best guess of the originator
  IP address collected from the Received trace, complementing similar macros
  %t, %a and %g; suggested by Gregor Weiss;

- add the result of macro %e to the default 0-level log entry;

- provide new macros %u and %U to evaluate to a timestamp of the message
  reception similar to an existing macro %d (RFC 2822 local date-time);
  the (%u) as Unix time (seconds since 1970-01-01T00:00Z as a decimal integer,
  suggested by Gregor Weiss), and (%U) as ISO 8601 (EN 28601) UTC date-time;

- avoid some empty lines in default DSN templates and fix some inconsistencies
  in their formatting;

- internal: collect existing common code for time formatting as new
  subroutines iso8601_timestamp and iso8601_utc_timestamp; collect existing
  common code to find IP address in the Received trace as a new subroutine

- bump up the version number in $myversion - the 2.1.1 still presented
  itself as 2.1.0;

- add a note about a data structure difference between @score_sender_maps and
  $per_recip_blacklist_sender_lookup_tables (amavisd.conf-sample, amavisd)

                                                            August 24, 2004
amavisd-new-2.1.1 (amavisd-new-20040824) release notes

- unconditionally initialize @banned_filename_maps to (\$banned_filename_re),
  otherwise $banned_filename_re is ignored by default (unless amavisd.conf
  explicitly assigns to @banned_filename_maps); a patch by Thomas Jarosch;

- fixed inappropriate log entry in SQL whitelisting:
     wbl: (SQL) recip <> whitelisted sender <...>,
     unexpected wb field value: "1";
  reported by Carlos Horowicz;

- added missing import of &ca to the amavisd-new-courier.patch; by Martin Orr;

- produce a warning when there is an unknown field in the policy bank
  to be loaded;

- with delivery method 'bsmtp:' prepend a directory $QUARANTINEDIR to the
  file path if not explicitly specified, to behave like the 'local:' delivery
  method, making it possible to hide full path from the X-Quarantine-Id
  and notifications; a patch by Thomas Jarosch;

- pre-load SA 3.0.0 module Mail::SpamAssassin::Plugin::Hashcash
  to make it available in the chroot jail;

- pre-load modules Mail::SpamAssassin::SpamCopURI and URI::*
  for SA older than 3.0.0;

- enhancement to amavisd-nanny: when terminating a process and SIGTERM
  produces no result for some time, try SIGKILL; contributed by Philip Engdahl;

                                                            August 15, 2004
amavisd-new-2.1.0 (amavisd-new-20040815) release notes

The 2.1.0 release is mostly a maintenance release over 2.0, with
only a handful of smaller features added. Based on a manual code audit
the number of smaller internal code changes is rather extensive,
some changes dealing with long-standing known deficiencies, minor bugs,
documentation problems and typos. Only a few fixes are for new bugs
introduced in 2.0.

The files amavisd.conf, amavisd.conf-default and amavisd.conf-sample
have been extensively reworked, with the hope to suit better the
new installations, while possibly causing some head-scratching for
existing installations when looking at a diff. The file amavisd.conf
is the one that should serve as a sound base for the initial config file,
while keeping an eye on the list of all variables and their defaults
in amavisd.conf-default. The amavisd.conf-sample is being phased out
of active maintenance, and should serve mostly as a set of examples and
the source of documentation until better documentation is available.

Two nice features are available:

- the use of BerkeleyDB and libdb is now optional;
  controlled by variables $enable_db and $enable_global_cache;

- a new program 'amavisd-nanny', with the accompanying instrumentation
  in amavisd, displays the general health of all amavisd child processes,
  reports crashed ones and attempts to kill long overdue processes;
  It is still experimental and minimalistic, problem reporting is currently
  only to stdout.

Other changes:

- incompatible change since 2.0: the use of BerkeleyDB is now off by default;

  The use of BerkeleyDB and libdb is made optional, controlled by
  variables $enable_db and $enable_global_cache, both false by default.

    enables the use of BerkeleyDB/libdb
    (for SNMP counters database and nanny, and optionally for cache);

    enables the use of libdb-based cache when $enable_db is also true;

    If either the $enable_global_cache or $enable_db are false,
    cache of mail body MD5 digests is kept in child-local memory
    as in pre-2.0 versions, and is therefore local and short-lived,
    with lower expected cache hit rate;

- incompatible change: DSPAM 3.0 or better is needed (if $dspam is enabled),
  no longer works with 2.x;

- incompatible change: changed name of the (hardly ever needed) configuration
  variable auth_supported_out to auth_required_out, to better reflect its
  semantics (should be true if MTA server to which amavisd is sending
  notifications and forwarding mail requires authentication (AUTH smtp

- a new small program included: amavisd-nanny is a program to show the status
  and keep an eye on the health of amavisd child processes (experimental);

- fixed a bug in lookup_acl where a "." did not act as a catchall;
  thanks to JP;

- fixed a problem in SQL lookup which could return undef even when
  not all the matched records had NULL in the field;

- fixed compatibility with old 'amavis' helper program
  ('delivery_care_of' defaulted to 'client', instead of depending on
  the presence of ldaargs), reported by Charlie Schluting and Christer;

- fixed long standing problem in do_ascii, which could return without calling
  Convert::UUlib::CleanUp, occasionally spilling state into subsequent
  mail checks within the same process;

- fixed macros %D, %O and %N when log_recip_templ is being expanded;
  a patch by Ed Walker;

- fixed recognition of separators in a nested call during macro expansion;

- pre-load missing modules Net::LDAP, Net::LDAP::Schema, Net::LDAP::Search,
  and Net::LDAP::RootDSE; suggested by Paul Jacobson;

- fix locale-related bug in rfc2822 date generation, where we were restoring
  the saved LC_TIME value to LC_CTYPE (!);
  a patch by Henrique de Moraes Holschuh / Debian support team;

- protect from signals while bdb cursor holds a lock;

- new subroutine inherit_header_edits() and slight code restructuring
  makes possible for spam_scan() and other code before the final delivery
  to start submitting common header edits into $msginfo->header_edits,
  avoiding the need for passing them through global variables;

- now a loglevel-based automatic syslog priority assignment can no longer
  lower a message syslog priority below the syslog priority specified
  in the $SYSLOG_LEVEL, it can only increase it; the violation of the
  least-surprise principle pointed out by Andy Dills;

- a small optimization in logging: a new subroutine ll allows to save time
  in preparing complex log entries when we know their log level exceeds
  the current log level and won't be logged;

- in default macro templates $log_templ and $log_recip_templ:
  * placed 'spam' condition before 'bad header' for consistency
    with program behavior;
  * added reporting of tag/tag2/kill levels in $log_recip_templ
    (experimental: macro names may change in future versions);
  * rewritten templates using negation (i.e. [?x|1] ) to avoid unsightly
    selector nesting;

- MIME decoding is now allowed to exceed the decoding quota, avoiding
  the problem when a small quota settings might not allow even a plain
  mail through;

- override LC_TIME to "C" on every log message, to work around issues
  with Unix::Syslog, which would log stuff with the date stamp localized,
  which syslog would dislike and add its own, and the resulting mess
  is not recognized by amavis log processors;
  a patch by Henrique de Moraes Holschuh / Debian support;

- changed dspam command line options to work with dspam 3.0 (no longer
  with 2.x), a patch by Ron Ohmer, Nabil Sefrioui, and Reech;

- dspam header fields are now inserted into passed mail if all recipients
  are local;

- supply better defaults for $hdrfrom_notify_sender, $hdrfrom_notify_recip,
  $hdrfrom_notify_admin and $hdrfrom_notify_spamadmin, similar to defaults
  from amavisd-new-20030616-p10;

- when parsing output from the 'AVG grisoft' virus scanner,
  don't include CR in virus name; reported by Vernon A. Fort;

- new file 'amavisd-new-qmqpqq.patch' provided by Martin Solciansky,
  (similar to fixes by Christopher Odenbach) making it work again
  with the current version of amavisd-new;

- use lstat instead of stat, and test for soft links wherever appropriate;

- remember inode and device number when creating temporary directory
  and temporary file, and test for change before removing them;

- enhanced security: certain tainted values are allowed to enter deeper
  into program, untainting them only where and when really necessary;

- avoid a taint problem in Mail::ClamAV;

- added AV entry for CAI eTrust Antivirus; by Stephane Lentz
  (requires a suid shell wrapper around inocmd32);

- added status 9 to the set of infected statuses for the
  drweb command line scanner (DrWeb Antivirus);

- use our subroutine q_encode to Q-encode header fields from the
  notification templates, instead of MIME::Words::encode_mimeword
  (a similar fix in 2.0 applied only to encoding of modified headers
  in passing mail);

- add attribute 'x-spam-type=original' to the Content-Type header field
  (the SpamAssassin's code to recognize an original email) when defanging
  spam, facilitating reporting spam via SA to other spam fighting tools;
  a patch by Brian May;

- add a note to amavisd.conf that $sa_auto_whitelist has no effect on SA
  since 3.0.0 - SA has now a configuration file option 'use_auto_whitelist';

- turn off timer in post_process_request_hook() to avoid periodically
  recreating child processes on an idle machine;

- added @mynetworks_maps and enhanced lookup_ip_acl() to take a list
  of lookup tables: currently members can be an array ref (as before),
  or a hash ref (new) or a plain constant (new);

- generalized @debug_sender_acl into @debug_sender_maps along the lines of
  other lookup tables, and make it part of policy banks;

- added @warnvirusrecip_maps, @warnbannedrecip_maps, @warnbadhrecip_maps;

- added @spam_subject_tag_maps and @spam_subject_tag2_maps,
  to allow per-recipient spam tags string; suggested by Ed Walker;

  Note, there is an inconsistency in names of legacy variables and the
  new @*_maps, in an attempt to rectify an unfortunate choice of name
  for seldomly used variable $sa_spam_subject_tag1:
    @spam_subject_tag_maps  = (\$sa_spam_subject_tag1);  # exotic
    @spam_subject_tag2_maps = (\$sa_spam_subject_tag);   # in common use

  Note also that corresponding SQL fields are 'spam_subject_tag' and
  'spam_subject_tag2'; usually only the 'spam_subject_tag2' would be used;

- added configuration variable $auth_reauthenticate_forwarded, which
  directs amavisd to apply its own credentials ($amavis_auth_user and
  $amavis_auth_pass) to unauthenticated forwarded (passed) mail, besides
  using them for submitting original messages (notifications and quarantine).
  This is similar to how mailing list managers are allowed to work (rfc2554).
  Note that the Perl module Net::SMTP in its current version is unable to
  specify the 'submitter' in its 'MAIL FROM' command, this should be rectified;

                                                               July 1, 2004
amavisd-new-20040701 / amavisd-new-2.0 release notes

MAJOR NEW FEATURES (since amavisd-new-20030616-p10):

- security improvements: no shell required in chroot jail, checks performed
  to see if dropping privileges was successful, can drop privileges before
  reading config file;
- the 'amavisd reload' command reimplemented, it now works even
  if running chroot-ed;
- new feature: policy banks hold sets of configuration variables that may
  be switched with another predefined set based on incoming port number or
  original SMTP client IP address, avoiding the need to run more than one
  instance of amavisd daemon;
- new feature: @score_sender_maps is a soft variant of black- and whitelisting;
- extended semantics of SQL field wblist.wb for soft black/white-listing;
- redesigned mail structure representation allows better control
  over 'banned' names and types;
- MIME defanging can wrap the entire original mail in a MIME container;
- more flexible control on lookups: configurable list of lookup tables
  observes the specified order of tables and permits arbitrary number
  of tables of any available type;
- level-0 logging either by-message or by-recipient;
- syslog priorities are now dynamically derived from the log level;
- constantly updates a small database of SNMP-style counters, providing
  real-time measurements for status monitoring and statistics reporting,
  reducing the need for analyzing a log file; includes a sample/demo
  program 'amavisd-agent';
- new 'policy delegation protocol' between helper program and the daemon
  can pass more information to the daemon and allows the daemon to
  pass more instructions back to MTA (useful for sendmail milter setup);
- persistent cache of recent virus and spam checks, common to all child
  processes, improves the cache hit rate;
- support for IPv6 address formats;
- provisional/experimental support for DSPAM spam checker;
- support for ClamAV virus scanner via Perl module Mail::ClamAV;
- cleaned amavisd.conf :
    amavisd.conf         configuration file with the essentials
    amavisd.conf-default lists all configuration variables with their defaults
    amavisd.conf-sample  traditional-style commented amavisd.conf with examples


INCOMPATIBLE CHANGES since amavisd-new-20030616 (any patch level)

- requires Perl module BerkeleyDB with libdb version 3.1 or later
  (tested with db 4.1); This requirement will be made optional
  at a later date (in amavisd-new-2.1.0);

- a directory at $db_home (default /var/amavis/db) must be manually created
  to store cache and snmp DB files. It should be writable by user running
  amavisd. The db files within are removed and re-created at each restart
  to avoid having to deal with db recovery (but need not be, as far as the
  program logic is concerned);

- sending signal HUP in order to restart amavisd no longer works (previously
  it only worked in non-chrooted environment and relied on guessing amavisd
  absolute path); please use 'amavisd reload', or 'amavisd stop' and restart;

  If the HUP method is really still needed, please replace the line
    commandline => [],  # disable
    commandline => ['/usr/local/sbin/amavisd','-c',$config_file],
  in file amavisd, adjusting the path if necessary.

- due to changed names of temporary files, the old 'DrWebD' av entry needs
  to be replaced with the current one;

- changed $final_virus_destiny default from D_BOUNCE to D_DISCARD
- changed $final_spam_destiny  default from D_REJECT to D_BOUNCE

- changed defaults for variables $virus_quarantine_to, $banned_quarantine_to,
  $bad_header_quarantine_to and $spam_quarantine_to from undef (no quarantine)
  to values 'virus-quarantine', 'banned-quarantine', 'bad-header-quarantine'
  and 'spam-quarantine' respectively.  Set them to undef or '' (empty string)
  to disable quarantine;

- add address extension at tag2 level, not at kill level as before;
  suggested by Jacob Elder and others;

- because of the reorganization of lookup tables, a new way
  of quickly disabling virus or spam checks in amavisd.conf is used.
  Instead of:
    # @bypass_virus_checks_acl= qw( . ); # uncomment to DISABLE anti-virus code
    # @bypass_spam_checks_acl = qw( . ); # uncomment to DISABLE anti-spam code
  the new recipe is:
    # @bypass_virus_checks_maps = (1);   # uncomment to DISABLE anti-virus code
    # @bypass_spam_checks_maps  = (1);   # uncomment to DISABLE anti-spam code

- @virus_admin_maps and @spam_admin_maps now take as lookup keys recipient
  addresses, not sender address as before. The new semantics was often
  requested, the old semantics was not useful because modern viruses and
  spam fake sender address, so the choice was made to incompatibly change
  semantics and use the same config variable names, rather than introduce
  new names and leave behind useless variables;

  An implication is that with multiple-recipient mail it is now possible
  to have more than one admin notification generated when recipients have
  different admin addresses associated. Still, each distinct admin address
  receives only one admin notification;

- removed old compatibility variable $mailto. Use $virus_admin and $spam_admin
  variables instead, as suggested in amavisd-new-20021116 release notes,
  or the more recent inventions @virus_admin_maps and @spam_admin_maps,
  or their equivalents in policy banks;

- removed support for old configuration variable $mailfrom. Use variables
  $mailfrom_notify_admin, $mailfrom_notify_sender, $mailfrom_notify_recip,
  and $mailfrom_notify_spamadmin as introduced in amavisd-new-20020630
  for the purpose;

- removed ancient variable @local_domains, use @local_domains_acl instead,
  or the more general @local_domains_maps

- removed old compatibility code which allowed for a couple of traditional
  variables to treat value "no" as false. The use of "no" for false has been
  deprecated since amavisd-new-20021116.

- revoke an old compatibility measure where a missing (undef) tag2 level
  would fall back to the kill level value;

- lookup_hash incompatible change (but hardly anyone will notice):
  a key presence in the Perl hash but with undefined hash value
  used to be interpreted as true, but now it is treated as undef,
  causing search to continue with remaining lookup tables (if any).
  This is more general and more in spirit with other lookup mechanisms;

- changed a default to initialize SAVI-Perl every time a child process is
  started, no longer at master process startup time only. This is to avoid
  the need to restart amavisd every time the Sophos IDE database is changed.
  One can revert to the previous behavior by uncommenting a call to
  Amavis::AV::sophos_savi_init in subroutine fetch_modules_extra;


- no longer invoke shell to call gzip for compressing quarantined files
  or to call the sendmail command for submitting messages; the most important
  consequence is that a shell is no longer needed in a chroot environment
  and should preferably be removed;

- not to forget what was introduced in -p10: inserted a security check
  for a missing Net::Server patch, and abort if vulnerable;

- new command line options '-u user' and '-g group' are available.
  These are pretty much equivalent to doing a su(1) to the specified user
  first (in which case the use of these options is redundant).

  By doing 'su' or by specifying a command-line option '-u username'
  one can prevent a potential security risk on misconfigured sites where
  amavisd.conf is writable by UID running amavisd (e.g. not owned by root).
  If a (non-root) username or UID is specified, privileges are now dropped
  _before_ opening and evaluating a configuration file. The consequence
  is that the configuration variables $daemon_user and $daemon_group
  (in amavisd.conf) can not have an after-effect (a warning is issued
  if different).

  If -u is not specified, the behaviour is as before, i.e. the config file
  is read and evaluated under the current UID (as root unless 'su' was done),
  and the values of $daemon_user and $daemon_group from the config file
  are passed to Net::Server, which changes UID during its startup after
  chroot-ing (if requested).

  If chroot is desired, the -u must not be used: the root privilege is
  required to do chroot, and the config file must be read _before_ doing
  chroot. A case of Catch-22. Be doubly careful of who can modify the
  configuration file.

  Another consequence of specifying -u is that any external files (e.g.
  templates, lookup hashes) as possibly read from amavisd.conf, are now
  accessed as unprivileged user and no longer as root. The same goes for
  opening the log file when not logging via syslog.


- policy banks hold sets of configuration variables controlling most of
  per-message settings, including: static lookup tables, IP interface
  access rules, forwarding address, log level, templates, administrator
  addresses, spam trigger levels, quarantine rules, lists of anti-virus
  scanner entries (or just a subset), banned names rules, defang settings,
  etc. The whole set of these settings may be replaced with another
  predefined set based on incoming port number, making it possible for
  one amavisd daemon to cope with more diverse needs of served user
  communities which could so far only be implemented by running more than
  one instance of the amavisd daemon, each with its own configuration file;

  This mechanism brings new potentials for the future: in principle policy
  banks could be swapped not only based on port number or SMTP client
  IP address, but on any characteristics pertaining to a mail message as
  a whole (not specific to each of its recipients), or to characteristics
  of a connection from a mailer (e.g. the interface address or protocol);

  Until a better mechanism is available, a policy bank named 'MYNETS' has
  special semantics: this policy bank gets loaded whenever MTA supplies a
  SMTP client IP address (Postfix XFORWARD extension or a new AM.PDP protocol)
  and that address matches the @mynetworks list.

  A hash %$interface_policy is a current mechanism of assigning a
  policy bank to an incoming TCP port number (port must be in the list
  @$inet_socket_port). Whenever the connection from MTA is received,
  first a built-in policy bank with an empty name - the $policy_bank{''}
  gets loaded, which bringings in all the global/legacy settings.
  Then it is overlaid by whatever configuration settings are in the bank
  named in the $interface_policy{$port} if any, and finally the bank 'MYNETS'
  is overlaid if it exists and the SMTP client IP address is known
  (by XFORWARD SMTP extension command from MTA) and it matches @mynetworks.
  See amavisd.conf-sample for examples.

  When a new policy bank is overlaid over an existing set of configuration
  variables, variables not present in the new policy bank retain their value.

  The built-in policy bank (with empty name) is predefined, and includes
  references to most other variables (the dynamic config variables),
  which are accessed only indirectly through the currently installed
  policy bank. Overlaying a policy bank with another policy bank may
  bring in references to entirely different variables, possibly unnamed.

  Configuration variables are referenced from a built-in policy bank (which
  is implemented as a perl hash, i.e. an associative array) by keys of the
  same name, e.g. { log_level => \$log_level, inet_acl => \@inet_acl, ...}.
  For scalars one level of indirection is allowed, e.g.
  a policy bank { log_level => \$log_level }; $log_level=2;
  is equivalent to { log_level => $log_level } or { log_level => 2 },
  but in the first example with an indirect reference, the $log_level
  may be assigned to even _after_ the policy bank has already been formed.

  A word of caution: the syntax of entries within a policy bank hash
  is slightly different from assignments to configuration variables.
  This is because entries within policy bank are not asssignments, but
  key=>value pairs as in any Perl hash. And these pairs are delimited by
  commas, unlike statements, which are delimited by semicolons.
  Value is separated from its key by '=>' (or by a comma), whereas the
  assignment operator is '='. Keys of a policy bank are without leading
  $ or @ or %, unlike variable names. Values of a hash can only be scalars
  (e.g. strings or references).

  - value of a policy bank is a reference to a Perl hash, e.g.:
    { log_level => 3,
      forward_method => 'smtp:[]:10025',
      spam_admin_maps => ["spamalert\@$mydomain"],
  - normal assignments look like:
      $log_level = 3;
      $forward_method = 'smtp:[]:10025';
      @spam_admin_maps = ("spamalert\@$mydomain");

  And a final note: Perl can detect and report typing mistakes in variable
  names, but mistyped key is just some unused hash entry lurking in a hash,
  never used and never reported as mistyped/useless.

- @score_sender_maps is a soft variant of black- and white-listing,
  contributing positive or negative score points based on sender
  e-mail address. Btw, the 'score' in the name '@score_sender_maps' is meant
  as a verb, recipient is scoring a sender (= to grade, to determine the
  merit of);

  Whitelisting is becoming deprecated because it is often and easily abused
  (but blacklisting can still be useful);

  # Instead of hard black- or whitelisting, a softer approach is to add
  # score points (penalties) to the SA score for mail from certain senders.
  # Positive points lean towards blacklisting, negative towards whitelisting.
  # This is much like adding SA rules or using its white/blacklisting, except
  # that here only envelope sender addresses are considered (not addresses
  # in a mail header), and that score points can be assigned per-recipient
  # (or globally), and the assigned penalties are customarily much lower
  # than the default SA white/blacklisting score.
  # The table structure is similar to $per_recip_blacklist_sender_lookup_tables
  # i.e. the first level key is recipient, pointing to by-sender lookup tables.
  # The essential difference is that scores from _all_ matching by-recipient
  # lookups (not just the first that matches) are summed to give the final
  # score boost. That means that both the site and domain administrators,
  # as well as the recipient can have a say on the final score.

- the MIME defanging (defang: to make harmless or less powerful)
  wraps the entire original mail in a MIME container of type
  'Content-type: multipart/mixed', where the first part is a text/plain
  with a short explanation, and the second part is a complete original
  mail, enclosed in a 'Content-type: message/rfc822' MIME part.
  Defanging is only done when enabled (selectively by malware type)
  and the malware is allowed to pass (*_lovers or *_destiny=D_PASS).
  The feature is global, i.e. not available on a per-recipient basis.

  Conventional mail header fields are retained, and header fields
  Resent-{From,Date,Message-ID} are added. A header field X-Amavis-Modified
  is inserted to indicate that the mail body has been modified.

  Note that defanging changes mail body and makes subsequent DomainKeys/DKIM
  verifications to fail. The DomainKey-Signature header field is not
  retained in defanged mail (but is retained in the attached original mail).

  It is an experimental feature (disabled by default, except for mail bombs),
  and not very efficient for large mail.

  MIME defanging is unconditionally done for mail bombs, i.e. when the
  X-Amavis-Hold header field is added. The text in the first MIME part
  describes the reason, e.g.:

      WARNING, possible mail bomb, NOT CHECKED FOR VIRUSES:
      Exceeded storage quota 29089500 bytes by do_unzip; ...

  When MIME defanging is enabled for passed spam, the first MIME part
  contains the full SpamAssassin report.

  The MIME defanging feature is not available in the sendmail milter setup.

- new parameter: @mynetworks
  It is an IP access list which determines if the original SMTP client
  IP address belongs to our internal networks. It is much like the Postfix
  parameter 'mynetworks' in semantics and similar in syntax, and its value
  should normally match its Postfix counterpart. It affects the value
  of a macro %l (=sender-is-local), and the loading of policy 'MYNETS'
  (if present). Note that '-o smtp_send_xforward_command=yes' (or its lmtp
  counterpart) must be enabled in the Postfix service that feeds amavisd,
  otherwise client IP address is not available to amavisd-new and new
  features based on @mynetworks do not work (the %l macro works as before
  and bases its decision on sender e-mail address matching local_domains);


- syslog priorities are now dynamically derived from the log level of each
  log message (the first argument of sub do_log). The priority as specified
  in the $SYSLOG_LEVEL configuration variable is ignored in 2.0 (no longer
  in 2.1), but the 'facility' is not ignored. This makes it possible
  to influence the log verbosity by syslog.conf settings. Here is an
  example of a useful syslogd.conf setting (some levels may be left out;
  assumes the $log_level is high enough, e.g. 2, to produce any low-priority
  log messages at all):

    mail.err      /var/log/messages
    mail.notice   /var/log/amavisd.log     /var/log/amavisd-info.log
    mail.debug    /var/log/amavisd-debug.log

  (On linux do not forget to prefix filenames in syslog.conf with
   a '-' to disable sync, which has much impact on syslog performance!)

  At the moment the mapping of message log levels to syslog priorities
  is hardwired:
    level <= -3: LOG_CRIT
    level <= -2: LOG_ERR
    level <= -1: LOG_WARNING
    level <=  0: LOG_NOTICE
    level <=  1: LOG_INFO  (in version 2.0)
    level <=  2: LOG_INFO  (since 2.1)
    else:        LOG_DEBUG

  Not to be confused with the $log_level configuration variable setting,
  which still works as before, suppressing generation of all log messages
  with log levels above $log_level.

  For efficiency reasons one should not specify unnecessarily high $log_level
  and then discard low syslog priority messages in the syslogd.

- added $log_recip_templ variable, which is similar to $log_templ,
  but gets called for every recipient (the $log_templ is evaluated only once
  per message). Normally one or the other log template should be disabled
  by assigning undef or an empty string to the corresponding variable;

- the default $log_templ no longer shows quarantine ID if quarantining is

- added a macro %. (a dot), which might be useful in the $log_recip_templ;
  Its value is empty when $log_templ is expanded, and is a recipient counter
  (starting by 1) when $log_recip_templ template is expanded. Based on
  this macro one can provide a single template for both the $log_templ
  and the $log_recip_templ if needed, or perhaps let the log entry for
  the first recipient be more verbose that the rest;

- added a macro %T which expands to a list of triggered SA tests, but only
  when $log_templ and $log_recip_templ are expanded. In notifications
  the %T is still a list of To: addresses. An overlaid semantics, but we
  are running out of letters and a macro expander rewrite would be needed;

- new macros %k, %1, %2, %O, please see README.customize


- new configuration variables make it more flexible to specify arbitrary
  list of lookup tables. Legacy configuration variables are still available
  and are referenced from the default values of @*_maps lists. If these lists
  are redefined, legacy variables are not used.

  The new variables (lists) are:

    @bypass_virus_checks_maps @bypass_spam_checks_maps
    @bypass_banned_checks_maps @bypass_header_checks_maps
    @virus_lovers_maps @spam_lovers_maps
    @banned_files_lovers_maps @bad_header_lovers_maps
    @virus_admin_maps @spam_admin_maps
    @banned_quarantine_to_maps @bad_header_quarantine_to_maps
    @spam_quarantine_to_maps @spam_quarantine_bysender_to_maps
    @spam_tag_level_maps @spam_tag2_level_maps @spam_kill_level_maps
    @whitelist_sender_maps @blacklist_sender_maps @score_sender_maps
    @addr_extension_virus_maps  @addr_extension_spam_maps
    @addr_extension_banned_maps @addr_extension_bad_header_maps


- for more informative logging of lookup operations, a new object type
  Amavis::Lookup::Label can be inserted to lists of lookup tables
  for the purpose of labeling the main purpose of the list;

- all lookup* subroutines can now return matching key when called in a
  list context;

- lookups can now return a list of all matching entries (not just the first
  match); used for the new soft- white/blacklists (@score_sender_maps);

- sub lookup() now allows for one level of list elements dereferencing,
  which makes possible the construction of the argument list and later
  still be able to modify its members (e.g. creation of regexp lookup table
  objects in the configuration file); It facilitates transition from old
  hard-wired lists of lookup tables to new @*_maps list variables which
  permits specifying an arbitrary number of lookup tables and to specify
  their search order;

- simplify and unify calls to lookup() by collecting arguments (references
  to lookup tables) in lists, e.g. @local_domains_maps, @virus_lovers_maps,
  @virus_admin_maps. These array variables default to lists of legacy
  variables, which are now never directly used by the program. Either
  the individual legacy variables may be assigned to, or the entire list
  replaced, in which case the legacy variables no longer have any effect.

- lookup_acl: respect $localpart_is_case_sensitive setting;

- lookup_hash and lookup_sql: rewritten lookup_hash and factored out the
  common code from lookup_hash and lookup_sql to make_query_keys();

- lookup_hash bug fix: avoid splitting address literal as if it were
  a domain name; (a bug with key '.' not being tried for address literals
  fixed thanks to Uwe S. Fuerst);


- extended semantics of SQL field wblist.wb, which can hold a score value
  boost, which is interpreted as soft black/white-listing (the same semantics
  as the value in @score_sender_maps);

- recognize SQL server error 2013/"Lost connection to" and treat it
  the same as 2006/"MySQL server has gone away"; by Max Kalika;

- full domain stripping: @.

- lookup_hash and lookup_sql: limit the list of subdomain
  search keys to 10 levels as a sanity measure; e.g. for address the subdomains keys search list
  starts at; (domain names are limited by standard
  to 127 levels);

- prepare SELECT statements on demand;

- recognize all-zero and all-null boolean fields as false;

- recognize new (optional) fields in the table 'policy':
    virus_quarantine_to, banned_quarantine_to, bad_header_quarantine_to
    addr_extension_virus, addr_extension_spam,
    addr_extension_banned, addr_extension_bad_header

  (the addition of virus_quarantine_to was suggested independently
  by Harald Kapper and by Dipl.Ing. Martin Boeck);

- consider the SQL user id a string (no longer required to be numeric);
  thanks to Max Kalika / Gentoo support;

- LDAP white/black list support by Jacques Supcik (similar to contribution
  from Scott Dier and Eric Dorland, which I forgot about, sorry);

- added amavisSpamQuarantineTo to the LDAP schema;
  new version of LDAP schema; by Jacques Supcik:

  In the previous schema, the tag levels have been defined as integer.
  This is too restrictive, and have now been changed to strings
  (there is no float type in LDAP);

- added LDAP attributes for completeness: amavisBadHeaderLover,
  amavisBypassBannedChecks, amavisBypassHeaderChecks, amavisVirusQuarantineTo,
  amavisBannedQuarantineTo, amavisBadHeaderQuarantineTo; by Jérôme Schell;


- provide optional ability to retain complete email message in its
  un-decoded form (alongside its decoded parts) for a virus scanner
  to see (enabled if $keep_decoded_original_re matched string 'MAIL');
  suggested by Tomasz Papszun; (partly backported to amavisd-new-20030616-p8);

- rewrite code that generates new file names (Amavis::Unpackers::Part->new),
  and rewrite code dealing with banned names. Keep information about each
  part organized as a tree, matching the descendence of each part, new
  package/object Amavis::Unpackers::Part to collect such information;

- make Amavis::Unpackers an optional module: the interface code to external
  decoder/unpackers/dearchivers does not get compiled and does not consume
  virtual memory if $bypass_decode_parts is true; (previously it just
  didn't get called, but was sitting in memory nevertheless);

- decode RPM archives by converting them to cpio, if rpm2cpio and cpio
  are available;

- do_tnef: extract $tnef->message if it exists, not just $tnef->attachments;

- support extracting MS cabinet files (CAB) by calling cabextract, if enabled
  and found. Beware: Lars Hecking warns that cabextract 0.6 is quite buggy
  and the author has been notified.


- support for ClamAV virus scanner via Perl module Mail::ClamAV,
  based on code by Roberto Pereira da Rosa;

- don't call virus scanners if there are no files in the directory to be
  scanned (e.g. mail with an empty body); some virus scanners don't like
  to be given an empty directory (e.g. Symantec newer savsecls);
  reported by Marco Bicca;

- rewritten/unified/generalized subroutines ask_daemon and sophos_savi
  based on the new subroutine ask_av;

- scan parts directory for file names exactly once regardless of the number
  of virus scanners and their arguments (containing '{}/*' or not);

- supply full original mail to virus scanners in case of MIME parse errors
  (in addition to any possibly decoded parts);

- when collecting file names to be virus scanned, prepare a hash which maps
  base file names to Amavis::Unpackers::Part objects, and make it available
  to virus scanner interface routines, which may benefit from the additional
  information about the file to be scanned.

  In particular, the new interface to Mail::ClamAV now turns on the option
  CL_MAIL, and the interface to SAVI turns on the option MIME, when entire
  mail is passed to AV scanner for checking. This enables ClamAV and SAVI
  to attempt MIME decoding the file by itself.

  TODO: The same option (--mime) would need to be specified when calling
  'clamscan' and supplying a non-decoded mail for checking; pointed out
  by Riccardo Ghiglianovich and Michael Boelen;


- The 'amavisd reload' command is now implemented differently:
    old: signals SIGHUP to a running amavisd process and exits immediately;
         the running amavisd process (under control of Net::Server) when it
         receives a SIGHUP starts its own copy with same arguments and exits;
    new: signals SIGTERM to a running amavisd process, waits for it to finish,
         then continues (same as 'amavisd start') to become a new daemon;
  The new method works even when chrooted, and is more reliable when the
  existing process is slow to terminate, as it actively waits for the previous
  daemon to finish before proceeding to promote itself to become a new daemon.

- a simple demo program 'amavisd-agent' is provided, allowing
  for continuous inspection of SNMP counters; a path to the
  /var/amavis/db is hardwired in the program, modify it if necessary.

- server-side support for optional Postfix SMTP/LMTP command XFORWARD:
  information about the original SMTP client IP, its DNS name, HELO name
  and protocol used is now made available to the amavisd program for
  logging and other purposes. The same information can also be obtained
  from the 'Amavis policy delegation protocol (AM.PDP)' if the helper
  program supports it (useful for sendmail milter setup);

- client-side support for optional Postfix SMTP command XFORWARD:
  if MTA announces in its SMTP EHLO response that it supports XFORWARD,
  amavisd will provide additional information about the original SMTP client
  if the information is available (either from XFORWARD on the receiving
  side, or by the 'Amavis policy delegation protocol';

- server side support for the new amavis helper protocol AM.PDP which allows
  for header modifications, removal of recipient addresses (e.g. non- spam
  lovers) or rewrite of recipient addresses (e.g. adding address extensions),
  and specification of full SMTP response;

- modified search logic for matching mail parts against $banned_filename_re;

  The old search order for names did not result in what one might
  expect when pattern list elements with a value of false were used
  in $banned_filename_re. Namely, all three components were searched
  independently (Content-Type, declared name, and file(1) type)
  and a logical or was used. Because searches for each mail part were
  independent, it was not possible to specify for example that anything
  within a zip would be allowed. If any of these searches returned true,
  mail was blocked.

  To make this useful, a complete rewrite of mail unpacking was needed
  and all information be made available in one place after the unpacking
  is over, so that checking for banned names can be done all at once.

  The search order is now much the same as used in rsync and its server,
  see man rsync, section 'EXCLUDE PATTERNS'. The new comments in
  amavisd.conf-sample explain the new situation.

- replaced $relayhost_is_client by a more flexible specification:

  To make it possible for several hosts to share one content checking daemon,
  the IP address and/or the port number in $forward_method and $notify_method
  may be specified as an asterisk. An asterisk in the colon-separated
  second field (host) will be replaced by the SMTP client peer address
  (i.e. the MTA host). An asterisk in the third field (tcp port) will be
  replaced by the incoming SMTP/LMTP session port number plus one. This
  obsoletes the previously used less flexible configuration parameter
  $relayhost_is_client. An example:

    $forward_method = 'smtp:*:*'; $notify_method = 'smtp:[]:10025';

  The same functionality can also by achieved by using a bigger hammer,
  the policy banks. These may completely replace the global settings
  for $forward_method and $notify_method, based on incoming port number;

- turn address extension variables (the so called "plus addressing")
  into recipient-based lookup tables, including the SQL lookups. For example:
    @addr_extension_virus_maps  = ('virus');     # defaults to empty
    @addr_extension_spam_maps   = ('spam');      # defaults to empty
    @addr_extension_banned_maps = ('banned');    # defaults to empty
    @addr_extension_bad_header_maps = ('badh');  # defaults to empty
  or perhaps:
    @addr_extension_virus_maps = (
      {''=>'infected', ''=>'malware'}, 'virus' );
  suggested by Gentoo modification, Jacques Supcik, and others;

- log and report hits and tag/tag2/kill levels rounded to three decimal
  places (trailing zeroes trimmed), no longer rounded to one decimal place;

- added @spam_dsn_cutoff_level_maps, making it possible to specify
  different DSN cutoff levels for different recipient domains or users.
  In multi-recipient messages where recipients can specify different values,
  the maximum value is used for deciding whether DSN should be suppressed;
  suggested by Ales Casar;

- configuration variable $gets_addr_in_quoted_form is no longer used;
  knowledge about address form (quoted or not) is now implicit in the
  receiving protocol;

- if tag level turns out to be undef, it will not be shown in X-Spam-*
  header fields, and will be interpreted as having a value lower than any
  spam score when deciding whether to insert X-Spam-* header fields or not;

- added macros %a and %g:
    * %a  original SMTP session client IP address (empty if unknown)
    * %g  original SMTP session client DNS name (empty if unknown)
  (like macros %I and %M that were once proposed by Dibo in his 2002-07 patch)
  This information may be available from Postfix when XFORWARD protocol
  extension to SMTP is enabled, and can be made available by helper program
  (e.g. from sendmail milter) when the new AM.PDP protocol is used;

- added macro %p, expanding to a current policy bank name (or empty
  if a built-in policy bank is still in place);

- added macro %r, expanding to the contents of the first Resent-Message-ID
  header field, or empty if no such field exists. Include reporting the
  Resent-Message-ID in the log and in the sender notification;
  suggested by Oliver Gorwits;

- new configuration variable $addr_extension_bad_header for completeness;

- added $bad_header_quarantine_to, @bad_header_quarantine_to_maps,
  $bad_header_quarantine_method, and $warnbadhrecip for completeness;
  suggested by Robin Lynn Frank;

- MIME::Parser errors now contribute to bad-header checks, so that the
  header checking is now conceptually extended to MIME sub headers
  (Postfix similarly considers MIME subheaders part of mail headers);
  MIME::Parser 6.1xx or later is recommended.

- allow $*_quarantine_method to be undef as a quick way of disabling
  some quarantine (it also can be disabled as previously, by using
  method 'local:' and following its rules);

- persistent cache of recent virus and spam checks, global to all child
  processes, can improve the cache hit rate. Uses BerkeleyDB database (hash
  and queue) and its interlocking mechanisms (Berkeley DB Concurrent Data
  Store) for the purpose. The V3.1 or better is required, V4 is recommended.

- include version information in the 'Usage' text;

- rewritten lookup_ip_acl() and added ip_to_vec() to allow for IPv6
  address syntax as specified in rfc3513 to be used in IP lookups;

- @inet_acl now defaults to ('', '::1'), i.e. it adds the
  IPv6 loopback address to the list;

- new configuration variable $sa_spam_level_char (defaults to '*')
  allows specifying another character for X-Spam-Level bar. Empty or
  undefined value disables inserting the X-Spam-Level header field;

- added configuration variable $sa_spam_report_header to enable/disable
  inserting the X-Spam-Report header; patch by Craig Sanders;

- added $banned_quarantine_to configuration setting to make possible the
  quarantining of banned mail to a different place from viruses;

- don't insert virus-, banned- and bad headers- related headers for passed
  mail to recipients with corresponding bypass_*_checks, making them believe
  the mail was not spam-checked (as they are not expecting such headers
  anyway); This was already done in version amavisd-new-20030616-p6
  but only for spam-related headers;

- for choosing address extensions use the same criteria as for adding
  header fields, e.g. pretend to not know the result of a certain test
  (virus, spam, ...) when recipient chooses to bypass such test,
  even if the result of the test is known;

- added variable $sa_spam_subject_tag1 (undef by default).
  If $sa_spam_subject_tag will not be inserted (at tag2 level), and
  $sa_spam_subject_tag1 is nonempty, this string (e.g. '***possible SPAM*** ')
  will be inserted into the Subject header field for spam levels above
  tag level; suggested by Immo Goltz;

- added separate configuration variables $banned_files_quarantine_method
  and $bad_header_quarantine_method. Quarantining of banned files and
  bad headers were previously controlled by $virus_quarantine_method;

- rewritten read_hash, it is now possible to specify key value (right-hand
  side) for each key. If value is not specified, a '1' is assumed as before;

- SMTP server support for rfc2554 authentications (PLAIN and LOGIN only)
  allows client authentication to be relayed to the MTA when message is
  forwarded. Might be useful if amavisd-new is used in a Postfix SMTP proxy
  setup, but is not needed for other setups. Disabled by default,
  see variables $auth_required_inp and @auth_mech_avail.

- SMTP client support for rfc2554 authentications (any authentication method
  as supported by Net::SMTP and Authen::SASL Perl modules). Authentication
  of forwarded mail (PLAIN and LOGIN only) is carried over from the incoming
  mail, authentication to be used when submitting notifications is controlled
  by configuration variables $amavis_auth_user and $amavis_auth_pass.
  Disabled by default, see variable $auth_supported_out (later renamed
  to $auth_required_out).

- when passing envelope sender address to SpamAssassin, supply it as a
  rfc2822-standard header field Return-Path, and no longer as X-Envelope-From
  (the change came with a pre-release amavisd-new-20040301).

- provisional/experimental support for DSPAM spam checker (pre 3.0):
  if configuration variable $dspam is nonempty and represents a path to
  a 'dspam' program, a message is passed to dspam and its inserted headers
  of the form X-DSPAM-* are axtracted and then made available for
  SpamAssassin rules to check and score if desired.

  All messages are currently presented to dspam as the same user, affecting
  how its database is being built. False negatives and false positives
  (based on SA assessment) are fed back into DSPAM as a simple form
  of auto-learning. Works reasonably well, but do not expect miracles.
  See subroutine spam_scan.

  Here is how DSPAM can be installed to be able to be used by amavisd-new:

    dspam 2.x:
    ./configure --enable-alternative-bayesian \
      --with-userdir=/var/amavis/dspam --enable-signature-headers \
      --without-local-delivery-agent --without-quarantine-agent

    dspam 3.0.0:
    ./configure \
      --with-dspam-home=/var/amavis/dspam --enable-signature-headers \
      --without-delivery-agent --without-quarantine-agent

    make install
    chmod u-s,a+rx /usr/local/bin/dspam
    chown vscan:vscan /var/amavis/dspam

  User 'vscan' may need to be added to DSPAM trusted.users file.

  The following can be inserted into the SA config file (
  to make it recognize and incorporate DSPAM's assessment:

    header DSPAM_SPAM X-DSPAM-Result =~ /^Spam$/m
    describe DSPAM_SPAM DSPAM claims it is spam
    score DSPAM_SPAM 0.5

    header DSPAM_HAM X-DSPAM-Result =~ /^Innocent$/m
    describe DSPAM_HAM DSPAM claims it is ham
    score DSPAM_HAM -0.1

  Eventually DSPAM support should be removed from amavisd-new, as soon as
  SA will be able to call it on its own.


- reformatted the whole program, reducing indentation from 4 to 2
  and replacing TABs with spaces (with some dubious help from perltidy,
  plus lots of manual adjustments);

- completely rewritten code to handle both the old and the new amavis helper
  protocol, as well as Postfix 'TCP client/server table lookup protocol'
  as specified in the Postfix documentation: tcp_table(5);
  (process_policy_request, prepare_policy_query, check_amcl_policy)

- type_short may now be a list of short types, not necessarily just a single
  value. Typical use is to classify a MS executable as both an 'exe' and
  as 'exe-ms', which makes more specific banned rules possible without
  unnecessary complication in regexps;

- parts now carry attributes, which can be inspected for banned checks;
  current attributes are U for undecodable, and C for (en)crypted;

- opened another can of Perl worms (taint bugs): turn on Perl pragma
  "use re 'taint'" in all modules, and selectively turn it off where needed.
  It replaces cumbersome manual preservation of taintedness when regexp
  saved ranges are used without intention to untaint. Because of Perl bugs,
  strategically placed local($1,$2,...) are needed, otherwise previous
  taint flag in $1, $2, ... can be brought on to new variables, which can
  all of a sudden become tainted out of nowhere;

- catch and report throws (die) in pre_loop_hook() to properly report
  problems during initialization;

- introduced subroutine exit_status_str and unify reporting of
  subprocess status;

- enhanced sub best_try_originator to ignore IP addresses from private,
  local and dedicated IANA networks (rfc3330) and look for the first
  public address in the 'Received' path;

- examine first four (chronologically) Received header fields (instead of
  first two) when looking for an originator IP address, and ignore those
  with private IP addresses;

- moved code dealing with body digest cache to a new package
  Amavis::Cache to facilitate transition to shared or persistent cache;

- new explicit cache expiration time controls (time to live in seconds):
    $virus_check_negative_ttl $virus_check_positive_ttl
    $spam_check_negative_ttl $spam_check_positive_ttl

- discard cache db ($db_home) and recreate it during restart;

- more informative changes of child process $0, which may show in the ps(1)

- store tempdir of a current message to the Amavis::In::Message object;

- gather some statistics about idle time;

- reorder and adjust mapping from file(1) results to file type classes;

- optimization: instead of invoking file(1) utility program for each
  part to be analyzed, now call it once for each round of currently
  available parts, giving it the list of all available parts as arguments;

- shorten the names of parts from part-..... to p..., to be able to stash
  more files names into a command line, e.g. when calling file(1) or
  external virus checker which can not deal with directories;

- use regexp lookup table mechanism (table $map_full_type_to_short_type)
  to match long types (output of file(1) utility) to short types (.exe,
  .jpg, .doc, ...).  The default table can be replaced by a customized
  table in amavisd.conf;

- replace $(?!\n) with \z in regular expressions throughout;
  replace most of the remaining $ with \z in regular expressions.
  The regular expression primitive \z is available since perl 5.005.

- TODO: disregard $MAXFILES during initial MIME unpacking;
  reported by Stephane Lentz and Robert LeBlanc (done in 2.3.2);

Some un-edited notes on the new banned rules mechanism:
(wrapped log lines, and replaced \\ by \ for clarity:

| Feb 24 19:07:29 hauptpostamt amavis[29847]: (29847-04-5) p.path BANNED:
|  "P=p002,M=application/octet-stream,T=zip, |
|   P=p003,T=exe,T=exe-ms,N=document.htm   .scr",

part p003 is of type (file(1)) MS executable, with suggested
name "document.htm   .scr" (lots of spaces in the name)

its parent resides on temp file p002 (i.e. p003 was extracted from it),
which is of type (T) zip archive, with suggested (MIME) name
(N) "", and has a MIME type (M) "application/octet-stream".

such a component p003 lying within such p002 is considered banned
by the following regexp rule (one rule within the $banned_namepath_re list):

|     matching_key="(?mix-s:^ (.*\t)? N= [^\t\n]* \. [^./\t\n]* \.
|                    (exe|vbs|pif|scr|bat|cmd|com|dll) (\t.*)? $)"

which says that any component at any level must not have a name (N)
matching a pattern:
  any number of characters,
  a dot,
  any number of non-dot and non-slash characters
  a dot,
  and ending with: exe or vbs ...
(basically: double extension ending with listed extensions)

The complications such as using [^\t\n]* instead of .* are there
to keep regexp contained within fields and ancestors/descendents.

There is one detail to remember when comparing logged p.path log entries
and the actual matching rules:

- for the sake of readability the logged entry has \n (newlines) converted
  into ' | '. The \n is a separator between components in the tree
  from the root (the mail itself, hidden) to the leaf component
  which can not be further expanded (i.e. not an archive)

- for the sake of readability the logged entry has \t (a tab) converted
  into comma, separating information fields such as P=...  M=..  T=..  M=..

So the above logged string:
  P=p002,M=appl...,T=zip, | P=p003,T=exe,T=exe-ms,N=document.htm.scr
is actually a single string:

and a Perl regexp is applied directly to it.

The raw string is rather unsightly, but the \n and \t were chosen
to minimize clash with valid characters within file names.

If a \n or \t is present in a name of the components, such character
is converted into a space to avoid clashing with separators.

| Feb 24 19:11:58 hauptpostamt amavis[31505]: (31505-01-5) p.path BANNED:
|   "P=p002,M=application/octet-stream,T=zip, |
|    P=p003,T=exe,T=exe-ms,N=paypal.scr",

a MS executable named "paypal.scr" within a zip archive ""

|      matching_key="(?mix-s:^ (.*\t)? N= [^\t\n]* \.
|                    (exe|vbs|pif|scr|bat|com) (\t.*)? $)"

block component at any level with a name (N) terminating
by dot followed by any of the listed extensions.

| Feb 24 19:18:25 hauptpostamt amavis[32159]: (32159-01-2) p.path BANNED:
|   "P=p002,M=application/octet-stream,T=zip, |
|    P=p003,T=exe,T=exe-ms,N=text.txt     .exe",

a MS executable named "text.txt     .exe" (with lots of spaces in the name)
within a zip archive named ""

|      matching_key="(?mix-s:^ (.*\t)? N= [^\t\n]* \. [^.\t\n]* \.\n
|                    (exe|vbs|pif|scr|bat|cmd|com|dll) (\t.*)? $)"

blocked by the double-extension rule.

| Feb 24 19:30:15 hauptpostamt amavis[1690]: (01690-02-8) p.path BANNED:
|   "P=p002,M=application/octet-stream,T=zip, |
|    P=p003,T=exe,T=exe-ms,N=jokes.doc   .exe",
|      matching_key="(?mix-s:^ (.*\t)? N= [^\t\n]* \. [^.\t\n]* \.\n
|                    (exe|vbs|pif|scr|bat|cmd|com|dll) (\t.*)? $)"

same thing

                                                              June 29, 2004
Patch: amavisd-new-20030616-p10

- insert a security check to test for a missing Net::Server patch, and abort
  if vulnerable;

- provide and use our own subroutine q_encode to do the Q-encoding when
  editing an existing invalid header field with non-encoded 8-bit characters,
  e.g. when inserting ***SPAM*** or ***UNCHECKED*** into Subject.
  The MIME::Words::encode_mimeword() does not encode spaces and does not limit
  encoded words to 75 characters, which violates the RFC 2047 and breaks mail
  readers; reported by Sebastian Hagedorn and Gregor Hoffleit;

- fixed a bug in inserting the tag_level header field, which was missing
  if sender was blacklisted and tag_level was greater or equal to 0;
  thanks to Joerg Thaler;

- amavis-milter.c
  * log envelope sender address at the same log level (DBG_INFO) as
    recipient addresses;
  * remove the log message "(mlfi_eom) header already present", it was
    inappropriate, the call to smfi_chgheader succeeds even if no such
    header was already present;
  * relax permissions on created directory and temporary file to allow
    group read access (needed if virus scanner runs under a different
    user id but within the same group); by Adam C. Migus;
  * add queue id (when available) to most log messages;

- do not preserve evidence just because a message gets an X-Amavis-Hold
  header field;

- when logging directly to a file and started as root, change UID of a
  log file to $daemon_user to avoid restart problems; based on a patch
  by Carsten Hoeger and Gregor Weiss;

- added configuration variable $first_infected_stops_scan
  to stop anti-virus scanning when the first scanner detects a virus;
  the default is false, all scanners in a group are called (as usual);

- in the AV entry for clamscan added the option '--tempdir=$TEMPBASE';

- in the AV entry for 'Norman Virus Control v5 / Linux'
  changed the command name 'nvccmd' into 'nvcc'; correction by
  Michael Ramke of the Norman Data Defense Systems GmbH;

- insert debug reports into sub ip_addr_to_name to help recognize DNS problems;

- revoke deleting an existing 'X-Scanned-By' header field, which was
  introduced in -p8; removing it gets in a way when more than one content
  filter is being chained;

- load a missing SpamAssassin v3.0 module, needed when running in chroot;

- contributed (for now in the form of an optional patch): support for
  the incoming qmqpqq protocol over a TCP socket - to be used with qmail.
  Apply the provided patch 'amavisd-new-qmqpqq.patch', it updates files
  amavisd.conf and amavisd in the current directory; by Martin Solciansky;

- documentation updates;

                                                              April 2, 2004
Patch: amavisd-new-20030616-p9

- avoid choking on undefined $banned_filename_re; thanks to Ales Casar,
  Sebastian Hagedorn and E. Falk;

- if Subject mail header field got $undecipherable_subject_tag inserted, it
  would also receive the spam tag $sa_spam_subject_tag unconditionally;
  fixed, thanks to Francis Stevens;

- updates to the @av_scanners list in amavisd.conf:
  * DrWebD now works with a new DrWeb daemon 4.31,
    thanks to information provided by Krzysztof Cegielski, DrWeb Polska;
  * updated BitDefender bdc, thanks to Alfredo Milani Comparetti;
  * clamscan: use documented option --no-summary instead of the
    undocumented/old(?) option --disable-summary; by Georgy Salnikov;
  * updated Kaspersky aveclient and F-Secure fsav, thanks to Tomi Hakala;
  * recognize that KasperskyLab antivirus in demo mode turns on the bit 0x10
    in status; avp and avpdc now use the same two sets of status codes;
    avp: statuses 3 and 6 moved to infected (to match avpdc);
    statuses 2 and 5 considered infected; all suggested by Georgy Salnikov;

- when original undecoded mail is to be kept for virus scanners (requested
  by patterns /^MAIL$/ or /^MAIL-UNDECIPHERABLE$/ in $banned_filename_re),
  the preserved file is now named parts/part-00000 instead of parts/email.txt
  to preserve the size of the name, upon which some virus scanners depend
  (e.g. DrWebD);

  A note to Courier users: due to the way a file name in Courier is created
  and passed to amavisd-new, it is currently not possible to use the triggers
  /^MAIL$/ and /^MAIL-UNDECIPHERABLE$/ in $keep_decoded_original_re, or a
  failure to create a hard link occurs; Thanks to Bowie Bailey for helping to
  troubleshoot the problem; (this limitation is lifted in amavisd-new-2.2.0)

- remove option -c when calling gzip, bzip2, compress, lzop and unfreeze
  (as has long been done in the development version); the option -c is not
  needed when no file argument is present, and some implementations of gzip
  and compress may choke on it;

- make loading of Perl module Carp::Heavy optional, versions of Carp that
  came with Perl 5.005 did not have it; reported by Jefferson Pizzolatti;

- load module Mail::SpamAssassin::BayesStore::DBM, required by SA 3.0
  running in chroot jail;

- look for program 'gcpio' ( $cpio = ['gcpio','cpio'] ), as on OpenBSD the
  plain cpio does not support required options (the --no-absolute-filenames
  is essential), but GNU cpio does; thanks to Manfred Gloiber;

- relax parsing of file(1) output to allow tab as well as space to follow
  a file name; the file(1) on Solaris uses tab instead of space;
  suggested by Glen Harris;

- make LHA understand self-extracting archives (SFX); patch by Georgy Salnikov:

  Although lha unpacks any non-SFX archive independently if its extension,
  it unpacks the SFX lha archives only if they have the extension .exe.
  Now the file is symlinked to $part.exe for checking by lha: this will now
  work for SFX, and will still work for non-SFX. Also, do_lha will return 0
  if $exec and the part cannot be de-archived, the same as in do_unrar.

  The executable file formats are checked for being zip/rar/lha SFX archives.
  However, if the SFX archive is not a zip archive, do_unrar always returns
  success, so that if the archive is also not a rar, it will be never checked
  by lha. Now do_unrar returns 0 if $exec and the corresponding part can not
  be un-RARed;

- do_executable and do_unarj: added checking for SFX arj; by Georgy Salnikov:
  (commented out the call to do_unarj in do_executable until more experience
  is gained on how well unarj survives certain mail contents; Mark);

- do_unarj: let arj/unarj work on file named part*.arj;

- when calling IO::File::open() use '+>' instead of 'w+' to avoid
  Perl taint bug ($mode turns tainted) (bug still present in 5.8.2)
  triggered by expression in IO::Handle::_open_mode_string();

- attack the Perl 5.8.0/5.8.1/5.8.3 taint bug (once variables $1,$2,etc
  get tainted they start spreading taintedness to other variables):
  * insert local($1,$2,$3,...) in blocks of code which call external
    modules which trigger the bug (Mail::SpamAssasin, MIME::Parser, ...)
  * insert local($1,$2,$3,...) in blocks of code which depend on these
    variables to be clean, and which demonstrated through bug reports
    and experience with various version of Perl that these variables
    were not always taint-clean;
  The last taint incident triggered by SA 3.0.0 (svn) reported by Luc de Louw;

- recognize status EX_NOUSER when forwarding via pipe to sendmail (old setup);
  previously it was treated as a temporary failure; patch by The Mindflayer;

- turn error message 'error reading from client socket' back into
  'client broke the connection without a QUIT' for consistency with P7;

- when %virus_admin lookup table is used, prefer $msginfo->sender_source
  (unmangled sender domain) over $msginfo->sender; suggested by Pawel

- nicety: in virus recipient notifications now supply the To: header field
  with the true recipient address instead of "undisclosed-recipients:;"
  in case of single-recipient mail; suggested by several people;

- amavisd would insert differently capitalized header fields (either
  X-AMaViS-Alert or X-Amavis-Alert), depending on the reason being reported;
  now use the X-Amavis-Alert throughout; reported by Carsten Hoeger;

- fetch Perl auto-loaded modules auto::POSIX::setgid and auto::POSIX::setuid
  if they exist; they are needed by older versions of Perl when running in
  chroot. A manual change was needed until now (documented in README.chroot),
  which should no longer be needed; a FreeBSD problem report 64636.

- helper program amavis-milter.c (used in the sendmail milter setup):

  * amavis-milter.c only included the client IP and client host name on the
    first mail transaction of a multiple-transaction SMTP session, but not
    in subsequent transactions; pointed out by Stephane Lentz. The solution
    was once already provided back in July 2002 by Radoslav Dibarbora - Dibo,
    and forgotten; credits where credits are due:

  * make a clear distinction between message data and connection data,
    which required code reshuffling and revealed previous unclean solutions;

  * add error checking and reporting to mkdir/rmdir/open/unlink/write
    system calls; previously an error could pass by unnoticed or just
    caused a tempfail without an explanation;

  * change final milter status ACCEPT into CONTINUE to allow further
    milters in the milter chain to examine the mail;

  * code cleanup;

  * regenerated helper-progs/configure with Autoconf 2.57 to make it
    capable of detecting mkdtemp and still be able to find sendmail
    libmilter files; thanks to Sebastian Hagedorn for a problem report,
    for testing, and for revealing a bug in the pre-released version;

  * use different syslog priority for different internal message
    log levels; by Sebastian Hagedorn;

  * adjust log levels of messages, set default verbosity 1 (DBG_WARN);

- documentation updates, including the updated comments in amavisd.conf
  regarding whitelisted senders, to reflect the change indicated in the
  amavisd-new-20030616-p4 release notes;

                                                              March 9, 2004
Patch: amavisd-new-20030616-p8

- be compatible with SpamAssassin version 2.70 and 3.0 as well as 2.6x;
  SA changed its API, replacing Mail::SpamAssassin::NoMailAudit
  with $spamassassin_obj->parse and belonging objects;

- as a stop-gap solution to the W32/Bagle-{F,...} detection problem
  (password-protected zip archives), three new measures are available:

  * ability to present the full non-decoded original message to virus
    scanners was partly back-ported from the development version.
    Enabled by adding qr'^MAIL$' or qr'^MAIL-UNDECIPHERABLE$' to the
    list in $keep_decoded_original_re, as illustrated in amavisd.conf;
    similar to a patch by Ted Cabeen. The following keys are used for a
    lookup into $keep_decoded_original_re:

      always provide a full original message to virus scanners
      (besides its successfully decoded components);

      same as for 'MAIL', but only if it contains undecipherable components
      such as password protected archive members, unsupported compression
      methods or encrypted parts (e.g. with PGP). Don't put too much trust
      into this, as some more exotic file formats may not be understood
      and not flagged as undecipherable;

  * a key 'UNDECIPHERABLE' is matched against $banned_filename_re when mail
    contains any undecipherable components, and if lookup returns true,
    mail will be banned. For example:

    $banned_filename_re = new_RE(
      qr'^UNDECIPHERABLE$',   # contains any undecipherable components
    ... );

  * a string can be prepended to Subject (for local recipients only)
    if mail could not be decoded or checked entirely, e.g. due to
    password-protected archives or non-decodable mail bombs:

    $undecipherable_subject_tag = '***UNCHECKED*** ';  # undef disables it

  NOTE: this solution is a quick-fix response to popular demand.
  Although the same or similar functionality will probably remain
  in future versions, the syntax and exact semantics may be refined.

- bring do_unarj in line with the rest of de-archivers: provide the
  same degree of mail bomb protection that was available in do_unrar;
  retain original archive for inspection by a virus scanner if it contains
  any members that can not be extracted (e.g. password protected members
  or unsupported compression schemes). (The .arj size checking deficiency
  was mentioned in the AMaViS Security Announcement (ASA) 2004-01-19);

- don't call a virus scanner if there are no files (no mail parts) to scan;
  it caused problems with certain scanners like aveclient, which expect
  a list of file names as arguments and complain if the list is empty;
  reported by Daniel Luttermann;

- a much needed feature: can specify $sa_dsn_cutoff_level in amavisd.conf
  to suppress sending a DSN (delivery status notification) when spam level
  is above this value, effectively turning D_BOUNCE into D_DISCARD for
  this message; undef disables the feature and is a default. A good
  first approximation value is 10, or with some risk go down to 8.
  The parameter has no effect if DSNs are already disabled (e.g. when
  $final_spam_destiny is D_DISCARD or D_REJECT);

- do_unrar: double check the archive size (against summary line as well);

- derive the value of macro %l from sender_source instead of from a declared
  sender address, so as to not rise false alarms when sender address is known
  to be faked; patch by Joerg Friedrich;

- reduce the number of retries to connect to a daemonized virus scanner
  from 3 to 2, so that a fallback to backup scanners occurs sooner;

- updated $banned_filename_re example in amavisd.conf, modify it to will;

- updates to the @av_scanners list in amavisd.conf:

  * KasperskyLab AVP - aveclient: *IMPORTANT* updated entry for the exit
    status only reflects the result of the last file scanned, we must use
    regular expression to detect viruses; fix provided by Andreas Triller;

  * Trend Micro FileScanner - vscan: added option -za, otherwise
    some broken archive may sneak-in a virus; suggested by Stephane Lentz;

  * Sophos sweep: added options -cab -tnef --no-reset-atime;

  * Dr.Web command line scanner: updated options, recognize exit status
    when using evaluation license;

  * ClamAV clamd: change socket location to a more usual value, adjust to will;

- amavisd.conf: added new virus names to the $viruses_that_fake_sender_re
  list, and uncommented the [qr/.*/=>1] line, so that by default any unknown
  virus will be treated as a sender-faking virus; adjust to will;

- do not send bad header notifications in response to messages
  from mailing lists;

- added header check for folded header field lines made up entirely of
  whitespace (a 'header space gap' violation to rfc2822); a check is only
  made when other header checks (bad character in header) are enabled;

- distinguish an empty string from undef in $mailfrom_to_quarantine,
  making it possible to specify a null return path when quarantining
  to a mailbox;

- helper-progs/amavis-milter.c (sendmail) enhancement based on work
  by Stephane Lentz:

  * the name of a temporary directory is derived from the sendmail queue ID,
    making it easier to match sendmail and amavisd-new log entries;
    the queue ID is also a part of the quarantine file name;

  * a phantom 'Received:' header field is prepended on the temporary file
    to preserve the information on the original SMTP client IP address,
    host name and queue ID. This trace header field does not propagate to
    recipients (is not inserted into the original message), but is available
    in the quarantined messages and is visible to SpamAssassin.
    Adding a milter macro {b} on ENVFROM is advised to preserve the MTA
    timestamp in the log, although not mandatory (falls back to current time):

              confMILTER_MACROS_ENVFROM``, {b}'')dnl

  If the new amavis-milter.c gives you trouble, switch the soft link to the
  previous version in the same directory. Keeping macro {b} does not hurt.

- catch possible errors in pre_loop_hook and report them properly;

- better check for I/O errors on SMTP input socket;

- add the following modules to the list of pre-fetched modules:
    Carp::Heavy, IO::Handle, IO::Socket::UNIX, IO::Socket::INET.
  The absence of Carp::Heavy could mute error report or a backtrace
  when running chroot-ed; the rest are for completeness only;

- added a macro %z which expands to the original mail size (in bytes);
  it could be useful in the $log_templ; thanks to Nick Leverton;

- make SpamAssassin timeout value configurable: variable $sa_timeout;
  by Henrique M. Holschuh (Debian);

- 'Received:' header field cosmetics: use 'unknown' in case the HELO argument
  was empty, and put-in a 'unix socket' if message was received from a helper
  program; by Henrique M. Holschuh (Debian);

- remove more pre-existing X-Spam* header fields from other scanners:
  X-Spam-Tests, X-Scanned-By; by Henrique M. Holschuh (Debian);

- helper-progs/amavis.c updates (old sendmail setup): report errors,
  log to syslog, change default dir to something obvious for bug reporting;
  by Henrique M. Holschuh (Debian);

- updated documentation;

                                                            January 5, 2004
Patch: amavisd-new-20030616-p7

- do_unzip and do_unrar: retain an archive if any of its components is
  password protected or encrypted (plus unpack what can be unpacked,
  as before). This gives virus checkers a chance to examine the original
  unpacked archive as a whole (e.g. scanning for variants of W32/Mimail),
  matching it in non-decoded form against virus patterns even if containing
  password-protected components. As a consequence, some virus scanners
  may now log their complaint when encountering such protected archives
  which previously didn't reach them. Such log entries should be considered
  informational only;

- add module Net::DNS::RR::AAAA to the list of Perl modules to be fetched
  before chroot takes place; thanks to Per olof Ljungmark;

- preload Perl modules DBD::*, based on @lookup_sql_dsn;
  required when using SQL lookups from a chroot jail;

- updated example dsn in @lookup_sql_dsn (file amavisd.conf) to use the
  new DBD::mysql syntax, and to show how to force accessing SQL server
  via inet socket which makes it easier to use from a chroot jail;

- disregard cached spam results for mail with small body.
  The most pronounced undesirable effect was in mail with
  an empty body where spam score was derived from header only;

- change the default value for local_domains_sql lookup for the catchall key
  '@.' under conditions: when using SQL lookups and user record with key '@.'
  is present in the database and a field 'local' is not present. Previously
  it surprisingly defaulted to true, now it falls back to static lookup table
  defaults, the same as if the record '@.' were not present in the table;

- fixed ugly text formatting in recipient notifications template;
  reported by Florian Effenberger;

- fixed parsing of 'Received:' header field for some unusual cases;

- updated amavisd-new-courier.patch to apply cleanly against -P6 (now P7);
  by Martin Orr;

- added av_scanners entry for 'AVG Anti-Virus', kindly provided by
  Grisoft s.r.o. from Czech Republic,

- make Dr.Web Daemon av entry work with evaluation or regular license;
  thanks to Andrew I Baznikin;

- added av_scanners entry for new AVP client (aveclient) that is shipped
  whith avp 5.x.x.x.  Moved kavscanner to the @av_scanners_backup list
  (it is presumably slow and less reliable than aveclient), updated other
  AVP/Kaspersky entries; thanks to Nabil Sefrioui;

- added status codes 10 and 15 to the list of ok statuses for kavscanner
  (10=Password-protected archives, 15=Corrupted files);

- add an example dummy virus scanner 'all-clean' to the @av_scanners_backup
  list which always succeeds, always returning false (= status clean).
  Uncomment it if desired to avoid mail requeue when all other scanners fail,
  and to just pass the mail unchecked;

- disregard rfc2821 recommendation that 552 smtp response code
  should be treated as 452. It is unnecessary in amavisd-new setups,
  and it is wrong because 552 has other meanings assigned to it
  besides "too many recipients";

- allow IP address in $forward_method and $notify_method to be bracketed,
  which is needed for IPv6 addresses containing colons. Both formats
  are now allowed: 'smtp:' and 'smtp:[]:10025',
  the later is now preferred;

- change IP address bracketing in log entries 'SEND via SMTP' and
  'FWD via SMTP' from [] into []:10025 to match
  the syntax in the connect log entry and to facilitate parsing of IPv6
  addresses from the log;

- update received_line() to generate valid Received header field even for
  IPv6 client addresses;

- modified helper program amavis.c to allow it to be run non-root and to
  set temp file mode for group accessibility. Change its default log level
  to DBG_WARN;

- fix an incorrect SELECT example in README.lookups; thanks to Nabil Sefrioui;

- documentation updates;

                                                          November 10, 2003
Patch: amavisd-new-20030616-p6

- change SQL lookup code to better handle SQL database server restarts.
  After a SQL server restart amavisd-new would previously TEMPFAIL (4xx)
  all messages until amavisd child process would run down, then resume
  normal operation with the new child birth. Now the SQL server reconnect
  is done when the next mail arrives, so only one mail with each amavisd
  child process TEMPFAILs during SQL server restart, and normal operation
  resumes faster;

- in flatten_and_tidy_dir and in do_ascii: sanitize protection of files and
  directories which may otherwise be made inaccessible to virus scanners;
  based on patch by Henrique de Moraes Holschuh (Debian amavisd-new support);
  problem reported by Aspa and by Tomasz Papszun;

- fix a potential security problem: don't let rmdir_recursively and
  rmdir_flat follow symbolic links; this might be exploited by attempting
  to delete some foreign file using privileges of the amavisd process
  (which should not be root);

- use cpio option -d when unpacking cpio archives;

- don't insert spam-related headers for passed mail to recipients with
  bypass_spam_checks, making them believe the mail was not spam-checked
  (as they are not expecting such headers anyway);

- text added to the banned mail sender notification template (at the end
  of file amavisd), explaining to the sender what happened and how to avoid
  the restriction; edit to will;

- amavis-milter.c program in the helper-progs subdirectory is now based
  on the most recent version from the AMaViS CVS (maintained
  by Lars Hecking), but hacked a bit to make its options mostly compatible
  with the previous version. Start it with option -h to get current usage text.
  See helper-progs/README if using sendmail milter setup;

  Please revert to the old one if the .40 gives you trouble;
  both version are included in the helper-progs subdirectory.

- fix parsing of unrar info lines;

- consistency with other virus/banned logic: don't send recipient notification
  (reporting banned name) if mail contains both a virus and a banned name,
  but $warnvirusrecip is false; reported by Nathan G. Grennan and Urska Brinar;

- check for possible I/O errors when reading from SMTP socket,
  and distinguish error condition from normal TCP session teardown;

- do not redirect stderr to /dev/null when calling file(1). This way the
  diagnostics from file(1) will now at least be visible in the debug session;

- determine_file_types: make 'bzip compressed' pattern match older
  bzip format (v1) as well; thanks to Davaeron;

- sub run_command: explicitly close STDERR before reopening it; this way
  the reporting of possible problems in each operation would be separate,
  and it seems to avoid a rare problem (open STDERR '/dev/null' failing)
  reported by Sam Hart;

- don't attempt to do lookups in regexp table $viruses_that_fake_sender_re
  if it is undefined; reported by Chris Paul;

- remove existing X-Spam-Score along with other X-Spam* header fields
  if spam scanning is enabled;

- do not skip inserting 'X-Spam-Status: No, hits=- ... WHITELISTED'
  for whitelisted senders which caused SA to be bypassed - artificially
  assume score is -10 for the purpose of comparing it to tag_level;
  reported by Mike Vanecek;

- explicitly qualify wblist.rid field in $sql_select_white_black_list
  just in case someone has a field 'rid' in the other table;

- fix showing the value of LC_TYPE environment variable in the log,
  show LC_CTYPE as well;

- SAVI-perl: remove MIME option from the default set of options.
  Even with recent versions of Sophos SAVI there are cases where the
  library goes into a spin while trying to decode broken MIME message
  (same applies to Sophie - one may want to change its configuration);

- updated vfind entry in @av_scanners to work with the new version,
  changing '--vexit {}' to '--vexit {}/*'; thanks to Lowell Filak;

- updated kavdaemon entry in amavisd.conf; thanks to Michael Hall and
  Daniel Melanchthon;

- added option '-packed' to 'FRISK F-Prot Daemon' entry;
  thanks to Manfred Gloiber;

- NAI uvscan entry: commented pre/post actions in the entry to show
  how to remove environment variable LD_PRELOAD after finishing,
  suggested by David Tilley; added option --mime, suggested by Max and
  Kevin W. Gagel; added note on how to treat password-protected files
  as viruses, by Seth Parker;

- added 'Dr.Web Daemon' entry which talks directly to Dr.Web daemon over
  its Unix socket, speeding up a single-file check more than 200 times;
  provided by Andrew I Baznikin;

- another entry for Symantec AntiVirus Scan Engine provided by Guido
  R. Rolon A.; I'm not sure which is which, check your documentation;

- added 'dumaru', 'parite', 'gibe' and some other virus names to

- add examples to $viruses_that_fake_sender_re to show how to make
  a default result true, and only list exceptions;

- placed a comment in amavisd.conf pointing out the proper syntax
  for $hdrfrom_notify_* variables; thanks to Wouter de Jong;

- bump up the size of $sa_mail_body_size_limit to 150 kB in amavisd.conf

- documentation updates; fixed typos and spelling mistakes in the
  documentation files and in amavisd.conf comments;

- new documentation file README.protocol, specifying the new (and the old)
  protocol between helper programs and amavisd daemon, to be made available
  with the next major release. The description of the current (traditional)
  protocol was contributed by Stephane Lentz.

                                                            August 25, 2003
Patch: amavisd-new-20030616-p5

- fix 'Modification of non-creatable array value attempted' bug when
  no 'Received' header field was present in an infected mail;
  reported by Paul Miner;

- caching of SQL lookups on white/blacklist was based on sender address only,
  instead of sender _and_ recipient. This could lead to white-/blacklisting
  of one recipient to affect other recipients of the same message;
  reported and debugged by Paul Gamble;

- added LDFLAGS to helper-progs/ It is needed at least on NetBSD.
  Patch by Julian C. Dunn (the NetBSD package maintainer for amavisd-new);

- more obvious logging of HOLD reason in sendmail/milter setup
  (reported by Pascal Martinez);

- $MAXLEVELS zero or undef should disable the limit according to docs,
  but was not honoured; reported by Rob Hutton;

- if notifications delivery encounters a temporary failure (4xx), propagate
  this status to the final result instead of only logging a warning;

- amavisd.conf:
  * VirusBuster entry changed to match newer version of the product;
    information from Marcus Schopen;
  * another entry in amavisd.conf for KasperskyLab kavscanner (v4.5?),
    contributed by Simone Marx;

                                                            August 12, 2003
Patch: amavisd-new-20030616-p4

- revert to using alarm() instead of Time::HiRes::alarm(). It is nonstandard
  to mix the two, and is causing problems on some operating systems
  (e.g. Solaris); thanks to Geoff Gibbs;

- rise log level for log entries on intentional mail drops in case a
  DSN (a bounce) should be sent, but will not be; such as on rejected
  bounces, viruses with forged names, and spam from mailing lists.
  The new log entries now say: '... Not sending DSN ...' and provide
  more information on the reason for dropping DSN;

- if sender is whitelisted, don't insert 'X-Spam-Flag: YES' header field,
  don't append spam address extension, and don't quarantine. This makes it
  less surprising, although previous behaviour was according to documentation.
  NOTE: the documentation still describes former behaviour, this needs
  to be fixed;

- report deaths of command line scanners and some external programs
  distinctly from normal exits with nonzero exit status;

- fix replacing * or {}/* in the pattern with actual file names, causing
  MkS_Vir (mks32), VFind, Dr.Web Daemon, and KasperskyLab aveclient to receive
  its file arguments glued together; based on patch by Rafael J. Wysocki;

- update 'Panda Antivirus for Linux' entry to work with new (and older)
  versions of pavcl; updated entry kindly provided by Panda Software;

- stop the timer after SMTP transaction is over to better behave in
  persistent SMTP/LMTP sessions; start the timer at the beginning of a
  SMTP transaction, in addition to restarting it when DATA mode is entered;

- sub mail_via_smtp_single: properly report SMTP response code when all
  recipients are rejected by MTA, instead of logging a 'mail_via_smtp:'
  without a value. The problem was commonly seen with the W32/Mimail-A
  virus which fakes an often invalid local sender address and gets rejected
  by Postfix outright; reported by Turgut Kalfaoglu;

- use SMTP response code 554 (instead of 550) for rejecting syntactically
  invalid header (according to rfc4409 (ex rfc2476));

- add am_id to SMTP response code generated by one_response_for_all()
  to make it easier for MTA log to be correlated with amavisd-new log;
  some cosmetic improvements in the generated SMTP response text;

- added 'Return-Path:' in notifications to make it more obvious to see
  envelope address from reports;

- the 'Message-ID:' in neutral DSN notifications template was inadvertently
  pushed into the DSN body;

- report undefined spam score in X-Spam-Status header field as 'hits=-'
  instead of 'hits=0.0' which can be misleading;

- indicate blacklisting in X-Spam-Status header field of quarantined messages;

- add X-Envelope-From header field to quarantined messages;

- added virus names: tanatos, lentin, bridex  (alternative names
  for bugbear, yaha and braid) to the $viruses_that_fake_sender_re;
  thanks to Harrie Overdijk;

- set environment variables LINES and COLUMNS to sensible defaults
  to avoid some external program get puzzled about the terminal settings
  (e.g. older versions of pavcl from Panda Software);

- another attempt at fixing the Subject header field duplication. The patch
  amavisd-new-20030314-p2 fixed the case of entirely missing Subject header
  field, but did not fix the case of Subject header field present but with
  an empty text. Reported by Steven Cobb and Francois Rolland;

- rewritten 'Received' header fields parsing to better cope with valid,
  as well as with more common cases of broken syntax; used when trying
  to report originator IP address for believed-to-be-faked senders;

- more permissive parsing of SMTP addresses and options on MAIL FROM
  and RCPT TO commands;

- 'neutral' (=space) field in SQL black/whitelists now terminates the lookup
  search, avoiding fallback to static black/whitelists. It enables recipient
  to explicitly express its neutral stance towards the sender, overruling the
  site default;

- $sa_mail_body_size_limit now takes into account some portion of the
  mail header size so that huge mail headers that can cause slow SA calls
  are avoided (such degenerate cases were reported by Ralf Hildebrandt);

- taint fix in read_l10n_templates (as used by the Debian distribution),
  patch by Henrique de Moraes Holschuh;

- don't send recipient notifications to recipients that have
  bypass_virus_checks/bypass_banned_checks; suggested by Joe Breu;

- replace /bin/false with a more usual /usr/bin/false as a last resort exit;

- fetch modules 'Net::Ping' and 'bytes', which seem to be needed in certain
  chrooted setups;

- change log level from 0 to 1 for the log entry 'BAD HEADER from';
  reported by Thomas Lamy via Debian bug reports;
  same for 'unrar: all %d members are encrypted';

- fix typo in variable name: $spamassasin_obj -> $spamassassin_obj

- explicitly set pipes and sockets to binmode, as is a default since Perl 5.8.1

- documentation updates:
  * new file LDAP.schema, by Jacques Supcik, PhD
  * updated README.chroot to tell that /dev/urandom is needed
    in chroot jail (otherwise creation of MIME notifications fails);
    thanks to Lynn Duerksen and Jimmy Porter;
  * updated README.sendmail-dual, thanks to Robert LeBlanc and Stephane Lentz;
  * added URLs of external programs to INSTALL;
  * small updates to other doc files;

                                                              June 28, 2003
Patch: amavisd-new-20030616-p3

- avoid 'savemail: cannot save rejected email anywhere' sendmail panic when
  feeding mail via LMTP and using D_BOUNCE settings in dual-sendmail setup.

                                                              June 27, 2003
Patch: amavisd-new-20030616-p2

- when running chrooted, fix pre-loading of modules needed by SpamAssassin
  and Razor agents, which SpamAssassin forgets to pre-load by itself;
  reported by Neil Camara;

- updated KasperskyLab AVPDaemonClient entry to protect against
  carriage returns in collected virus names; thanks to Harrie Overdijk;

- updated 'Symantec CommandLineScanner' entry to look also for the new
  scanner name 'savsecls'; thanks to Guido Rolon;

- documented extended uses of read_hash() in the release notes;

- updated README.chroot;

                                                              June 24, 2003
Patch: amavisd-new-20030616-p1

- bug fix: allow stderr to be joined with stdout (>&1), instead of creating
  a file with a name "&1" when calling command-line scanners or doing
  LHA decoding (it got broken in 20030616); noticed by Henrik Larsson;

- bug fix: the X-Spam-Level header field would carry 64 asterisks instead
  of none when spam is configured to pass (mail tagging only) and when SA was
  not called (e.g. sender is whitelisted, ...) and tag_level or tag2_level
  is set to 0 or less; reported by Ralf Hildebrandt;

- bug fix: untaint e-mail addresses when forwarding via pipe;
  reported by Sam Tilders;

- modify code to match documentation: $relayhost_is_client should influence
  $notify_method too, not just $forward_method (adding an extra argument to
  mail_dispatch() was necessary to make this possible); reported by Zhu Yicun;

- SQL failure modes consistency: when initial connect to SQL failed, previous
  versions of amavisd-new would fall back to static defaults, disabling
  SQL lookups. Now initial connect failure is fatal, just like if connect
  failure occurred during operation - mail flow just stops during SQL servers
  outage, and is resumed when servers become reachable again;

- commented-out adding of 'X-Spam-Report' header field to retain behaviour
  from amavisd-new-20030314 - I find this header too long and intrusive
  for regular use. Uncomment it if adding X-Spam-Report is desired;

- support lzop (.lzo) and freeze (.F) compressors if available;

- use external program cpio (if available) to efficiently and safely handle
  the following archive formats: cpio binary, HPUX binary cpio, cpio crc,
    old ASCII cpio, new ASCII cpio, and HPUX old ASCII cpio, as well as
    POSIX.1 tar (also GNU tar) and old tar if allowed (see below)
  without reading whole archive members into memory;

- if your cpio(1L) can read tar format (as is common on FreeBSD and Linux),
  it is recommended to uncomment the following line (in file 'amavisd'):

    # /^\.tar$/  && defined $cpio && return do_cpio($part,$tempdir);

  which will cause tar archives to be safely and efficiently decoded
  by cpio(1L) instead of Archive::Tar, which loads the whole archive
  into memory;

- use separate file(1) 'classifications' ".uue" and ".hqx" (instead of ".asc")
  for uuencoded and binhex formats;

- merge 'H+BEDV AntiVir' and 'CentralCommand Vexira Antivirus'
  entries into a single entry - they are basically the same product
  (the Vexira entry didn't work for new version of Vexira);
  thanks to Vivek Khera;

- do_unrar: some switches to rar/unrar are not recognized by older versions
  of rar/unrar; issue a more informative log when this happens, and retry
  without newer switches for compatibility; thanks to Geoff Gibbs;

- crop backtrace on SA timeouts at 980 characters;

- preload Perl modules Net::DNS::* if spam scanning is enabled; they could not
  be loaded if running in a chroot jail, so SpamAssassin did not use Razor;

- documentation: collected documentation pieces pertaining to LDAP
  into README.lookups; fix SQL example table data in README.lookups;

( the patched MIME-tools by David F. Skoll is recommended over 5.411,
  as it better handles broken/bad MIME syntax: -> Download section )

                                                              June 16, 2003
amavisd-new-20030616 release notes

This is mostly a maintenance release. Only a handful of new features
were added that were small or easy enough to provide.

- revised the incoming SMTP session abort procedure to properly shut down
  the remote SMTP client in case of unexpected trouble (e.g. disk full)
  during SMTP data reception;

- subroutine mail_via_smtp_single() overhaul to produce cleaner diagnostics
  and to properly abort outgoing SMTP session in case timeout occurs
  in SMTP 'DATA' mode (avoids recipients repeatedly receiving partial
  message in successive delivery attempts when the receiving MTA responds
  very slowly); problem reported by Scott Vintinner;

- fixed the generation of LMTP response code when D_BOUNCE was chosen,
  but DSN was not actually sent because sender was thought to be faked
  or mail came in from a mailing list. Since DSN was not really sent,
  the final response remained at 5xx (causing bounce by MTA) instead of
  being converted to 2xx as D_BOUNCE would suggests; reported by Peter Bates;

- security: avoid inadvertently untainting values (e.g. mail addresses)
  at several places throughout the program;

- work around a Perl 5.8.0 taint bug where global variable $1 could become
  tainted, making further untainting attempts unsuccessful;

- work around another Perl 5.8.0 taint bug where Encode::encode fills up
  all available memory when given a tainted string with a non-encodable
  character. New subroutine safe_encode() provides a wrapper around
  the Encode::encode;

- treat 'Maximum number of files exceeded' the same as 'Exceeded storage
  quota', i.e. inserting the X-Amavis-Hold header field; noted by
  Christopher Odenbach;

- call exec with explicitly specified program path, preventing exec
  from even considering calling a shell;

- catch and better report the Perl's failure to fork on open;

- avoid inappropriately reporting 'Illegal seek' in run_av (and elsewhere)
  even though nothing went wrong; reported by William Yodlowsky;

- new subroutine flatten_and_tidy_dir() recursively descends into a directory
  containing potentially unsafe files with unpredictable names, soft links,
  etc., rename each regular nonempty file giving it a generated name, and
  discard all the rest.

- do_unrar and do_unarj now let the archiver itself recursively unpack
  the archive and then use flatten_and_tidy_dir() to tidy up the result.
  With previous method it was not possible to extract archive members
  with names containing non-ASCII characters from rar archives due to
  a bug in unrar and rar programs. Problem reported by Pan Bambaryla.
  Protection from mail bombs in do_unrar is retained.

- do_unrar: use program rar or unrar, whichever is available. Both
  rar and unrar recognize the same options to extract. (amavisd.conf:
  $unrar=['rar','unrar'] ). Under some circumstances unrar falls in a
  loop while rar extracts the archive correctly.

- do_unrar: don't bail out on exit status 1 (warning) from unrar after
  collecting file names from the archive. Be more careful when extracting
  file names, such as archive members with names starting with '-' or spaces;
  Disable showing archive comment and file comments which can easily
  break member-name parsing. Avoid extracting files in subdirectories
  twice: once when directory itself is listed, and the second time when
  each file in a directory is listed. Licence note: unrar is free,
  rar is not ( see for fresh unrar sources ).

- do_unarj: prefer more versatile arj over unarj (amavisd.conf:
  $unarj=['arj','unarj'] ). Both programs support subcommand 'e' that we use;
  suggested by Guillem Jover (via Debian support);  ARJ since 2.78/3.10
  is Open Source (GPL license). When arj is available (recognizing several
  new archive types, archive versions and options), put these options
  to good use, recovering several archive members that are unaccessible
  by the demo program unarj, such as old member versions, memebers with
  equal names, etc.  Disable showing comments.

- add X-Virus-Scanned header only if a mail was actually scanned
  for viruses (i.e. av scan not bypassed for this set of recipients);
  pointed out by Phil Regnauld;

- updated entry for 'KasperskyLab AvpTeamDream' virus scanner;
  by Daniel Melanchthon;

- truncate instead of round the spam level when producing the bar
  in the X-Spam-Level header field for compatibility with SA.
  Clip the level bar at 64 characters instead of 60 (more would not
  be allowed by RFC 2822, although SA used to crop at 100);

- subroutine hdr() incorrectly fixed an illegal header field body
  which did not have a space after a line fold; noticed by Virna Gupta;
  (the buglet did not show unless one modified the program);

- replace several remaining regular expressions /...$/ with /...$(?!\n)/
  as a matter of principle (as was done in amavisd-new-20030314-p2 for the
  more urgent cases). The use of simple $ was almost always subtly
  semantically wrong, but fortunately in most cases without consequences;

- specify open mode explicitly in several 'open's, and protect special
  characters in file names from being 'too cleverly' interpreted by
  Perl open (still not all cases done);

- SpamAssassin changes umask to 0077 - restore our mask after the SA call;

- be a bit more careful regarding platform-independency, distinguishing
  between use of \012 and \015 against \r and \n (note the \n on Mac is CR);

- change log level of several less important av scanner log messages
  from 2 to 3;

- small documentation updates in amavisd.conf;

- updated README.lookups: the new substring capture and reference mechanism
  in regexp lookups explained. The SQL example modified to work under
  PostgreSQL, thanks to Phil Regnauld;

- updated documentation: README.sendmail-dual, README.postfix, README.chroot,
  new file: README.exim_v4_app2 - by Louis Erickson;


- provide a command line option 'debug-sa', which is equivalent
  (but more convenient) then setting $sa_debug to true and starting
  as '# amavisd foreground';

- list version numbers of the more important Perl modules at startup;

- lookup_re(): RE pattern now allows for capturing of parenthesized substrings,
  which can then be referenced from the result string using the $1, $2, ...
  notation, as with the Perl m// operator. The number after the $ may be
  a multi-digit number. To avoid possible ambiguity the ${n} or $(n) form
  may be used. Substring numbering starts with 1. Nonexistent references
  evaluate to empty strings. If any substitution is done, the result inherits
  the taintedness of the key. Keep in mind that the $ character needs to be
  backslash-quoted in qq() strings (but not in q() strings). Example:
    $virus_quarantine_to = new_RE(
      [ qr'^(.*)@example\.com$'i => 'virus-${1}' ],
      [ qr'^(.*)(@[^@]*)?$'i     => 'virus-${1}${2}' ] );

- read_hash() now returns the hashref (first argument) also as a return
  value, making it easier to create new hashes (instead of only adding values
  to existing ones). Also the first argument (hashref) is now optional,
  and if missing a new hash is created. The following three cases are now

  a) $sa_tag2_level_deflt = {};
     read_hash($sa_tag2_level_deflt, '/var/amavis/tag2_levels.dat');

  b) $sa_tag2_level_deflt = read_hash({}, '/var/amavis/tag2_levels.dat');

  c) $sa_tag2_level_deflt = read_hash('/var/amavis/tag2_levels.dat');

- extend the semantics of the configuration variable $inet_socket_port,
  which can now either be a scalar as before (i.e. a single port number),
  or a ref to a list of port numbers, e.g:
    $inet_socket_port=[10024,10026,10028]; # accept SMTP on all these TCP port

  When $relayhost_is_client is true, the semantics of $forward_method has
  changed slightly: instead of taking the port number from $forward_method,
  it is now calculated as being one higher than the port number on which
  the incoming SMTP connection came in (one from the $inet_socket_port list).

  This allows for multiple MTA pairs on the same host to share the same
  amavisd daemon without having to use multiple loopback interfaces.
  Useful for example if incoming and outgoing mail is handled by separate
  mailers on the same host.

- add field 'spam_quarantine_to' to SQL table 'policy' - a patch
  provided by Vivek Khera. For compatibility a missing field defaults
  to undef, causing lookup search to proceed with the next lookup map;

- add fields: spam_lover, banned_files_lover, bad_header_lover,
  bypass_banned_checks and bypass_header_checks to SQL table 'policy'
  for completeness;  For compatibility a missing field defaults to undef,
  causing lookup search to proceed with the next lookup map;

- add a 'SPAM-TAG, ...' log entry at log level 2, which is produced when
  inserting spam-related header fields for each cluster of recipients
  with same settings. It complements the 'SPAM, ...' log entry, which
  is triggered at kill level. Feel free to adjust their log levels
  if you think one or the other is redundant;

- insert a 'X-Spam-Report' header field at tag2 level when spam is passed.
  Feel free to comment out the ...append_header('X-Spam-Report'... line
  if this is undesired.

- fold-in the 'bad_headers.patch' to the main code
  (available in the distribution since amavisd-new-20030314-p1),
  but renamed *bad_headers_lovers* to *bad_header_lovers* and added
  $final_bad_header_destiny and *bypass_header_checks*.

  To refresh memory:

  Enables checking headers of each mail for invalid (non-encoded) 8-bit
  characters, and produces a bounce (non-delivery status notification,
  or a SMTP REJECT if desired) with the full explanation of the problem,
  with offending header fields trimmed, sanitized and included in the text.

  New setting $final_bad_header_destiny, defaults to D_PASS for backwards
  compatibility, but a value D_BOUNCE is suggested. Similarly to other
  *lovers* settings, a hash and an ACL lookup %bad_header_lovers and
  \@bad_header_lovers_acl are available, and a setting $warnbannedsender.

  Certain recipients may be exempt from the checking (*bad_header_lovers*).
  Similarly mail from mailing lists (Precedence: list or bulk), and mail
  with null reverse-path mail (e.g. bounces) is passed, even if violating
  the RFC 2822 header syntax. A log entry is produced nevertheless.

  Postfix users: this is similar to the Postfix strict_7bit_headers=yes
  functionality, but produces a much more informative problem report
  (non-delivery notification) to the sender. One difference is that
  amavisd-new header check takes into account only the RFC 2822 header,
  not MIME headers in the mail body. It is reasonably efficient to use
  amavisd-new as a header checker only, without any anti-virus or anti-spam
  checks, if desired.

- added configuration settings $warnbannedsender and $warnbannedrecip,
  separating this function from $warnvirussender and $warnvirusrecip,
  which previously applied to both virus and banned files;

- new macro %c which evaluates to spam level/hits (mnemonic: sCore)
  as provided by SpamAssassin; useful in $log_templ;

- new configuration settings *bypass_banned_checks* and *bypass_header_checks*;

- new parameter $spam_quarantine_bysender_to (in contrast to the usual
  $spam_quarantine_to) makes possible collecting quarantined spam
  by sender names or domains, instead of (or in addition to) the more
  usual by-recipient map;

- new configuration setting $remove_existing_spam_headers (defaults to true);

- bring-in the localization (l10n) support contribution from Debian Linux,
  by Henrique de Moraes Holschuh; see README.l10n, and comments at the
  call to read_l10n_templates() in file amavisd.conf;

- supply X-Envelope-From and X-Envelope-To header fields containing envelope
  addresses to the messaged passed to SpamAssassin for checking;

- there is now an official OID:
  under which LDAP schema for amavisd-new LDAP lookups is to be defined
  ( 'ijs.amavisd-new.ldap' );

                                                                May 6, 2003
Patch: amavisd-new-20030314-p2

- blacklisted sender only caused spam headers to be inserted,
  but did not cause rejecting the message and other 'evasive actions'
  due to over-optimization; reported by Lawrence Farr, Robin Elfrink
  and Eric Vollmer;

- fix improper evaluation of bypass_spam_checks* in multi-recipient mail
  where some recipients match 'bypass_spam_checks*' and others do not.
  Spam-check was bypassed if _any_ (instead of _all_) recipient matched
  'bypass_spam_checks*'; thanks to Joseph W. Breu;
  (it got broken in amavisd-new-20030314, the amavisd-new-20021227 was ok);

- adding address extensions was not done for *_lovers, only for
  *_destiny=D_PASS; reported by Steve Khoo. The "deal_with_spam" section
  needed to be overhauled for this reason. For consistency with the
  documentation and with the change in adding address extensions,
  quarantining is now done if mail is considered spam, regardless of
  it being delivered (e.g. to spam lovers), or not;

- if only spam checking is performed, but no virus scanning nor banned
  file names checking is enabled, and the spam is detected but configured
  to pass, and the Subject is configured to be edited, the Subject header
  field was duplicated because the existing Subject header field was not

  Solution: make MIME entity available even if virus and banned files
  scanning is not performed. Header information is needed by
  add_forwarding_header_edits_per_recip() when deciding whether to add
  or replace the Subject header field in case spam is passed.

  (workaround for previous versions: don't leave $banned_filename_re
  undefined. If not needed, set it to an empty list: new_RE() );

- when quarantining to a mbox file, excape 'From ' with '>' only if it
  follows an empty line, as sendmail mail.local man page prescribes;

- added parameter $notify_xmailer_header to control the body of the header
  field X-Mailer as placed in the notification messages (e.g. bounces)
  generated by amavisd-new; by default this header is no longer inserted,
  as it was felt it revealed too much information. Choose from:
    $notify_xmailer_header = undef;  # no X-Mailer inserted (default)
    $notify_xmailer_header = '';     # X-Mailer: MIME-tools (Entity
    $notify_xmailer_header = 'your-text-here';  # X-Mailer: your-text-here

- replace regular expressions /...$/ with /...$(?!\n)/ or with string
  equality operator in some places (many more to come in the next major
  release); the use of simple $ was almost always subtly semantically
  wrong, but in most cases without consequences;

- enable timeouts after SMTP connection has been established, to be able
  to handle cases where MTA is very slow in accepting checked mail,
  which could result in multiple deliveries if MTA on the input side
  times out;

- debugging aids: call stack backtrace is logged if
  Mail::SpamAssassin::NoMailAudit::check exceeds allowed time;

- LDAP enhancement: allow "hostname" to be a string or a reference to an
  array, in which case each entry will be tried in order until a connection
  is made; by Jacques Supcik;

- X-Spam-Status now says WHITELISTED if sender was whitelisted;
  contributed by Joseph W. Breu;

- if mail rejection was due to sender blacklisting, say so in the SMTP
  response; thanks to Robin Elfrink;

- updated file bad_headers.patch to apply cleanly against this version.
  The bad_headers.patch will be folded-in with the next major release,
  please try it out, comments welcome. For patch description see release
  notes for amavisd-new-20030314-p1;

- specify option 'ExactAddresses=>1' in a call to Net::SMTP::new()
  in anticipation for the next release of Net::SMTP;

- amavisd.conf: updated documentation / comments;

- amavisd.conf: removed option '-ni' in the DrWeb virus scanner entry;
  thanks to Mike Boev and the FreeBSD community
  ( )

- amavisd.conf: added entry for 'MkS_Vir daemon', thanks to Dariusz Grzegorski;

- amavisd.conf: updated entry for 'F-Secure Antivirus' to cope with fsav 4.50
  as well as with 4.1x; solution by Juhan Tamsalu;

- documentation: updated README.chroot; new README.sendmail describing
  dual-MTA sendmail setup, small updates to other doc files;

                                                             March 21, 2003
Patch: amavisd-new-20030314-p1

- fix passing of remote MTA reject code back to the input side
  (DSN was not sent if MTA rejected the message);

- fix broken logic which allowed a virus to pass if a recipient was a
  'banned file lover' and the same message also contained a 'banned file';
  (or the other way around: pass a banned file if containing a virus and sent
  to virus lovers); thanks to Ortwin Gentz, I'm glad somebody is looking
  over my shoulder;

- remove a log entry:  one_response_for_all: SHOULDN'T HAPPEN
  The test was inappropriate, but harmless;

- enhanced sub run_av() to fix a problem with MkS_Vir virus scanner,
  where a '*' in the template command was not expanded to actual
  file names; reported by Pan Bambaryla;

- trim virus scanner output (in sub run_av) to a manageable size
  of 900 characters so as not to clutter log entries and notifications
  too much; provoked by the Panda pavcl;

- move the check for 'Precedence:' mail header field from
  sub delivery_status_notification() to its caller for flexibility;

- replaced the text in notification templates:
    The message WAS delivered to:
    The message WILL BE delivered to:
  to reflect the truth. The message may still be rejected
  by the MTA when forwarding takes place.

- small documentations edits/updates;

- cosmetic: trim trailing whitespace in the program;

- cosmetic: make shorthand subroutine nf() locally scoped; by Jacques Supcik;

- make available an experimental patch (file: bad_headers.patch,
  not applied by default), which, when applied (with patch(1) utility)
  to amavisd-new, makes it check headers of each mail for invalid
  (non-encoded) 8-bit characters, and produces a bounce (non-delivery
  status notification, or a SMTP REJECT if desired) with the full
  explanation of the problem, with offending header fields trimmed,
  sanitized and included in the text.

  Certain recipients may be exempt from the checking (*bad_header_lovers*).
  Similarly mail from mailing lists (Precedence: list or bulk), and mail
  with null reverse-path mail (e.g. bounces) is passed, even if violating
  the RFC 2822 header syntax. A log entry is produced nevertheless.

  Postfix users: this is similar to the Postfix strict_7bit_headers=yes
  functionality, but produces a much more informative problem report
  (non-delivery notification) to the sender. One difference is that
  amavisd-new header check takes into account only the RFC 2822 header,
  not MIME headers in the mail body. It is reasonably efficient to use
  amavisd-new as a header checker only, without any anti-virus or anti-spam
  checks, if desired.

  The patch is experimental in the sense that it may change in future
  versions, but is fully functional when applied;

- Jacques Supcik cleaned up and generalized his LDAP lookups
  support code (which is off by default). In his words:

  The major change is in the definition of ldap lookups.
  In the config file, you have to enable ldap first:

  | $enable_ldap = 1;

  Then you can define defaults for your ldap queries:

  | $default_ldap = {
  |   hostname => 'localhost', tls => 0,
  |   base => 'ou=hosting,dc=example,dc=com', scope => 'sub',
  |   query_filter => '(&(objectClass=amavisAccount)(mail=%m))'
  | };

  And then the lookups themselves:

  | $virus_lovers_ldap = {res_attr => 'amavisVirusLover'};
  | $banned_files_lovers_ldap = {res_attr => 'amavisBannedFilesLover'};
  | ...

  With this method, you can define every parameter individually.
  You could have a different ldap server for each lookup! Like that,
  the configuration is closer to the one in postfix or courier.

  The hashes are converted into lookups objects in the
  Amavis::Lookup::LDAP class method.

                                                             March 14, 2003
amavisd-new-20030314 release notes


- per-user white- and blacklisting, including a SQL lookup mechanism;
  refined semantics for white/blacklists;

- easier to run in chroot jail, see README.chroot;

- can specify two sets of virus scanners: the main list and a backup list,
  which gets consulted only if all scanners from the main list fail;

- split REJECT into separate destiny settings: REJECT and BOUNCE,
  giving you a choice of who is responsible for sending a non-delivery
  notifications to the sender; see comments in amavisd.conf;

- edited configure file amavisd.conf, providing more informative comments,
  cleaner examples and simplified common settings;

- made it work better with the Unicode-aware Perl 5.8.x (not against it)
  and distinguish between character data / byte data / octet data (binary),
  provide UTF-8 (Unicode) support for notification messages,
  while avoiding some possible nasty surprises in the UTF-8 locale
  (which is enabled by default in Red Hat 8.0);

- SQL SELECT clause settable from the config file;

- can store quarantined or forwarded messages as BSMTP files (RFC 2442);

Changes since amavisd-new-20021227 patch level 2


- fixed caching problem when bypass_*_checks are used selectively: when only
  one of the virus/spam tests were performed but the other bypassed due to
  a per-recipient setting, the result of a non-performed test was inserted
  into cache as clean. This could cause tests to be bypassed for subsequent
  message with the same body but different recipients with different bypass*
  settings, but only within a lifetime of a child process (10 messages
  by default). Noticed by Jürgen Louis Fluk, thanks!

- changed caching of SQL lookups to match the documented behaviour,
  i.e. SQL lookup result is cached (for the benefit of field lookups)
  for a single message only, and no longer for the child lifetime.
  It is possible to revert to the previous behaviour by commenting-out
  one line (look for $sql->clear_cache). Pointed out by Paul Miner;

- see also further down about the field 'banned_files_lover'
  in the SQL table 'policy', and the UTF-8 (Unicode) workarounds.


- running amavisd-new in chroot jail is now easier to set up due to added
  configuration variable $MYHOME and its simplification effects on other
  settings in the config file. Relevant settings are better commented;

- avoid the need to have a shell in chroot jail (use of the new subroutine
  run_command() instead of Perl functions qx() and system() );

- new file README.chroot with guidelines on how to run amavisd-new
  in a chroot jail, to prevent possible vulnerabilities from affecting
  the rest of your system;

- added a security notice (web page and other places) that the Unix utility
  file(1) 3.39 and older contains an exploitable security vulnerability,
  which can cause system access with access rights of the user running
  amavisd-new daemon (not root, unless you violated the recommendations).
  Please upgrade file(1) to 3.41 or newer:

- new section (Security considerations) on the web page;


- improve explanation texts (comments) in amavisd.conf,
  edit the file to improve legibility and structure, change some examples.
  The daemon is thought to be compatible with amavisd.conf from previous
  version, but it may be worthwhile to switch to the new file to avoid
  asking FAQ;

- introduce a configuration variable $MYHOME, which is never used directly
  by amavisd daemon, but is a handy way of setting default values for
  other configurable variables in the config file. Also makes default
  settings more chroot-friendly.

- introduce a configuration variable $mydomain, which is never used directly
  by amavisd daemon, but is a handy way of setting default values for
  other configurable variables when there is only a single domain (with
  subdomains) behind the mailer. This avoids the need to edit half-a-dozen
  configuration variables in simple setups;

- moved debugging-related settings to a separate section
  (Section VIII - Debugging);

- make SQL SELECT clause settable from the config file;

- renamed variable @local_domains to @local_domains_acl for consistency;
  The old name became a synonym for @local_domains_acl, so compatibility
  with existing config files is retained;

- comment out the example: read_hash(\%whitelist_sender, ...);
  it slipped into the amavisd-new-20021227 released code from my tests;

- comment out the setting $relayhost_is_client=1;  It is only useful
  for multi-access daemon setup, and potentially confusing for others;


- introduced a list of backup virus scanners: @av_scanners_backup,
  complementing the normal list of virus scanners.

  If no virus scanners from the @av_scanners list produce 'clean' nor
  'infected' status (e.g. they all fail to run or the list is empty),
  then all scanners in the @av_scanners_backup list are tried.
  When there are both daemonized and command-line scanners available,
  it is customary to place slower command-line scanners in the
  @av_scanners_backup list;

- removed the Mac and SafeMacDfHandling options from SAVI-Perl,
  as they are no longer supported by Sophos library since 2003-03;
  set new option NamespaceSupport to SOPHOS_DOS_AND_MAC_FILES;
  thanks to Paul B. Henson;

- changed a default to initialize SAVI-Perl only at startup time,
  and no longer every time the child process is started. This is because
  starting with SAVI 3.0 (March 2003) the initialization is slower.

  NOTE: it is now necessary to restart amavisd master process
  after SAVI database is updated, to make it notice new virus descriptions.

- in 'MkS_Vir for Linux' entry replaced options -e -c with -s,
  to make mks work, suggested by Pan Bambaryla;

- added status 8 ('corrupted') to the list of clean statuses for
  'KasperskyLab AVPDaemonClient', as is already the case for some other
  virus scanners. See amavisd-new web page (Security considerations)
  if the idea bothers you;

- RAV command line scanner no longer supported off-the-shelf,
  as interfacing to it seems to be violating their license terms.

  Btw, if RAV command line scanner does not work as user amavis,
  check permissions and ownerships of /usr/local/rav8/* or reinstall
  RAV as the user that will run amavisd; thanks to Oyku Gencay;

- RAV does not like '</dev/null' (segfault) (reported by Johnny Ljunggren),
  most others don't care, but possibly Panda needs it. Tried to remove it,
  but later put is back, as I don't like external command to have access
  to the amavisd socket over which SMTP or AM.CL session is running;

- H+BEDV AntiVir: changed option -allfiles to --allfiles;
  thanks to Dr Patrick Atlas, confirmed in the H+BEDV documentation;

- comment out the default entry for Sophos sweep, as the name 'sweep'
  clashes with the Debian sweep package (audio samples editor).
  If you uncomment the entry, make sure the correct 'sweep' is found
  in the specified path; thanks to Guido Guenther and Brian May;

- added av definition for BitDefender (based on amavis-ng code by
  Hilko Bengen); thanks to Stephane Lentz for testing and suggesting
  command line options;

- added av definition for Ikarus AntiVirus for Linux, based on
  amavisd-0.1 code by Lars Hecking;


- make program aware of the type of data coming-in from external files,
  pipes and sockets - distinguish between binary data, byte data, and
  character data / plain text.

  This is necessary in order to be able to survive in the UTF-8 locale-aware
  Perl 5.8.0 environment (e.g. to avoid 'Malformed UTF-8 character'
  and 'Wide character in print' warnings/errors, among others).

- check character ranges in regular expressions throughout the program,
  and fix/rewrite them to avoid assumption that all characters are in the
  range 0..255, which is no longer true! Some nasty surprises could have
  been lurking in that corner.

  Note the Perl 5.8.0 terminology (see perluniintro and Encode man page):
  * character: a character in the range 0..(2**32-1) (or more)
    (what Perl's strings are made of);
  * byte: a character in the range 0..255
    (a special case of a Perl character);
  * octet: 8 bits of data, with ordinal values 0..255
    (term for bytes passed to or from a non-Perl context, e.g. a disk file)

- modified sanitize_str() and hdr() to cope with UTF-8 characters;

- modified string_to_mime_entity() and hdr() to encode RFC2047-permitted
  headers according to $hdr_encoding, and notification body according
  to $bdy_encoding settings (both default to iso-88591-1, aka Latin1).
  Notification template texts are supposed to be stored internally as
  native or UTF-8 Perl strings, which is handled (mostly) transparently
  by Perl 5.8.x;

- added option $hdr_encoding_qb to be able to explicitly specify
  MIME encoding ("Q" or "B") for $hdr_encoding MIME charset;

- rfc2822_timestamp: produce correct date format regardless of
  LC_TIME locale setting;

- read_text() subroutine (which may be used in amavisd.conf to read
  notification text templates) may be given a second argument, specifying
  character encoding used in the external file. If Perl is Unicode-aware
  (5.8.0), this argument is given to binmode to influence the reading
  of the file and its conversion into internal Perl string representation.

  When notification text is generated from the internal Perl string
  representations (UTF-8), the following settings influence the character
  encoding (charset in MIME terminology):

    $hdr_encoding = 'iso-8859-1';  # ... header field bodies

  The header is first encoded into $hdr_encoding, which is then encoded
  into 7-bit ASCII as required by RFC 2821 and defined by RFC 2047.
  The specified encoding (charset) will appear in the RFC 2047 Q-encoded
  strings in header fields where this is allowed, i.e. in Subject,
  Comments, and X-* header fields. In other header fields where RFC 2047
  encoding is not allowed, encoding into plain ASCII is used, where invalid
  characters are treated as Perl Encode module does it, i.e. by converting
  them to '?'.

    $bdy_encoding = 'iso-8859-1';  # ... notification body text

  Notification body text is encoded from the internal Perl representation
  according to $bdy_encoding. The encoding (charset) is also inserted
  in MIME header as Content-Type.charset.


- make it work with Net::Server version 0.85, which changed the handling
  of the CHLD signal, this is breaking qx and system calls used by previous
  versions of amavisd-new;

- with Net::Server 0.85 you may see a message:
    Net::Server: Couldn't POSIX::setuid to ... []
  If the process UID remained 0 (root), the program will terminate,
  otherwise just consider the message harmless (or complain to the
  author of Net::Server);

- temporarily removed utility amavisdconf, it is not yet updated to cope
  with the new configuration variables;

- determine_file_types: 'script text executable' is now considered .txt,
  not .exe; thanks to Stephane Lentz and Michael Dengler;

- report correct system error if Convert::UUlib::decode fails
  (e.g. permission denied); thanks to Nick Murtagh for figuring out
  that UUDecode() in uunconc.c tries to decode into a temporary file
  in /root/tmp/whatever;

- avoid the need for a Unix utility file(1) to support option -b;
  suggested by Clifton Royston considering OpenBSD;

- new macro %b for customized notifications and log: it evaluates to
  the message digest of mail body (MD5, hex).  Used as a caching key,
  can serve to identify messages in quarantine with same body text.

- provide a new BSMTP output method that can be used as $forward_method,
  $notify_method, $virus_quarantine_method or $spam_quarantine_method.
  It is simple in a sense that it generates one file for each message,
  so it is not terribly efficient. The parameter after the colon is used
  as a file name prototype, where %i in the prototype name is replaced
  by current time, %n by amavis internal message id, and %b by message
  digest of the body (MD5,hex). For example:

    $forward_method = "bsmtp:$TEMPBASE/out-%i-%n.bsmtp";
    $spam_quarantine_method = "bsmtp:$QUARANTINEDIR/spam-%b-%i-%n.bsmtp";

  The bsmtp file has ".tmp" appended to a name during writing, and is
  renamed to the final name when it is done, to avoid being seen by possible
  fetcher prematurely. It is experimental in a sense that configuration
  and implementation details may change in the future, including possible
  changes in the old 'local:' method;

- besides the default (implied) quarantine method 'local:', it is now possible
  to specify 'bsmtp:...' as a quarantine method, and have different settings
  for virus and spam quarantines. New settings: $virus_quarantine_method
  and $spam_quarantine_method, e.g.:
    $virus_quarantine_method = "bsmtp:$QUARANTINEDIR/virus-%i-%n.bsmtp";
    $spam_quarantine_method  = "bsmtp:$QUARANTINEDIR/spam-%b-%i-%n.bsmtp";

- if ***SPAM*** is to be inserted into Subject header field,
  but the original mail does not contain the Subject field, insert one;
  suggested by Christopher Odenbach;

- per-recipient SQL field and static flag (spam_modifies_subj) for modifying
  Subject: header field when spam is passed;

- updated README.lookups, describing new SQL fields and tables;

- split kill_level setting into two independent settings: tag2_level,
  and kill_level. A quick reference:
    tag_level  controls adding the X-Spam-Status and X-Spam-Level headers,
    tag2_level controls adding 'X-Spam-Flag: YES', and editing Subject,
    kill_level controls 'evasive actions' (reject, quarantine);
  The default is tag2_level=kill_level, which retains compatibility.

- work around a bug in Perl 5.8.0 in UTF-8 environment, which could
  produce error "Bad RFC822 field name 'C0c0\x{0}\x{0}T-Type'" when building
  notifications (the Perl bug is triggered in Mail::Header::_tag_case
  by statement $tag =~ s/\b([a-z]+)/\L\u$1/gio; reported on several
  Red Hat 8.0 hosts, seemingly only on Intel platforms).

  Many thanks to Chris Kloiber and Steve Sloan for making available
  their test machines for debugging.

- replace all calls to qx by run_command, to avoid 'Malformed UTF-8 character'
  when reading from external utilities that do not honour locale settings,
  e.g. Unix file(1) utility, which may return non-printable characters with
  codes above 127. This also ensures better error reporting.
  run_command() moved into Util module, part of its code moved into a
  new subroutine run_command_copy().

- improved diagnostics when certain external programs fail to run,
  such as when fork or exec fails;

- new LDAP lookups (experimental), by Jacques Supcik, PhD,
  IP-Plus Internet Services - Swisscom Enterprise Solutions Ltd,
  Genfergasse 14, 3050 Bern, Switzerland (
  Many thanks for the contribution!

- changed enhanced status code 5.7.0 to 5.7.1 when message is considered
  bad (virus, banned file, spam);

- when sending message via SMTP, don't go into DATA part if any RCPT TO
  returned 4xx code, since without implementing proper queueing we wouldn't
  be able to handle the situation unless message came in via LMTP;

- updated README.postfix to describe:
  * the now recommended Postfix 2.0 setup with pre-cleanup service,
    instead of the (post-)cleanup2;
  * how to specify a backup content filter(s) in case the primary fails;
  * the Postfix setting if you want X-Amavis-Hold header field to cause
    message to be placed on hold by Postfix, e.g. when amavisd-new encounters
    a mail bomb which can not be checked for viruses.

- unconditionally delete old headers X-Spam-Status, X-Spam-Level,
  X-Spam-Flag and X-Spam-Report, when anti-spam code is enabled.
  Previously deletion was done only when new headers were to be inserted.

- replace sloppy file accessibility tests (Perl functions -e, -f, -d, -r, ...)
  with more reliable method, which also produces more informative

- set $SIG{CHLD}='DEFAULT' in Net::Server hooks to make qx work
  in Net::Server::Single environment (useful for debugging),
  as well as with the version of 0.85 of Net::Server;

- adjust the log level in do_ascii() (reported by Pan Bambaryla),
  and in few other places;

- fix $log_templ - an unconditional comma was causing the log entry to be
  always issued, regardless of mail content; reported by Stephane Lentz;

- enhance reject/discard/pass destinies with reject/bounce/discard/pass,
  making it more flexible to specify what action to perform on detection
  of bad content; (thanks again to Chris Hastie for earlier discussion
  on the topic); see explanations in amavisd.conf;

- added hash lookup table %local_domains for those that have a great number
  of hosted domains and prefer to read a list from a separate file;

- a special shorthand provision when SQL lookups are used: when a match
  for recipient address or domain is found in SQL tables (regardless of
  field values), the recipient is considered local, regardless of static
  @local_comains_acl or %local_domains lookup tables. This simplifies
  life when a large number of dynamically changing domains is hosted.
  To overrule this behaviour, add an explicit boolean field 'local'
  to table 'users' (missing field defaults to true, meaning record presence
  implies locality).

- changed SQL table 'policy' field name 'banned_files_lover' to a name
  'banned_files_lovers' (added an 's') for consistency with variables
  %banned_files_lovers and @banned_files_lovers_acl, and updated

- SQL field treated as boolean now recognize '\0' character as false
  (not to be confused with NULL SQL field, which is still mapped to undef);

- when generating Received: header field, move receiving port number
  into the second comment field - some software were too picky
  about the syntax of the first comment field ('TCP-info' in rfc2821);

- consistency: apply late 'string sanitation'; meaning that strings
  such as virus names, banned attachment names, e-mail addresses, etc,
  are sanitized at the point where the string is leaving amavisd-new,
  not at the point where it enters is. This allows the quotation rules
  to depend on the output properties (log entry, notification text,
  SMTP response text, ...) / (ASCII, iso-8859-1, UTF-8, binary, ...);

- delegate responsibility to call sanitize_str to write_log,
  instead of having a caller do it;

- wrap long syslog entries at 980 characters (less than 1023 minus prefix);
  thanks to Ted Cabeen;

- skip spam-related actions if $spam_level is undefined (e.g. when
  anti-spam code is not enabled) so that missing $sa_kill_level_deflt
  setting in amavisd.conf won't cause a surprise. Thanks to Marcio Merlone.

- new version of amavis-milter.c based on amavis-milter.c
  (from amavisd-0.1);

- use the /m regexp modifier by default when parsing av-scanner output
  for clean and infected regular expressions, making it easier to specify
  patterns, and to make it like the expression which collects virus names;
  thanks to Daniel Hunziker for reporting a problem with Symantec
  CommandLine Scanner;

- refined the semantics of sender whitelisting/blacklisting,
  add per-recipient sender while/blacklisting mechanisms:


  WHITELISTING: use envelope sender lookups to ENSURE DELIVERY from whitelisted
  senders even if the message is recognized as spam. Effectively, for the
  specified senders, message recipients temporarily become 'spam_lovers', with
  further processing being the same as otherwise specified for spam lovers.
  It does not turn off inserting spam-related headers, if they are enabled.

  BLACKLISTING: messages from specified senders are DECLARED SPAM.
  Effectively, for messages from blacklisted senders, spam level
  is artificially pushed high, and the normal spam processing applies,
  resulting in 'X-Spam-Flag: YES', high 'X-Spam-Level' bar and other usual
  reactions to spam, including possible rejection. If the message nevertheless
  still passes (e.g. for spam loving recipients), it is tagged as BLACKLISTED
  in the 'X-Spam-Status' header field, but the reported spam value and
  set of tests in this report header field (if available from SpamAssassin,
  which may not have been called) is not adjusted.

  A sender may be both white- and blacklisted at the same time,
  settings are independent. For example, being both white- and blacklisted,
  message is delivered to recipients, but is tagged as spam.

  If ALL recipients of the message either white- or blacklist the sender,
  spam scanning (calling the SpamAssassin) is bypassed, saving on time.


  The same semantics as for global white/blacklisting applies, but this
  time each recipient (or its domain, or subdomain, ...) can be given
  an individual lookup table for matching senders. The per-recipient lookups
  override the global lookups, which serve as a fallback default.

  One can specify a two-level lookup table: the key for the outer table is
  recipient, and the result should be an inner lookup table (hash or ACL
  or RE), where the key used will be the sender.

  Per-recipient white/blacklisting is also available as a SQL lookup.
  See configuration variable $sql_select_white_black_list and README.lookups .

INCOMPATIBILITIES WITH amavisd-new-20021227

- the previous configuration file amavisd.conf from version
  amavisd-new-20021227 is usable with amavisd-new-20030314.
  Nevertheless it may be a good idea to adopt the new amavisd.conf,
  as it describes new options, is cleaner and better commented;

- reordered hash lookups from:,,,,, .com, . ,
    user+foo@, user@
    user+foo@, user@,,,, .com, .
  to make user@ still useful even when catchall ( . ) is used,
  and to bring it closer to the recommended SQL order;

  Not likely to affect existing setups, as previously the combination
  of user@ and catchall was not useful;

- SQL lookups:

  * incompatible change (avoiding unexpected behaviour): when the requested
    SQL field name does not exist in the SQL table, return a sensible default
    value for boolean fields (according to field name and its type),
    instead of 1 as previously. Logs a warning when no default is available.
    Thanks to Ortwin Gentz for noticing the problem;

  * new fields spam_tag2_level, spam_modifies_subj and local, with useful
    defaults if absent, so it is not necessary to immediately update an
    existing database;

  * changed SQL table 'policy' field name 'banned_files_lover' to a name
    'banned_files_lovers' (added an 's') for consistency with variables
    %banned_files_lovers and @banned_files_lovers_acl;

amavisd-new-20021227-p2 (2003-01-10):

- sender notification were not sent for messages containing banned
  filenames but no virus; fixed (thanks to Philip Ross);

- added virus name 'Sobig' to the default $viruses_that_fake_sender_re

- extended default $banned_filename_re to match also a longer
  (4 character) middle file extension (e.g. Movie_0074.mpeg.pif)

- add explicit path to the program vscan in the 'Trend Micro FileScanner'
  entry (amavisd.conf), thanks to Matthew Hobbs;

- improvements in the setup of helper programs (files,, config.h, amavis.c and amavis-milter.c), by Radu Greab;

- add Sophos error code 527 ('Not supported in this SAVI implementation')
  to the list of (mostly)harmless codes in the SAVI-Perl interface routine
  (thanks to Paul B. Henson and Tim Winders for discussion);

- initialize SAVI-Perl module (if available) in each new child, instead of
  only once during daemon startup time. This avoids the need to reload
  amavisd for every change in the Sophos IDE database - at the expense
  of additional SAVI-Perl startup time every $max_requests messages;

                                                          December 30, 2002

amavisd-new-20021227-p1 (2002-12-30):
  fix the problem when using multiple virus scanners, where the
  last scanner doesn't find a virus, but some other scanner does,
  and amavisd-new (incorrectly) does not report a virus.
  Thanks to Okke Timm, Henrik Larsson and Eduardas Paulavicius;

                                                          December 27, 2002
amavisd-new-20021227 release notes

amavisd-new-20021227 is primarily a maintenance release, consolidating
amavisd-new-20021116 with its mandatory patches, updating documentation,
but also brings a couple of niceties.

Changes since amavisd-new-20021116 patch level 5

- The new configure/makefile in the helper_progs subdirectory is
  contributed by Stephane Lentz, based on MIMEDefang
  It is needed to build helper programs for sendmail (milter and nonmilter)
  setups (but not needed for Postfix and Exim v4 setups, or any dual-MTA
  setup). Similar to the prior work done by Ramiro Morales - thanks to both!

- Updated documentation (README, INSTALL, RELEASE_NOTES, README.lookups,
  README.customize, README.postfix).

- new subdirectory test-messages, containing sample mail messages to test
  virus and spam detection - see README there

- For daemonized antivirus scanners (only) the semantics of few parameters
  in the @av_scanners list has been incompatibly changed and enhanced.
  The same internal mechanism now handles both Unix sockets and inet socket.
  More than one socket may now be specified, to be tried one after the other
  until one is successfully connected to (useful for 'FRISK F-Prot Daemon').
  Also the usual regexp pattern parameters in the @av_scanners are now
  observed by Sophie and Trophie interfaces as well, bringing them more
  in line with other scanners.

- The (?m) regexp flag in @av_scanners patterns is now turned-on
  by default, so it is no longer necessary to specify it in @av_scanners,
  and has been removed to shorten the expressions.

- The SAVI-Perl interface routine now turns on most Sophos engine options,
  enabling recognition of some more file formats.

- All active antivirus scanners are now called, the scan no longer stops
  when the first one finds a virus. This is on the average a very small
  price to pay (only payed in virus storms), but makes possible to see
  which virus scanners found the virus and which did not, and to get their
  timing in the logs. The list of virus scanners that noticed the virus
  is now included in the notification message to the virus administrator.
  A new macro %W is available for this purpose.

- More sensible logic when checking more than one file for those
  daemonized av scanners that need to be handed individual files instead
  of directory name.

- The 'Clam Antivirus-clamd' daemon is now supported, both via Unix
  socket, as well as via inet socket (but Unix socket should be used,
  due to lack of self-protection in this respect by clamd).

  clamd is a fast and open-source anti-virus scanner
  ( ), and I recommend it to be used in
  combination with other commercial virus scanners as the last resort
  scanner, to avoid bouncing mail in case primary scanner(s) stumble
  across some unhandled situation, like corrupted or encrypted mail.

- New utility program 'amavisdconf' to check the values of configurable
  variables set by /etc/amavisd.conf or defaulted (modeled after
  Postfix utility 'postconf', but -e option is not yet supported).

    amavisdconf [-d | -n] [-c conf-file] [variable]

   (none) show the value as will be used by amavisd;
    -d  show default value as in the absence of the config file;
    -n  only show variable if its value is different from the default;
    -c conf-file ... use the specified file instead of /etc/amavisd.conf

  If a variable is specified show only that variable, otherwise show all
  configurable variables. (Note: the leading $, @ or % must be included
  with the variable. Make sure to protect $ from shell evaluation,
  e.g. by enclosing variable name in single quotes, or prepending
  a backslash:
    $ amavisdconf -d '$forward_method'
    $ amavisdconf -n

  This is the first attempt at such a utility, some details would need
  to be polished (e.g. how to report code reference), but it is hoped
  to be usable already in its present form.

- Two new subroutines (read_hash and read_text) are now available
  for use from amavisd.conf file, with a primary intent to move larger
  texts out of the amavisd.conf file:

  * read_hash allows for lookup hashes (e.g. white/blacklist,
    static %spam_lovers table, ...) to be read from a file during startup;
    It takes quoted form of addresses, optionally enclosed in angle
    brackets, and converts them into unquoted (raw) form, as needed
    for the hash keys. Format: one address per line, everything from
    '#' to end of line is a comment, leading and trailing whitespace
    removed, empty (or comment-only) lines are ignored.

  * read_text allows for reading more complex strings (e.g. multi-line,
    or including special characters such as backslash) from a file.
    Useful to override the default template text for notification messages.

  See example usage in amavisd.conf.

- Three synonym (alias) subroutine names are available for use
  in amavisd.conf to shorten the expressions:

    new_RE       same as   Amavis::Lookup::RE->new
    ask_daemon   same as   Amavis::AV::ask_daemon
    sophos_savi  same as   Amavis::AV::sophos_savi

  amavisd.conf file is changed to use the short names, but full
  subroutine names are still valid.

- Provide protection mechanism for SMTP/LMTP server code to limit the
  number of recipients in a mail transaction to some sane value
  ($smtpd_recipient_limit, default 1000). Even if the number is exceeded,
  mail will still be normally delivered in more than one transaction
  (provided your MTA is rfc2821 compliant).

- SMTP client code now correctly handles the situation where remote
  SMTP server returns '452 Too many recipients'. When this happens the
  mail forwarding or submission is split into several SMTP transactions,
  and as long as the remote SMTP server is willing to accept at least
  one recipient per transaction, mail delivery is assured.

- Auto white-listing (AWL) by SpamAssassin may be turned on by
  setting $sa_auto_whitelist to true (contributed by Hamish Marson).
  This only works as one global database, the per-recipient AWL
  is presently not available.

- New configuration variables $hdrfrom_notify_sender, $hdrfrom_notify_admin,
  and $hdrfrom_notify_spamadmin, to remove the overloading (double semantics)
  from variables $mailfrom_notify_*. Now $mailfrom_notify_* control ONLY
  envelope addresses, while $hdrfrom_notify_* only controls From header
  in notifications. The default values for $hdrfrom_notify_* are chosen
  to preserve compatibility.

- The spam headers in quarantined message now reflect the lowest
  tag and kill levels when more than one recipient is given, instead
  of reporting the levels for the first specified recipient.

- Tested with SpamAssassin 2.50 and with its new Bayesian classification
  (used as a global database by amavisd-new). Given two sets of spam
  and non-spam (=ham) messages, one can build a database by using new
  SpamAssassin utilities sa-learn-spam and sa-learn-nonspam as user amavis,
  and the database built this way will be used by amavisd-new/SpamAssassin.

- Initialize SpamAssassin as $daemon_user (no longer as root),
  as a safety measure, and to ensure the files it creates are still
  accessible after the UID change.

Incompatible changes to 20021116:

- Different parameters in @av_scanners list for daemonized antivirus
  scanners (as described above).

- lookup_hash: an undefined hash value was previously treated as a
  special case, resulting in value true. This is no longer the case,
  which brings it in line with other lookup methods. It is not likely
  anyone will be affected, as this practice was only documented but
  not encouraged by any example.

Here is a brief description of patches that needed to be applied to
amavisd-new-20021116, and were available from the amavisd-new web page:

patch5 (2002-12-09):
- fix a simple test to decide if the anti-virus and anti-spam code is needed;
- use the /m regexp modifier by default when parsing av-scanner output
  for virus names - seems like (?m) in qr is ignored (have we found
  a Perl bug?);
- fix DSN to report original smtp response in case reporting MTA is remote
  (thanks to Chris Hastie for his keen eye in understanding DSN);
- fix reporting quarantine mailbox name(s) in case of per-recip quarantine
  or quarantine forwarding to MTA;
- macro %q (quarantine id) is now a list of quarantine mailbox names or
  addresses (may have more than one entry in case of per-user quarantines);
- macro %v (av-scanner output) is now a list of scanner output lines,
  to facilitate formatting of multi-line av-scanner output;
- fix spam admin notifications, which used the virus template instead of
  the spam template;
- fix problem of undefined entity causing spam notification failure
  in case other content checking (virus/banned) is not enabled;
- make SQL lookups (DBI access) optional - no DBI code is loaded if
  @lookup_sql_dsn of empty;
- support SAVI-Perl module 0.10 or later (direct access to the Sophos library
  engine) ( );
- add macro %j that evaluates to message 'Subject:' header field;
- fix minor SMTP protocol inconsistency when multiple MAIL FROM
  commands are issued by the client;
- more informative timing report when more than one av-scanner is used;
- make sendmail initial submission happy when null reverse-path is specified
  (instead of the correct empty argument, which works but makes sendmail
  issue a warning, we now use <> in arguments to specify null paths),
  (thanks to Sebastian Hagedorn and Dale Perkel);
- replace log entry "spam from=<%s>, to=%s" with "SPAM, <%s> -> %s ...".
- dropped (redundant) historical %bypass_checks @bypass_checks_acl
  (use bypass_virus_checks and %bypass_spam_checks instead);

patch4 (2002-11-21):
  In amavisd:
- SQL lookup fix;
- finished per-recipient header edits and split-forwarding, where
  one multi-recipient message may be forwarded as several messages
  if different headers need to be inserted for different recipients
  (not with milter);
- the check for banned file contents now looks for both the short
  and the original full type classification as provide by 'file' utility
  (thanks to Wolfgang Lumpp for the idea);
- #!/usr/local/bin/perl => #!/usr/bin/perl
- cosmetics

  In amavisd.conf:
- suggested group name change $daemon_group 'sweep'=>'amavis';
  so that non-Sophos users are not puzzled where 'sweep' group comes from;
- added path and fixed args for KasperskyLab AVPDaemonClient
  (thanks to Mike Hall);
- added path for Trend Micro FileScanner vscan
  (thanks to Eduardas Paulavicius);
- added 'use strict;' and '1;' at the end; prevent read_config from failing
  if the config file does not return true (thanks to Vivek Khera);
- cosmetics, fixed typos in comments;

patch2 & patch3 (2002-11-18, 2002-11-19):
- fixes a problem with 'recip_done' object method (causing TEMPFAIL)
  when $warnvirusrecip is enabled (thanks to Chris Hastie, Dale Perkel,
  and Didi Rieder);
- pass null return path to sendmail in such a way to make it happy
  (thanks to Sebastian Hagedorn and Dale Perkel);
- added a safety precaution to prevent somebody shooting himself in the
  foot by specifying an empty $forward_method while using the SMTP in/out
  setup (e.g. Postfix), which could send mail down the bit bucket;

patch1 (2002-11-17):
- fixes a problem in milter setup where per-recipient response codes
  needed to be handled differently (the problem was causing a TEMPFAIL)
  (thanks to Stephane Lentz and Didi Rieder for the initial milter testing
  and problem reports);

                                                          November 16, 2002
amavisd-new-20021116 release notes

- provide a mechanism to load only required code sections
  (anti-virus scanning, anti-spam scanning, SMTP/LMTP server module,
  traditional amavis client & milter server module),
  resulting in reduced memory usage and less installation dependencies;

- introduce the per-recipient status handling and make possible some
  per-recipient functionality that was not possible before, such as
  handling the per-recipient spam thresholds. This required a major
  rewrite and cleanup of some sections, hopefully providing cleaner
  mechanism for possible future advances in this area.

  Consider for example a previously incorrectly handled situation
  where there is more than one recipient, and some (but not all) of them
  get a reject from the (outgoing) relay MTA - it is not possible for a
  single SMTP response or status code to describe the situation;

- generate proper (non-)delivery status notifications (DSN), compliant with
  rfc1892 (now rfc3462) and rfc1894 (now rfc3464). This form supersedes
  the simple virus/spam sender notifications, but also covers the area of
  genuine delivery failures, such as selective recipient rejects by the
  outgoing relay MTA, which could previously lead to multiple deliveries;

  NOTE: to avoid sender getting two non-delivery messages (one from
  MTA, and another from amavisd), do not set $warnvirussender and
  $final_virus_destiny=-1 (REJECT) at the same time (and equivalently
  for spam settings);

- enhanced mail system status codes (rfc1893 (now rfc3463), and rfc2034)
  are now included with all SMTP responses and DSN notifications;

- added checking for banned MIME types and names. If any mail part
  matches, the whole mail is rejected, much like the way viruses
  are handled. A list in object $banned_filename_re can be defined
  to provide a list of Perl regular expressions to be matched against
  each part's:

  * Content-Type value (both declared and effective mime-type),
    including the possible security risk content types
    message/partial and message/external-body, as specified by rfc2046;

  * declared (recommended) file names as specified by MIME subfields
    Content-Disposition.filename and, both in their
    raw (encoded) form and in rfc2047-decoded form if applicable;

  * file content type as guessed by 'file' utility and classified into
    short type names such as .asc, .txt, .html, .doc, .jpg, .pdf,
    .zip, .exe, ... - see subroutine determine_file_types().
    This step is done only if $bypass_decode_parts is not set.

  NOTE: by default the $final_banned_destiny is set to 1 (pass),
  so detected banned file names only cause a header line to be added,
  quarantining, and added address extension - but the mail is delivered
  nevertheless. This default is set so that we can get initial experience
  without being too obtrusive. Change as you prefer.

- besides SMTP/ESMTP protocol, the server now also accepts LMTP protocol
  (rfc2033). This now explains why I had to implement PIPELINING,
  ENHANCEDSTATUSCODES and 8BITMIME SMTP extensions, as these are required
  by rfc2033.

  One advantage of using LMTP to feed content filter is that LMTP uses
  per-recipient status response, as opposed to 'one-size-fits-all'
  SMTP status response, which require that the SMTP client (e.g. content
  filter) is capable of generating DSN.

  Another advantage is specific to the Postfix setup, as Postfix
  LMTP client supports multiple transactions per session, saving on
  connection teardown/reconnect for every message being checked.

  To enable LMTP feed from Postfix, add 'max_use=10' to, and
  replace last argument 'smtp' with 'lmtp' in the line:
  'smtp-amavis unix - - n - 2 lmtp'. No changes are needed in amavisd-new,
  both protocols can coexist, the distinction is based on HELO/EHLO
  vs. LHLO command.

  Btw, older versions of Postfix lmtp client inappropriately lowercased
  the envelope addresses. This is fixed in Postfix Snapshot 1.1.11-20021015,
  and in the regular Postfix release 1.2 (when available). Lowercasing
  can be a problem for addresses where local part is case-sensitive,
  although such setups are rare.

- improved per-recipient SMTP response code handling when sending mail
  via SMTP. Previously one rejected recipient (by MTA) in a multi-recipient
  message caused the whole message to me rejected. This is now correctly
  handled with the new ability to send DSN;

- added command line option '-c config-file' so one can override the
  default location of the configuration file (/etc/amavisd.conf);

- explicitly set PATH and HOME environment variables
  (settings: $path and $helpers_home)

- added another form of lookups: Perl regular expression matching.
  See README.lookups for details. Corresponding new variables in amavisd.conf
  are: $virus_lovers_re, $spam_lovers_re, $bypass_virus_checks_re,
  $bypass_spam_checks_re, $local_domains_re, as well as $banned_filename_re,
  $viruses_that_fake_sender_re, $keep_decoded_original_re,
  $whitelist_sender_re, $blacklist_sender_re.

- besides whitelist_sender* lookup tables, there is now also
  a blacklist_sender* set of tables, which causes mail to be declared spam
  and to skip remaining spam checks. See examples in amavisd.conf.

- provide a configurable lookup table $keep_decoded_original_re
  of file types, for cases where unpacker is not very trustworthy.
  The lookup key is what 'file' utility returned. If the part contents
  match the lookup table, we keep both the original and the unpacked file
  for virus scanner to check;

- provide a configurable list of regular expressions
  $viruses_that_fake_sender_re, which recognizes viruses that usually
  fake envelope sender address. Don't send sender notification if a match
  is found. For syntax of the new regexp lookup tables see README.lookups;

- specifying per-recipient quarantine address or location is now possible
  by setting $virus_quarantine_to and $spam_quarantine_to be a ref to a
  hash lookup table. Thanks to Vivek Khera for the idea;

- no 'configure' and 'make' for the daemon; all configuration is done
  via amavisd.conf configuration file at the daemon startup time;
  Also the ./mta, ./av and ./notify configuration subdirectories
  are now gone;

- provided a sample init shell script (edit to will,
  and move it to /etc/init.d/amavisd if you want). Based on the script
  from Wil Cooley;

- one-shot debugging mechanism: if envelope sender matches @debug_sender_acl
  lookup table, turn debugging fully up just for this one message and
  cause temp file and directories not to be cleaned for this message.
  This facilitates debugging a particular problem even in the presence
  of regular traffic;

- cleaner debug log entries for multi-transaction SMTP/LMTP protocol sessions;

- cleaner log entries - avoid misleading and incomplete text
  when quarantine is disabled; thanks to Michael Leone;

- avoid using (nonstandard) field width in the %e format specifier
  when calling strftime, which lead to bad date syntax on some systems;

- local time zone offset is now automatically computed, no more '-0000'
  in rfc2822 dates;

- put Message-ID field value in angle brackets as required by rfc2822
  when generating notifications;

- add a In-Reply-To field to notifications when original Message-ID is known;

- don't bounce a virus (or spam) back to a mailing lists even if
  $final_virus_destiny (or $final_spam_destiny) is set to REJECT;
  a patch by Brian May from the Debian support crew, thanks;

- because some external module may play games with STDIN and STDOUT
  (like SpamAssassin seems to do when local_tests_only=>0)
  run the input protocol directly on the Net::Server's socket,
  not on STDIN and STDOUT, which are aliases to the socket;

- a patch to amavis-milter.c by Didi Rieder to support REJECTing mail,
  by instructing sendmail to return non-delivery notification to sender;

- a modified amavis.c helper program to make possible to invoke
  local delivery agent from it, for those still using such a setup;

- a new macro %l is available for use in notification messages (via
  'expand'); it evaluates to true (1) if the sender matches @local_domain,
  and returns empty otherwise; by default it is now used in creating
  a 'Subject:' line, inserting word 'LOCAL' before the sender name
  when appropriate;

- new macros %D and %N expand to lists of recipients that got the mail
  delivered (%D), or not-delivered (%N).  Union of both sets gives %R,
  i.e. a list of all recipients as specified in the envelope;

- dropped macro %a

- improved parsing of e-mail addresses according to rfc2821
  (full address literals syntax, etc.);

- proper line folding for generated rfc2822 header fields;

- proper SMTP response wrapping (for very long responses such as
  the ones that include a trouble report) according to rfc2821;

- allow for obsolete rfc822 syntax of permitting whitespace before
  colon in header field name;

- added optional spam-sender nondelivery notifications, based on patch
  from Lazslo E. Miranda ( and Fernando F. Morais

- dynamically change process name (Perl variable $0) to reflect
  the process state; suggested by Chip Paswater. Not all operating
  systems make this process state visible by ps(1);

- determine location of external programs (or their absence) at
  startup time; in amavisd.conf one may specify absolute path or just
  rely on PATH. This mechanism is also used to determine absolute path
  of the daemon itself, making reload (after HUP) more predictable;

- explicitly specify lock file for serialization to be used by the
  Net::Server::PreForkSimple module, instead of relying on default
  provided by POSIX::tmpnam. The default approach has a possibility
  for a minor security problem, because the lockfile is created with
  open()..., so it will follow symlinks. Observed, and a patch
  provided, by Jarno Huuskonen;

- new defense against mail bombs: for the cumulative total of all
  decoded mail parts we set max storage size. The formula is:
    quota = max($MIN_EXPANSION_QUOTA,
                min($MAX_EXPANSION_QUOTA, $mail_size*$MAX_EXPANSION_FACTOR))
  In plain words (later condition overrules previous ones):
    allow MAX_EXPANSION_FACTOR times initial mail size,
    but not more than MAX_EXPANSION_QUOTA,
    but not less than MIN_EXPANSION_FACTOR times initial mail size,
    but never less than MIN_EXPANSION_QUOTA

- if the permitted quota is exceeded (or the defense triggered by other
  similar safeguards), the virus scanning is skipped to protect the
  virus scanner from tripping over the mail bomb, a header field
  is inserted:
     X-Amavis-Hold: ... reason ...
  and a log entry 'Placing on HOLD: reason' at level 0 is produced.
  Also the temporary directory is preserved.

  The 'X-Amavis-Hold: ...' header field can be used by your MTA to
  put the message 'on hold' (freeze). If MTA is not set up to catch
  messages with this tag (the default), they are passed normally to
  recipients. This is likely to be the best action under the circumstances.

- tested with razor-agents-2.20 and SpamAssassin-2.43 (Razor2 is now called
  by SpamAssassin, and no longer directly by amavisd-new). New configuration
  variable $helpers_home, which defaults to $TEMPBASE. Thanks for hints
  by Chris Hastie and John Stewart;

- to avoid CPU loop in SpamAssassin-2.43 and earlier, my patch needs
  to applied to SA - see amavisd-new web page;

- to avoid taint problem in Razor 2.20 (if SpamAssassin-2.43 is configured
  to call it), my patch needs to be applied to it - see amavisd-new web page;

- chroot available (but not well tested): $daemon_chroot_dir = '/var/amavis'

- provide a fail-over mechanism for SQL database connect - given a list
  of SQL servers/databases, pick the first that is available.
  Thanks to Ken McKittrick for making available the patch,
  and to Ben Ransford <> for writing it.

- remove existing virus-related and spam-related headers (some of them
  optionally) if we'll be providing our own; suggested by Borut Mrak;

- avoid 'insecure dependency' in lookup_sql when calling DBI::execute
  (thanks to ric* at

- fixed macro %H to provide original header lines, not the ones
  stored internally by MIME::Entity, which may have been modified;
  (thanks to Chris Hastie for noticing the problem);

- no longer inserts X-Razor-id header field;


- when specifying boolean values to variables in the amavisd.conf file,
  please specify 1 (or old style "yes") for true, and 0 or "" or undef
  for false. The old style "no" yields true for Perl, and is only still
  supported (converted to 0) for some traditional variables for
  compatibility with amavisd(-snapshot);

- removed variables $sendmail_wrapper*, and changed the syntax for
  specifying $forward_method/$notify_method, with the intention to
  do all mail sending settings at one place. See amavisd.conf for examples;

- variable $mailto is now deprecated (but still works as a fallback default
  for compatibility with previous version). Use $virus_admin and $spam_admin
  lookup tables instead, they also offer a mechanism to specify per-domain
  administrator address;

- variable $warnadmin is no longer used. Use $virus_admin and $spam_admin
  lookup tables instead. Not specifying administrator address
  (e.g. leaving $virus_admin, $spam_admin and $mailto undefined)
  turns off admin warnings.

- dropped variable $LOGDIR, the variable $LOGFILE now specifies the full path;

- renamed: $warnsender -> $warnvirussender, $warnrecip -> $warnvirusrecip

- dropped variables $enable_relay, $sendmail_cf_orig, $QMAILDIR;
  they were never used in amavisd-new;

- dropped $MAX_ARCHIVE_NESTING, which is replaced by new storage limitations:

- SQL database is now specified differently (@lookup_sql_dsn)

MTA-SPECIFIC: sendmail/milter

- $notify_method now specifies deferred delivery mode ('-odd') by default,
  when submitting notifications to sendmail. This is to avoid calling milter
  immediately during submission, which in turn calls amavisd-new, possibly
  leading to a deadlock situation when the number of amavisd-new child
  processes is small. Seems like this change is needed since Sendmail 8.12
  or so.

  The following recommendation is from mimedefang-filter man page:

  | You MUST run a client-submission queue processor if you use
  | Sendmail 8.12.  We recommend executing this command as part of the
  | Sendmail startup sequence:
  |     sendmail -Ac -qp1m


- rewritten decode_parts() to allow for retaining source text if the
  unpacker sw is considered unreliable. This more cleanly resolves
  the problem reported on the amavis-user mailing list on 2002-06-06:
    "Amavisd passing through VBS/VBSWG.gen@MM"
  A nice side-product is that a directory search is avoided for
  each nested unpacking step;

- refined do_unzip to control and limit the size of decompressed
  members (among others it defends against the bombs);

- rewritten external decompressors interface to the gzip/bzip2/compress
  family. Instead of using a 'system' call, they are now called through
  fh_copy, making possible to control and limit the size of the
  decompressed contents on the fly, avoiding denial-of-service attacks.
  Affected: b(un)zip2, g(un)zip, (un)compress;

- fh_copy now uses IO::Handle object to assure the forked process
  gets reclaimed even in case of aborted contents extraction;
  stdin gets redirected to /dev/null or to a specified input file
  for the exec'd process;

- Convert::UUlib is called again (the amavisd-new-20020630 removed its
  usage due to problems with improper decoding). This time the originals
  are kept, so that virus checker sees both the original and the
  attempted-decoded part;

- completely rewritten do_ascii and its usage of Convert::UUlib
  to fix:
  * coding error (bitwise op treated as logical op and improperly negated):
      ... if (!$uu->state || !FILE_OK || -z $newpart);
  * it never reset the state, so if a successfully decoded ASCII file
    contained another ASCII file, each decoding level would decode all
    previously decoded parts again, plus add new ones at this level;
    recursion would not stop until the hard limit, resulting in TEMPFAIL;
  * it never checked nor reported errors that should have been detected
    (I/O errors, out of memory, trouble accessing or creating files);
  * more informative log entry;

- save MIME preambles and epilogues (if nonempty) as extra (pseudo)parts
  to be scanned. This also mitigates the problem of syntactically-incorrect
  MIME mail as produced by some user agents, which (rightfully) gets
  treated as one long preamble by MIME-Tools, and previously went by

- supports unpacking arc archives using 'nomarch' (by Russell Marks,, (thanks to David D. Kilzer
  for the initial code). Using 'nomatch' fixes a nasty habit of arc
  which gratuitously appends a form-feed at the end of file when using
  the 'p' (pipe) option, which might mislead a virus scanner.
  Besides, 'nomarch' is GPL licensed;

- fixed a 'broken pipe' problem when calling unrar,
  thanks to Ricardo Campos Passanezi and Rainer Link;


- rewritten interfacing for most command-line virus scanners.
  All settings for them is now done in amavisd.conf. New ones may be
  added without having to modify the daemon source. More complex scanners
  (e.g. daemonized scanners) still need to have a corresponding interface
  routine in the daemon;

- rewritten Sophie and Trophie interfaces to be more resilient
  to Sophie/Trophie daemon restarts during virus database reloads,
  avoiding an unnecessary retry (TEMPFAIL) - thanks to Cor Bosman
  for the suggestion and code, and to Dale Perkel for testing the
  Trophie interface;

- make sophos sweep tolerant to encrypted attachments: if all files are
  password protected, then the scanner failure is ignored and the message
  is allowed to pass. Based on patch by Radu Greab;

- updated nai uvscan interface to recognize the result
  'Found trojan or variant Exploit-CodeBase !!!', a patch by Anton Berezin;

- Clam Antivirus supported;

- fixed a problem in 'avp' scanner interface ($TEMPBASE not imported),
  thanks to Joshua E Warchol;

- avpdc (KasperskyLab AVPDaemonClient): recognize additional exit codes,
  based on patch from Christian Hammers;

- Panda new regexps, no TERM vt100 setting (thanks to Benjamin Zwittnig);

- Trend vscan exit code seems to be the number of infected parts.
  Updated the test to reflect that, based on observation from Stephane Lentz;

- MksVir scanner interface returns error code 2 if viruses are removed
  (if --clean option is passed) - a patch by Robert Litwiniec;

                                                               June 30, 2002
amavisd-new-20020630 release notes

Since it seems like several people are adapting amavisd-new
in details to their requirements, and certain improvement requests
have much in common, I'm making available the 20020630 release,
shortly before leaving on vacation. As this means my support
will be absent for the coming few weeks, and this release is
perhaps by few days premature, please consider it primarily a
development and new features release.

It is available at the usual location at:
  (or ask Google about 'amavisd-new')

Having said that, it is still a fully functional and tested version,
and it is running in production at a couple of sites now. Also it is
my version of choice in view of dependability, having it running
at our site while I'm away. If you get into trouble, you can still go
back to amavisd-new-20020517 with which it is fully upwards compatible.

The main changes and features since amavisd-new-20020517 are:

- the code is thoroughly rearranged, interfaces cleaned, separated
  into namespaces (packages), several sections generalized (e.g. lookups,
  appending/editing header lines). The AV scanner and unpacking sections
  are still mostly the same and compatible with amavisd, so whatever
  improvements and new AV scanner support becomes available for amavisd,
  applies almost without a change to amavisd-new;

- SMTP on the input side (used with Postfix and Exim) now talks ESMTP
  (rfc2821) and not just rfc821, including some SMTP extensions:
  command pipelining (rfc2920), message size declarations (rfc1870), and
  8bit-MIME transport (rfc1652). The main reason for this was the change
  in recent Postfix versions which can now do MIME transformations to support
  7bit transports (implied by SMTP). To ensure the transparent 8bit path and
  avoid message transformation by MTA, amavisd-new needs to declare it does
  present an 8bit-clean path. A side benefit is a little speedup in passing
  chunks of mailing list addresses due to pipelining support;

- split certain previous amavisd.conf settings (variables)
  into several variables or lookup tables:

  * sender address for notifications:
    $mailfrom -> $mailfrom_notify_admin, $mailfrom_notify_sender,
                 $mailfrom_notify_recip, $mailfrom_notify_spamadmin
    (these may also be empty to specify null reverse path <>,
    which is most useful for sender notification);

  * administrator address for notifications:
    $mailto -> $virus_admin, $spam_admin (per-sender lookups)

  * %bypass_checks -> %bypass_virus_checks, %bypass_spam_checks

- supports SQL database lookups via Perl module DBI (interface to
  popular database types). Some examples are provided, if you need
  other SQL lookups just modify the code by analogy. SQL lookups
  are most useful for per-user settings of virus_lovers, bypass_virus_checks,
  bypass_spam_checks, spam thresholds etc. when the user base is large
  and subgroups can not be identified through their (sub)domains.
  Another use is for dynamically changing settings without having
  to restart amavisd-new;

- can optionally insert 'Received:' header if acting as a mail relay
  (not with milter); does loop detection as required by rfc2821 section 6.2;

- notification messages now contain 'Date:' and 'Message-id:' headers;

- quarantined viruses contain X-AMaViS-Alert header line with names
  of detected viruses;

- quarantined spam contains X-Spam-Status and X-Spam-Level header lines;

- optionally send spam admin notifications, which include the full
  SpamAssassin spam report and message header;

- when started as root, changes UID and GID to $daemon_user, $daemon_group;

- to facilitate startup scripts and debugging, supports few simple
  command line parameters:
    amavisd        ... standard run: changes uid/gid and daemonizes
    amavisd start  ... same thing
    amavisd debug  ... starts with full debug level, stays attached
    amavisd reload ... finds amavisd master process and sends it a HUP
    amavisd stop   ... finds amavisd master process and sends it a TERM

- some more (minor) configurable options:
    $daemonize, $pid_file,
    $replace_existing_extension, $localpart_is_case_sensitive

- no longer calls Convert::UUlib for uuencoded, xxencoded, and binhex
  attachments. The first two are handled by MIME::Parser, the remaining
  are likely to be handled by anti-virus scanners, especially when some
  virus would use such encoding. The use of Convert::UUlib is dropped
  because it was causing recent problems with garbling virus so that it
  could no longer be recognized by AV scanners, and because the underlying
  library does not seem very dependable. See the thread 'Amavisd passing
  through VBS/VBSWG.gen@MM' from the beginning of June 2002 in amavis-user
  mailing list archives;

- quarantine files now include internal amavis id in the file name,
  instead of the process number; previous naming scheme could stumble
  across a name contention on a busy system;

- only a recommendation: logging via syslog is now preferred to direct
  logging to a file. It serializes the logging, and avoids locking/unlocking
  and reopening a log file by amavisd* for every log entry.
  The syslog daemon does it more efficiently and reliably.

  NOTE for Linux users: make sure you prefix the file name in syslogd.conf
  with a '-' tag to disable fsync after every write; this is most necessary
  for heavy logging such as from MTA and/or amavisd-new.  The downside
  is that you may lose the last few log entries in case of machine crash.
  See man page of your syslogd for details.

- changed examples according to rfc2606 and recent complaints on the
  postfix-users list.


- header rewriting is only available in SMTP-in/SMTP-out setup,
  i.e. with Postfix and Exim, but not with sendmail milter setup;

- the SQL lookups are a very recent addition and not so well tested
  as the rest of the program. Also since SQL lookups are supposed
  to introduce per-user settings (e.g. spam thresholds), the code
  still does not accommodate it, and spam thresholds of the first
  recipient in a message affects the whole message;

- as Razor2 is still pretty much unstable, I left the spam_scan()
  routine much as it was in the May version. Contributions welcome;

- as I ran out of my time for this release, I didn't prepare a
  separate version without SpamAssassin and Razor 1.20 support,
  so you will need to strip it out if you do not need it.
  Similarly for the required DBI Perl module. I'm very sorry;

- the time stamp used in the 'Date:' header in notifications,
  and in an optional 'Received:' header, does not contain true time zone
  offset, but -0000 instead (standard meaning for unknown). I did not want
  to include a fat Perl package for handling time zones. For now just edit
  the subroutine rfc2822_timestamp() if you want to change that;

- amavisd.conf settings $daemon_user and $daemon_group are not
  automatically set by ./configure options. Please set them manually.

I would like to thank many people on the amavis-user list, on the
postfix-users list, and in private conversations, who contributed
valuable ideas and improvements, and offered much appreciated

                                                               May 17, 2002
Available at:

amavisd-new-20020517 is primarily a response to popular demand for
Mail::SpamAssassin support. If amavisd-new-20020424 meets your needs,
there is no urgent need to upgrade. There are no incompatible changes
between these two versions, except some new (optional) amavisd.conf
variables, so you may keep old amavisd.conf file if you wish.

A new file README.exim is now provided, so that Exim 4.x is now a
supported and tested configuration, besides the usual Postfix and
sendmail/milter MTAs. Also included is a brand new qmail amavis client
by Lars Hecking (untested, please try it), plus his updated version of
amavis.c (no longer needed in recommended Postfix and Exim configurations,
and Sendmail milter and qmail configurations use a different client).

Several files are unchanged from the base amavisd CVS release 2002-05-13
and do not reflect the amavisd-new state: FAQ, HINTS, INSTALL, BUGS,
NEWS, TODO, ChangeLog, tests/, doc/. Please start with this file
README.amavisd-new-RELNOTES and follow it. More recent instructions
and last-minute changes are available from the web page.

Changes since amavisd-new-20020424:

- supports Mail::SpamAssassin and Vipul's Razor (1.20 required)
  for spam checking (but not for modifying mail body - only add headers
  and/or address extension, or reject/discard/quarantine spam).

  NOTE: spam checks are off in the default amavisd setup.
        Copy file ./amavis/ to ./amavis/
        before running ./configure and make, to use the
        spam-check-enabled version!

  Amavisd calls Mail::SpamAssassin directly, avoiding the need to set up
  spamc/spamd or to chain filters. This is more efficient, one daemon
  less to worry about, although maybe less flexible for some taste.
  At the moment the SA per-user database is not used. Feel free to
  experiment with it and let me know what you came up with.

  The usual SA config files are observed, but remote tests are disabled
  by default (Razor is handled directly, RBL lookups can/should be
  performed early by MTA which knows what IP address mail came from,
  and that information is (mostly) lost afterwards). If you feel otherwise,
  change the hard-wired settings in the call:

    $spamassasin_obj = Mail::SpamAssassin->new(
      { dont_copy_prefs => 1, local_tests_only => 1 } )

  (also the Razor score contribution is hard-wired,
  modify: '$razor_spam_found ? 3 : 0' to will).

  Amavisd-new handles Razor checks directly to be able to exercise
  more control over it than would otherwise be available through SA:
  timing, signatures are needed for insertion into header, skips one-liner
  body checks which are common Razor false-positives. To avoid SA calling
  Razor again, either keep the default setting 'local_tests_only=>1',
  or set 'score RAZOR_CHECK 0' in the SA configuration.

  Added are 'whitelist_sender' hash and ACL lookups (see README.lookups),
  which approves spam from specified SMTP originator addresses - SpamAssassin
  can only check and whitelist rfc822 headers, not the envelope addresses,
  and I see no way of passing envelope addresses to it - which is a pity,
  as important information is lost.

  SpamAssassin checks are computationally quite expensive compared to other
  amavisd activities, and the time needed for SA check goes up significantly
  with the message size. SA check are skipped (but not the Razor check)
  if mail size exceeds 64k. My analysis shows that presently less than 1%
  of spam exceeds 64k characters, and this is probably well below the
  false-negative SA rate, so it is not worth wasting time to check large mail.

remaining changes:

- new file README.exim (thanks to Jochen Erwied, Patrice Fournier
  and Igor D'Astolfo);

- updated README.postfix to describe how to avoid running header_checks,
  body_checks and dns_lookups in Postfix twice;

- put back file README.customize, which was omitted from amavisd-new-20020224
  by mistake (but available with previous version and on the web);

- added introductory paragraph to README, the rest still needs to be updated;

- added missing last-step check for '.' in hash lookups to match examples
  and make it more useful;

- added timeouts to certain tasks to make it better suitable for unattended
  operation (less, but still subject to certain DoS attacks, similar to
  the official amavisd). The main reason for adding timeouts is that we
  don't want spam checking to slow down amavisd operation too much: if it
  takes too long, just skip it (assume not spam) and move on;

- change unmangle_sender() to believe sender address for Klez viruses;
  this is sometimes wrong, but seems like people prefer to sacrifice
  few false accusations in favour of some warranted sender notifications.
  Feel free to hack this routine at will (and publish good ideas),
  it is intended to be modified;

- use Perl module Errno instead of to avoid using broken
  file on some popular platforms;

- make possible to have per-sender-domain administrator e-mail address
  for admin notifications (hash lookup %mailto, sub warn_admin() )

- relax temporary file/dir protection to allow them to be readable
  by the group. This makes possible for anti-virus (daemon) checker
  to run under a different user (but in the same group). Doing so should
  be safer as it makes impossible for virus-checking daemon to clobber
  files. It is only supported in the all-SMTP configuration. To use
  it with traditional amavis clients (e.g. milter), you will have to
  modify their sources to change umask and mods for file/dir they create;

- avoid changing sender address <> to <""> under certain circumstances
  (not strictly wrong, but still a bad idea);

- replaced header 'X-Razor-Warning:' with SA-compatible 'X-Spam-Status:' ;
  the 'X-Razor-id:' is still provided to facilitate user spam reporting;

- added '--' between options and argument when calling $sendmail_wrapper
  to be triply sure we avoid problems with some mailers (note that
  $sendmail_wrapper is only still needed in the sendmail setup,
  Postfix and Exim do not need it in the recommended configuration);

- tidy the %local_delivery_aliases mechanism for local delivery / quarantine;

- allow MIME::Parser to decode uuencoded parts, if it feels
  it can (and should) do it;

- write warnings from MIME::Parser to the log (at log level 1 or higher),
  instead of discarding them.

The package is available at the usual location at:

where also the most up-to-date version of FAQ
and certain other fresh documentation files live.

                                                             April 24, 2002
amavisd-new-20020424 is primarily a maintenance release
to summarize one week's worth of experience with amavisd-new-20020418
and to implement some good ideas from the amavis-user mailing list.
It also brings one or two new features.

It is available at:

or more specifically, at:

Changes since amavisd-new-20020418:

- removed Perl 'my' declaration from configurable variables which prevented
  them from being changed in amavisd.conf (thanks to Sebastian Hagedorn
  and Wouter de Jong for reporting)

- introduced child timeouts to prevent bad amavis client from monopolizing
  a child forever (thanks to Sebastian Hagedorn for reporting the problem)

- supported and documented (in README.postfix) configuration where multiple
  remote or local SMTP-in/SMTP-out MTAs (e.g. Postfix) can use the same
  amavisd server, by making it deliver checked mail back to the same
  IP address it came from (see variable $relayhost_is_client in amavisd.conf)
  (thanks to Wouter de Jong for the splendid idea which fitted naturally
  into the overall scheme)

- in certain log messages include the SMTP-in and SMTP-out MTA IP address

- new access list checking for IP addresses - used to limit SMTP access
  to authorized MTAs only: @inet_acl; access control is now enabled by default

- slightly more sophisticated hash-based access list lookups, modeled after
  Postfix map lookups. The sequence now goes through the following steps:
    - hash lookup for
    - hash lookup for  (only if $recipient_delimiter is nonempty)
    - hash lookup for
    - hash lookup for
    - hash lookup for
    - hash lookup for .in
    - hash lookup user+foo@
    - hash lookup user@  (only if $recipient_delimiter is nonempty)

- updated README.lookups

- cleaner quarantine code; new variable $mailfrom_quarantine allows
  to choose either the original envelope sender, or admin-specified
  fixed sender address; include a special per-user quarantine example
  (look for 'trouble-user-quarantine' in the amavisd)

- fixed problem with localized system error messages - use numeric errno
  instead of strings in Sophie and Trophie clients (thanks to Igor D'Astolfo
  for reporting the problem with Italianized version of Linux and for a good

- successfully tested with Exim
  (thanks to Igor D'Astolfo for testing and for reporting
  the <<doubly-quoted>> sender (but not recipient!) problem)

- wrong variable used in the spam section ($final_virus_destiny
  instead of $final_spam_destiny) (thanks to Wayne Smith for reporting
  and for testing Vipul's Razor 'plugin' patch)

- changed pattern match in the ./av/oav to support new version of
  OpenAntiVirus ScannerDaemon (thanks to Rainer Link)

- changed 'configure' to make it recognize Sophie 1.33rc1
  (thanks to Igor D'Astolfo and Lars Hecking)

- include acinclude.m4/acx_pthread.m4 macro
  (thanks to Rainer Link)

- updated README.postfix: includes instructions on how to avoid body_checks
  and header_checks for reinserted mail (contributed by Wayne Smith,
  works nicely, requires Postfix version 1.1.7-20020331 or later)

- clarified comments in amavisd.conf

- optionally keep existing X-Virus-Scanned: header lines, or remove them
  before adding our own header line - see $remove_existing_x_scanned_headers
  (requested by Darryl Harvey)

- avoid historical (misleading) parameter name $localhost_ip;
  use $relayhost and $relayhost_port instead, but take old variables into
  account for backwards compatibility with existing amavisd.conf files

- Here is an overall picture (sequence of events)
  of how pieces fit together:

    bypass_checks? ==> PASS
    no viruses?    ==> PASS
    log virus     if $log_templ is nonempty
    quarantine    if $virus_quarantine_to is nonempty
    notify admin  if $warnadmin
    notify sender if $warnsender
    notify recips if $warnrecip
    final_destiny==pass?  ==> PASS
    virus_lovers?         ==> PASS
    DISCARD or REJECT (depending on final_*_destiny)

                                                             April 18, 2002
This it to announce the second release of amavisd-new-20020418,
available at:

It is a version of amavisd (a daemonized AMaViS, which is an interface
between MTA and virus scanners), based on amavisd CVS from today (20020418)
(same configuration, amavis clients in C, instructions, AV client code),
while also being a successor of the initial release of amavisd-new-20020329,
which is a performance-enhanced pre-forked Net::Server -based amavisd
with SMTP-in/SMTP-out capability, written in Perl.

(I tried to make this release based on amavisd-snapshot-20020300, but failed,
as that version still has $errval semantics bug (the bitwise-operations
problem was discussed on the AMaViS-user list some time ago).
This was fixed in the CVS version and most AV clients were changed then.
As the CVS version contains AV client code for new AV scanners, this makes
it incompatible with config stuff from amavisd-snapshot-20020300.

If you have problems with ./configure or make, these problems would be
common to both versions. Also the unpackers and decompressors code is mostly
the same for all recent amavisd versions, so any problems in this arena
(like DoS mishandling) are most likely to be common for both the official
amavisd and the amavisd-new version.

Compared to the first release of amavisd-new-20020329, the second release
brings further significant performance improvements especially in the
SMTP-in/SMTP-out configuration, e.g. with Postfix, but also brings
some interesting new features and new configuration possibilities.

No important bug fixes were needed, so upgrade at your leisure if you are
running my initial version. The Postfix users would perhaps want to rush
a bit though, to put new performance improvements into use.

The summary of changes since the initial version:

- significant SMTP-in speedups (25% with fast AV scanner), file reuse
- pass reject reason to MTA on the input side
- more informative MTA log entries in the SMTP-in/SMTP-out setup
- amavis internal id (am_id) in log entries and passed to MTA in SMTP response
- ISP features: specify subgroups of users who want to receive viruses
- address extensions: e.g. user@domain -> user+virus@domain if virus detected
- can specify final_virus_destiny: reject, discard, pass
- quarantine new options: save to individual file, save to mailbox, pass to MTA
- new headers in quarantined viruses preserve envelope addresses
  and quarantine id (similar to the suggestion from Furio Ercolessi - see code)
- detailed timing breakdown report for each passed message
- anti-spam hooks and examples, example patch to integrate Vipul's Razor client
- body cache now always enabled - heavy speedup for mailing list bursts
- Sophie 1.33-ready
- rewritten Trophie client, based on new Sophie client code
- rewritten README.postfix, describing new setup possibilities
- new file README.lookups (to be used with virus_lovers and bypass_checks)
- new file README.customize (same as in the initial version)
- new file README.performance (unfinished)
- new amavisd.conf options, documented there
- code heavily commented, cleaned, generalized again
- does not accuse innocent users of sending viruses if we are suspicious
  of sender address (see FAQ below) (same as in the initial version)


- no test mode;

- only Postfix and sendmail milter are fully supported and tested;
  to integrate with other mailers one would need some understanding of their
  operation to set up properly; contributions are welcome;

(P.S. note: this is probably a non-issue, related to the file system problem:
  All amavisd versions seem to share one still unresolved problem, probably
  with amavis-milter.c client, its use of libmilter, or perhaps even in the
  libmilter code itself - see recent thread on the AMaViS-user list
  (subject: Leftover email.txt files with amavisd 'standard' also).
  The problem is most pronounced in burst of heavy traffic.


- Net::Server 0.82 triggers a Perl 5.005 bug (the problem is obvious:
  you get syntax errors). Either upgrade to Net::Server 0.83, or
  upgrade your Perl - 5.6.x should be ok.

- if you intend to play with customized notifications, it is wise
  to remove the '-t' option from $sendmail_wrapper_args in amavisd.conf.
  That way you are free to screw up notification mail headers any way you want,
  and the message would still be delivered to the correct recipients.
  Removing '-t' is now the recommended setting, but both variants should work.
  If you use SMTP-out method for notifications (e.g. Postfix recommended
  and default setting), this does not affect you.

- if you see virus notifications claiming the virus originator is <?>
  or <?@some.domain> and sender notifications are not sent, this is not a bug,
  but a feature - see comments at the subroutine unmangle_sender().
  The original idea comes from Furio Ercolessi: as some viruses tend to use
  forged or corrupted sender or 'From:' addresses, we try to determine
  the true virus sender, and if we can not do that, we avoid accusing
  innocent users of sending viruses.

- if you kill or HUP amavisd, temporary directories may be left undeleted;
  this is normal and mails are not lost;

- if amavisd does not restart after receiving HUP, a possible reason may be
  that amavisd can not be found in the path as set in the $ENV{PATH} variable
  (near the beginning of amavisd program). Another reason may be a syntax
  error if you changed the amavisd.conf file. Try to start it manually:
    $ su vscan -c amavisd
  If that does not make you wiser, set $DEBUG = "yes" and retry.
  This is also the recommended first-time start method.

- after changing $inet_socket_bind in amavisd.conf, you must stop amavisd
  and start it anew. The HUP method causes amavisd to stumble over its feet.

Below is my announcement notice for the initial version of amavisd-new,
just slightly edited to remove some mistakes.

Date: Sat, 30 Mar 2002 04:13:25 +0100
From: Mark Martinec <>
Subject: [AMaViS-user] ANNOUNCE: new amavisd - leaner and meaner
Message-id: <01KFYI1DJ2O200AMKT@CATHY.IJS.SI>

I would like to announce an updated version of amavisd,
based on recent CVS code of amavisd (which is not far
from the February amavisd snapshot). It is a result of my
three weeks work on the code, caused by our needs for:
- better reliability;
- higher throughput (less overhead);
- versatility (e.g. separating amavis and MTA hosts, load sharing),

Available at:

This is now finally a version which I can recommend to friends :)
TODO: better DoS handling in unpackers.

Main features - in brief:

- pre-forked reusable children - saving on process creations;
- persistent connections to certain AV scanners, e.g. Sophie, saving on forks;
- both SMTP and pipe (sendmail wrapper) interfaces independently available
  and configurable on all three sides: input, output, notifications;
- sendmail Milter interface supported and tested; Postfix supported and
  thoroughly tested (Exim untested; classical sendmail untested, no qmail);
- customizable notification messages;
- compatibility with existing configurations;
- cleanups, generalizations, speedups, fixes, better code documentation;
- HUP signal causes restart with new configuration;
- ISP feature: certain recipients may be allowed to receive viruses
  (with alert header line added (not with milter), notifications are
  still generated);
- anti-spam hooks, caching-ready, Vipul's Razor interface in a form
  of a patch included - should make integration with SpamAssassin easier.


  Unpack the tar over the checked-out CVS version of amavisd branch
  (or February snapshot). It overlays some files, the rest is unchanged.
  For milter interface make sure you use amavis-milter.c from CVS,
  as the snapshot version contains a bug which can cause message loss
  if amavisd dies.


- child and socket handling is now delegated to Perl module Net::Server,
  which gives us pre-forked children which are able to do more than one
  mail-check during their lifetime, saving on process creations
  and giving better response time;

- Net::Server controls number of children, does signal handling,
  takes care of dead children, handles listening on multiple sockets
  (both Unix and TCP/UDP), delegating tasks and synchronizing 'accept's;

- as a consequence, certain AV-scanner interfaces (most notably Sophie,
  and soon to follow Trophie) can keep persistent connections to the
  AV checking daemon, saving on AV scanner process creations
  (forks on accept), and socket setups/teardowns;

- on the input side: both SMTP (TCP) and traditional amavis client
  protocol (Unix socket) are now accepted - even both at the same time
  (by default) to ease transition to SMTP interface;

- new input-side SMTP interface (SMTP server) is easier to set up
  (no need to worry about file ownerships, UIDs and GIDs),
  and more versatile (e.g. SMTP responses carry more information
  that sysexits.h-based status codes, SMTP can talk to remote host);
  It is based on Perl native I/O and it is quite fast - no OO overhead,
  _not_ SMTP::Server-based (which I do not consider production quality);

- a nice by-product of SMTP input interface is the extra information
  available in MTA logs, e.g. Postfix log:
  postfix/smtp[7656]: DA7B147FA6:
    to=<user@domain>, relay=localhost[], delay=5,
    status=sent (250 Ok, discarded - VIRUS: EICAR-AV-Test)
- notifications messages can be delivered either via SMTP or
  as traditionally via pipe to sendmail wrapper / MTA pickup.
  This means that in the Postfix setup for example, one needs
  to worry only about one output interface (SMTP or pipe).
  This also saves unnecessary work of checking the just-generated
  notification messages for viruses;

- the output pipe interface is now more careful with status
  checking and forking;

- output SMTP interface error handling more compact;

- sender notifications can be sent to more than one contact address
  (not used at the moment, but the mechanism is available);

- sender address unmangling patch incorporated (but I'm willing
  to take it out if considered inappropriate for the base distribution);

- see file ./amavis/mta/postfix_init for comments describing
  how to select SMTP- or pipe-based output interface
  (for re-injection and notifications);

- customizable notification messages and log entry text
  (see README.customize);

- clean notion of when mail addresses are in their quoted and
  when in unquoted form (RFC2821). All internal handling uses unquoted
  form, addresses get quoted as required by the output interface,
  and quoting gets stripped away as required by each input interface;

- besides traditional choices where a mail can be forwarded/accepted,
  discarded, or temporary failure/retry indicated, there is now
  a fourth choice: REJECT. Depending on the MTA this requires a
  message rejection to be done by the input side MTA itself.
  This comes handy in cases when amavisd accepts a non-infected mail,
  but outgoing MTA does not want to take it back for final delivery,
  e.g. in case of some policy violation. Traditionally amavisd
  would indicate temporary failure on its input side, causing the
  message to be retried and re-scanned over and over again,
  without having a chance of ever being accepted;

- all existing virus scanners are still there (one little change
  in each: replaced a call to do_virus with return 1);
  Sophie client rewritten to take advantage of persistent connections;

- decoding sections are mostly unchanged. This area needs more work
  in the future;

- includes exit status codes from
  instead or having them hard-wired;

- anti-spam code easier to integrate into amavisd due to some code
  rearrangements; a patch to integrate Vipul's Razor is included
  as an example, but it may be better to tie amavisd with SpamAssassin.
  Anti-spam code will NOT be integrated into amavisd, but the least
  we can do is making it easier for people to add their own code;

- cache-ready (example in the included Razor patch). This works by
  calculating a message digest (hash, signature) of the message body
  and keeping it in storage for a short while, e.g. for the lifetime
  of a child process (10 consecutive requests by default).
  If another message with the same body content arrives in the near future
  we can skip a virus check. This comes handy where mailing list traffic
  is frequent, especially if we have to deal with poorly done mailing
  list managers or heavy spam traffic;

  To support this concept the functions of MIME decoding and
  unpacking of archives are now separated.

- MIME parsing is now supplied with our own Filer subclass.
  This was necessary to avoid MIME parser complaints when it tried
  to reconstruct file name extensions from file names in strange
  character sets. Its work was completely unnecessary and harmful,
  since we want to supply our own file names and do not care
  for file name extensions.

- as a consequence of our own Filer subclass, we now avoid the first
  (or the only) directory traversal (reading) in the first decoding pass;
  Not much, but every little bit counts.

- when using sendmail wrapper it is no longer necessary to supply
  the sendmail -t option. If -t is not specified, sendmail wrapper
  receives addresses via command arguments (exec, no shell),
  which is more reliable than having to parse mail headers - which are
  now more error-prone due to user-customizable notifications.
  The -t option is still supported though, but not recommended.

- do_unzip no longer complains with multi-line backtrace when it dies;
  just a single message is issued, like with other decoders;

- HUP signal causes restart and re-reading of config file;

- many new comments, code unifications, supplied some missing
  error checks, code generalizations;

MTA support:

- thoroughly tested with Postfix in all combinations of input,
  output, and notifications interfaces; in production use;

- big thanks to Sebastian Hagedorn who helped to test the milter
  interface on his Solaris!

- Exim and traditional sendmail interface untested, but should
  work without much work.

- Qmail interface is still missing, like in current CVS amavisd.
  I believe the amavis client for qmail can be written as
  a few-dozen line Perl program. Anybody want to try?

A quick cookbook on how to set up Postfix / amavisd interface
using SMTP on both amavisd input and output side (including notifications).
Amavisd by default now accepts both Unix socket and SMTP on the loopback
interface, so the transition is easier: first install new amavisd,
then at some other time change Postfix configuration (if desired) to:
  # MTA -> amavisd
  smtp-amavisd      unix - - n - 2 smtp

  # amavisd -> MTA
  localhost:10025   inet n - n - - smtpd -o content_filter=
  # choose transport to amavisd
  content_filter = smtp-amavisd:localhost:10024

[see new file README.postfix for details]

The amavisd can now easily be located on a different host than MTA,
also Postfix load-balancing transport methods can be used (e.g. multiple
MX records). Although amavisd now talks SMTP and incorporates some rudimentary
defenses against malicious SMTP clients, do not expose its SMTP server
directly to the world - always front-end it with MTA. By default it
binds to the local interface only, but other access restrictions are
also available.

Happy amavising!  Experiences and comments are most welcome.