1 This document contains detailed information about incompatibilities that might
2 be encountered when upgrading from one release of Exim to another. The
3 information is in reverse order of release numbers. Mostly these are relatively
4 small points, and the configuration file is normally upwards compatible, but
5 there have been two big upheavals...
8 **************************************************************************
9 * There was a big reworking of the way mail routing works for release *
10 * 4.00. Previously used "directors" were abolished, and all routing is *
11 * now done by routers. Policy controls for incoming mail are now done by *
12 * Access Control Lists instead of separate options. All this means that *
13 * pre-4.00 configuration files have to be massively converted. If you *
14 * are coming from a 3.xx release, please read the document in the file *
15 * doc/Exim4.upgrade, and allow some time to complete the upgrade. *
17 * There was a big reworking of the way domain/host/net/address lists are *
18 * handled at release 3.00. If you are coming from a pre-3.00 release, it *
19 * might be easier to start again from a default configuration. Otherwise *
20 * you need to read doc/Exim3.upgrade and do a double conversion of your *
21 * configuration file. *
22 **************************************************************************
25 The rest of this document contains information about changes in 4.xx releases
26 that might affect a running system.
32 * The value of $tls_peerdn is now print-escaped when written to the spool file
33 in a -tls_peerdn line, and unescaped when read back in. We received reports
34 of values with embedded newlines, which caused spool file corruption.
36 If you have a corrupt spool file and you wish to recover the contents after
37 upgrading, then lock the message, replace the new-lines that should be part
38 of the -tls_peerdn line with the two-character sequence \n and then unlock
39 the message. No tool has been provided as we believe this is a rare
42 * With OpenSSL 1.0.1+, Exim now supports TLS 1.1 and TLS 1.2. If built
43 against 1.0.1a then you will get a warning message and the
44 "openssl_options" value will not parse "no_tlsv1_1": the value changes
45 incompatibly between 1.0.1a and 1.0.1b, because the value chosen for 1.0.1a
46 is infelicitous. We advise avoiding 1.0.1a.
48 "openssl_options" gains "no_tlsv1_1", "no_tlsv1_2" and "no_compression".
50 COMPATIBILITY WARNING: The default value of "openssl_options" is no longer
51 "+dont_insert_empty_fragments". We default to unset. That old default was
52 grandfathered in from before openssl_options became a configuration option.
53 Empty fragments are inserted by default through TLS1.0, to partially defend
54 against certain attacks; TLS1.1+ change the protocol so that this is not
55 needed. The DIEF SSL option was required for some old releases of mail
56 clients which did not gracefully handle the empty fragments, and was
57 initially set in Exim release 4.31 (see ChangeLog, item 37).
59 If you still have affected mail-clients, and you see SSL protocol failures
60 with this release of Exim, set:
61 openssl_options = +dont_insert_empty_fragments
62 in the main section of your Exim configuration file. You're trading off
63 security for compatibility. Exim is now defaulting to higher security and
64 rewarding more modern clients.
66 * Ldap lookups returning multi-valued attributes now separate the attributes
67 with only a comma, not a comma-space sequence. Also, an actual comma within
68 a returned attribute is doubled. This makes it possible to parse the
69 attribute as a comma-separated list. Note the distinction from multiple
70 attributes being returned, where each one is a name=value pair.
72 * accept_8bitmime now defaults on, which is not RFC compliant but is better
73 suited to today's Internet. See http://cr.yp.to/smtp/8bitmime.html for a
74 sane rationale. Those who wish to be strictly RFC compliant, or know that
75 they need to talk to servers that are not 8-bit-clean, now need to take
76 explicit configuration action to default this option off. This is not a
77 new option, you can safely force it off before upgrading, to decouple
78 configuration changes from the binary upgrade while remaining RFC compliant.
84 * GnuTLS will now attempt to use TLS 1.2 and TLS 1.1 before TLS 1.0 and SSL3,
85 if supported by your GnuTLS library. Use the existing
86 "gnutls_require_protocols" option to downgrade this if that will be a
87 problem. Prior to this release, supported values were "TLS1" and "SSL3",
88 so you should be able to update configuration prior to update.
90 * The match_<type>{string1}{string2} expansion conditions no longer subject
91 string2 to string expansion, unless Exim was built with the new
92 "EXPAND_LISTMATCH_RHS" option. Too many people have inadvertently created
93 insecure configurations that way. If you need the functionality and turn on
94 that build option, please let the developers know, and know why, so we can
95 try to provide a safer mechanism for you.
97 The match{}{} expansion condition (for regular expressions) is NOT affected.
98 For match_<type>{s1}{s2}, all list functionality is unchanged. The only
99 change is that a '$' appearing in s2 will not trigger expansion, but instead
100 will be treated as a literal $ sign; the effect is very similar to having
101 wrapped s2 with \N...\N. If s2 contains a named list and the list definition
102 uses $expansions then those _will_ be processed as normal. It is only the
103 point at which s2 is read where expansion is inhibited.
105 If you are trying to test if two email addresses are equal, use eqi{s1}{s2}.
106 If you are testing if the address in s1 occurs in the list of items given
107 in s2, either use the new inlisti{s1}{s2} condition (added in 4.77) or use
108 the pre-existing forany{s2}{eqi{$item}{s1}} condition.
114 * The integrated support for dynamically loadable lookup modules has an ABI
115 change from the modules supported by some OS vendors through an unofficial
116 patch. Don't try to mix & match.
118 * Some parts of the build system are now beginning to assume that the host
119 environment is POSIX. If you're building on a system where POSIX tools are
120 not the default, you might have an easier time if you switch to the POSIX
121 tools. Feel free to report non-POSIX issues as a request for a feature
122 enhancement, but if the POSIX variants are available then the fix will
123 probably just involve some coercion. See the README instructions for
124 building on such hosts.
130 * The Exim run-time user can no longer be root; this was always
131 strongly discouraged, but is now prohibited both at build and
132 run-time. If you need Exim to run routinely as root, you'll need to
133 patch the source and accept the risk. Here be dragons.
135 * Exim will no longer accept a configuration file owned by the Exim
136 run-time user, unless that account is explicitly the value in
137 CONFIGURE_OWNER, which we discourage. Exim now checks to ensure that
138 files are not writeable by other accounts.
140 * The ALT_CONFIG_ROOT_ONLY build option is no longer optional and is forced
141 on; the Exim user can, by default, no longer use -C/-D and retain privilege.
142 Two new build options mitigate this.
144 * TRUSTED_CONFIG_LIST defines a file containing a whitelist of config
145 files that are trusted to be selected by the Exim user; one per line.
146 This is the recommended approach going forward.
148 * WHITELIST_D_MACROS defines a colon-separated list of macro names which
149 the Exim run-time user may safely pass without dropping privileges.
150 Because changes to this involve a recompile, this is not the recommended
151 approach but may ease transition. The values of the macros, when
152 overridden, are constrained to match this regex: ^[A-Za-z0-9_/.-]*$
154 * The system_filter_user option now defaults to the Exim run-time user,
155 rather than root. You can still set it explicitly to root and this
156 can be done with prior versions too, letting you roll versions
157 without needing to change this configuration option.
159 * ClamAV must be at least version 0.95 unless WITH_OLD_CLAMAV_STREAM is
160 defined at build time.
166 1. Experimental Yahoo! Domainkeys support has been dropped in this release.
167 It has been superceded by a native implementation of its successor DKIM.
169 2. Up to version 4.69, Exim came with an embedded version of the PCRE library.
170 As of 4.70, this is no longer the case. To compile Exim, you will need PCRE
171 installed. Most OS distributions have ready-made library and development
178 1. The internal implementation of the database keys that are used for ACL
179 ratelimiting has been tidied up. This means that an update to 4.68 might cause
180 Exim to "forget" previous rates that it had calculated, and reset them to zero.
186 1. Callouts were setting the name used for EHLO/HELO from $smtp_active_
187 hostname. This is wrong, because it relates to the incoming message (and
188 probably the interface on which it is arriving) and not to the outgoing
189 callout (which could be using a different interface). This has been
190 changed to use the value of the helo_data option from the smtp transport
191 instead - this is what is used when a message is actually being sent. If
192 there is no remote transport (possible with a router that sets up host
193 addresses), $smtp_active_hostname is used. This change is mentioned here in
194 case somebody is relying on the use of $smtp_active_hostname.
196 2. A bug has been fixed that might just possibly be something that is relied on
197 in some configurations. In expansion items such as ${if >{xxx}{yyy}...} an
198 empty string (that is {}) was being interpreted as if it was {0} and therefore
199 treated as the number zero. From release 4.64, such strings cause an error
200 because a decimal number, possibly followed by K or M, is required (as has
201 always been documented).
203 3. There has been a change to the GnuTLS support (ChangeLog/PH/20) to improve
204 Exim's performance. Unfortunately, this has the side effect of being slightly
205 non-upwards compatible for versions 4.50 and earlier. If you are upgrading from
206 one of these earlier versions and you use GnuTLS, you must remove the file
207 called gnutls-params in Exim's spool directory. If you don't do this, you will
210 TLS error on connection from ... (DH params import): Base64 decoding error.
212 Removing the file causes Exim to recompute the relevant encryption parameters
213 and cache them in the new format that was introduced for release 4.51 (May
214 2005). If you are upgrading from release 4.51 or later, there should be no
221 When an SMTP error message is specified in a "message" modifier in an ACL, or
222 in a :fail: or :defer: message in a redirect router, Exim now checks the start
223 of the message for an SMTP error code. This consists of three digits followed
224 by a space, optionally followed by an extended code of the form n.n.n, also
225 followed by a space. If this is the case and the very first digit is the same
226 as the default error code, the code from the message is used instead. If the
227 very first digit is incorrect, a panic error is logged, and the default code is
228 used. This is an incompatible change, but it is not expected to affect many (if
229 any) configurations. It is possible to suppress the use of the supplied code in
230 a redirect router by setting the smtp_error_code option false. In this case,
231 any SMTP code is quietly ignored.
237 1. The default number of ACL variables of each type has been increased to 20,
238 and it's possible to compile Exim with more. You can safely upgrade to this
239 release if you already have messages on the queue with saved ACL variable
240 values. However, if you downgrade from this release with messages on the queue,
241 any saved ACL values they may have will be lost.
243 2. The default value for rfc1413_query_timeout has been changed from 30s to 5s.
249 There was a problem with 4.52/TF/02 in that a "name=" option on control=
250 submission terminated at the next slash, thereby not allowing for slashes in
251 the name. This has been changed so that "name=" takes the rest of the string as
252 its data. It must therefore be the last option.
258 If you are using the experimental Domain Keys support, you must upgrade to
259 at least libdomainkeys 0.67 in order to run this release of Exim.
265 1. The format in which GnuTLS parameters are cached (in the file gnutls-params
266 in the spool directory) has been changed. The new format can also be generated
267 externally, so it is now possible to update the values from outside Exim. This
268 has been implemented in an upwards, BUT NOT downwards, compatible manner.
269 Upgrading should be seamless: when Exim finds that it cannot understand an
270 existing cache file, it generates new parameters and writes them to the cache
271 in the new format. If, however, you downgrade from 4.51 to a previous release,
272 you MUST delete the gnutls-params file in the spool directory, because the
273 older Exim will not recognize the new format.
275 2. When doing a callout as part of verifying an address, Exim was not paying
276 attention to any local part prefix or suffix that was matched by the router
277 that accepted the address. It now behaves in the same way as it does for
278 delivery: the affixes are removed from the local part unless
279 rcpt_include_affixes is set on the transport. If you have a configuration that
280 uses prefixes or suffixes on addresses that could be used for callouts, and you
281 want the affixes to be retained, you must make sure that rcpt_include_affixes
282 is set on the transport.
284 3. Bounce and delay warning messages no longer contain details of delivery
285 errors, except for explicit messages (e.g. generated by :fail:) and SMTP
286 responses from remote hosts.
292 The exicyclog script has been updated to use three-digit numbers in rotated log
293 files if the maximum number to keep is greater than 99. If you are already
294 keeping more than 99, there will be an incompatible change when you upgrade.
295 You will probably want to rename your old log files to the new form before
296 running the new exicyclog.
302 RFC 3848 specifies standard names for the "with" phrase in Received: header
303 lines when AUTH and/or TLS are in use. This is the "received protocol"
304 field. Exim used to use "asmtp" for authenticated SMTP, without any
305 indication (in the protocol name) for TLS use. Now it follows the RFC and
306 uses "esmtpa" if the connection is authenticated, "esmtps" if it is
307 encrypted, and "esmtpsa" if it is both encrypted and authenticated. These names
308 appear in log lines as well as in Received: header lines.
314 Change 4.31/2 gave problems to data ACLs and local_scan() functions that
315 expected to see a Received: header. I have changed to yet another scheme. The
316 Received: header is now generated after the body is received, but before the
317 ACL or local_scan() is called. After they have run, the timestamp in the
318 Received: header is updated.
320 Thus, change (a) of 4.31/2 has been reversed, but change (b) is still true,
321 which is lucky, since I decided it was a bug fix.
327 If an expansion in a condition on a "warn" statement fails because a lookup
328 defers, the "warn" statement is abandoned, and the next ACL statement is
329 processed. Previously this caused the whole ACL to be aborted.
335 Change 4.31/2 has been reversed, as it proved contentious. Recipient callout
336 verification now uses <> in the MAIL command by default, as it did before. A
337 new callout option, "use_sender", has been added to request the other
344 1. If you compile Exim to use GnuTLS, it now requires the use of release 1.0.0
345 or greater. The interface to the obsolete 0.8.x releases is no longer
346 supported. There is one externally visible change: the format for the
347 display of Distinguished Names now uses commas as a separator rather than a
348 slash. This is to comply with RFC 2253.
350 2. When a message is received, the Received: header line is now generated when
351 reception is complete, instead of at the start of reception. For messages
352 that take a long time to come in, this changes the meaning of the timestamp.
353 There are several side-effects of this change:
355 (a) If a message is rejected by a DATA or non-SMTP ACL, or by local_scan(),
356 the logged header lines no longer include the local Received: line,
357 because it has not yet been created. If the message is a non-SMTP one,
358 and the error is processed by sending a message to the sender, the copy
359 of the original message that is returned does not have an added
362 (b) When a filter file is tested using -bf, no additional Received: header
363 is added to the test message. After some thought, I decided that this
366 The contents of $received_for are not affected by this change. This
367 variable still contains the single recipient of a message, copied after
368 addresses have been rewritten, but before local_scan() is run.
370 2. Recipient callout verification, like sender verification, was using <> in
371 the MAIL FROM command. This isn't really the right thing, since the actual
372 sender may affect whether the remote host accepts the recipient or not. I
373 have changed it to use the actual sender in the callout; this means that
374 the cache record is now keyed on a recipient/sender pair, not just the
375 recipient address. There doesn't seem to be a real danger of callout loops,
376 since a callout by the remote host to check the sender would use <>.
382 1. I have abolished timeout_DNS as an error that can be detected in retry
383 rules, because it has never worked. Despite the fact that it has been
384 documented since at least release 1.62, there was no code to support it.
385 If you have used it in your retry rules, you will now get a warning message
386 to the log and panic log. It is now treated as plain "timeout".
388 2. After discussion on the mailing list, Exim no longer adds From:, Date:, or
389 Message-Id: header lines to messages that do not originate locally, that is,
390 messages that have an associated sending host address.
392 3. When looking up a host name from an IP address, Exim now tries the DNS
393 first, and only if that fails does it use gethostbyaddr() (or equivalent).
394 This change was made because on some OS, not all the names are given for
395 addresses with multiple PTR records via the gethostbyaddr() interface. The
396 order of lookup can be changed by setting host_lookup_order.
402 1. The new FIXED_NEVER_USERS build-time option creates a list of "never users"
403 that cannot be overridden. The default in the distributed EDITME is "root".
404 If for some reason you were (against advice) running deliveries as root, you
405 will have to ensure that FIXED_NEVER_USERS is not set in your
408 2. The ${quote: operator now quotes an empty string, which it did not before.
410 3. Version 4.23 saves the contents of the ACL variables with the message, so
411 that they can be used later. If one of these variables contains a newline,
412 there will be a newline character in the spool that will not be interpreted
413 correctely by a previous version of Exim. (Exim ignores keyed spool file
414 items that it doesn't understand - precisely for this kind of problem - but
415 it expects them all to be on one line.)
417 So the bottom line is: if you have newlines in your ACL variables, you
418 cannot retreat from 4.23.
424 1. The idea of the "warn" ACL verb is that it adds a header or writes to the
425 log only when "message" or "log_message" are set. However, if one of the
426 conditions was an address verification, or a call to a nested ACL, the
427 messages generated by the underlying test were being passed through. This
428 no longer happens. The underlying message is available in $acl_verify_
429 message for both "message" and "log_message" expansions, so it can be
430 passed through if needed.
432 2. The way that the $h_ (and $header_) expansions work has been changed by the
433 addition of RFC 2047 decoding. See the main documentation (the NewStuff file
434 until release 4.30, then the manual) for full details. Briefly, there are
437 $rh_xxx: and $rheader_xxx: give the original content of the header
438 line(s), with no processing at all.
440 $bh_xxx: and $bheader_xxx: remove leading and trailing white space, and
441 then decode base64 or quoted-printable "words" within the header text,
442 but do not do charset translation.
444 $h_xxx: and $header_xxx: attempt to translate the $bh_ string to a
445 standard character set.
447 If you have previously been using $h_ expansions to access the raw
448 characters, you should change to $rh_ instead.
450 3. When Exim creates an RFC 2047 encoded word in a header line, it labels it
451 with the default character set from the headers_charset option instead of
452 always using iso-8859-1.
454 4. If TMPDIR is defined in Local/Makefile (default in src/EDITME is
455 TMPDIR="/tmp"), Exim checks for the presence of an environment variable
456 called TMPDIR, and if it finds it is different, it changes its value.
458 5. Following a discussion on the list, the rules by which Exim recognises line
459 endings on incoming messages have been changed. The -dropcr and drop_cr
460 options are now no-ops, retained only for backwards compatibility. The
461 following line terminators are recognized: LF CRLF CR. However, special
462 processing applies to CR:
464 (i) The sequence CR . CR does *not* terminate an incoming SMTP message,
465 nor a local message in the state where . is a terminator.
467 (ii) If a bare CR is encountered in a header line, an extra space is added
468 after the line terminator so as not to end the header. The reasoning
469 behind this is that bare CRs in header lines are most likely either
470 to be mistakes, or people trying to play silly games.
472 6. The code for using daemon_smtp_port, local_interfaces, and the -oX options
473 has been reorganized. It is supposed to be backwards compatible, but it is
474 mentioned here just in case I've screwed up.
481 1. I have tidied and re-organized the code that uses alarm() for imposing time
482 limits on various things. It shouldn't affect anything, but if you notice
483 processes getting stuck, it may be that I've broken something.
485 2. The "arguments" log selector now also logs the current working directory
488 3. An incompatible change has been made to the appendfile transport. This
489 affects the case when it is used for file deliveries that are set up by
490 .forward and filter files. Previously, any settings of the "file" or
491 "directory" options were ignored. It is hoped that, like the address_file
492 transport in the default configuration, these options were never in fact set
493 on such transports, because they were of no use.
495 Now, if either of these options is set, it is used. The path that is passed
496 by the router is in $address_file (this is not new), so it can be used as
497 part of a longer path, or modified in any other way that expansion permits.
499 If neither "file" nor "directory" is set, the behaviour is unchanged.
501 4. Related to the above: in a filter, if a "save" command specifies a non-
502 absolute path, the value of $home/ is pre-pended. This no longer happens if
503 $home is unset or is set to an empty string.
505 5. Multiple file deliveries from a filter or .forward file can never be
506 batched; the value of batch_max on the transport is ignored for file
507 deliveries. I'm assuming that nobody ever actually set batch_max on the
508 address_file transport - it would have had odd effects previously.
510 6. DESTDIR is the more common variable that ROOT for use when installing
511 software under a different root filing system. The Exim install script now
512 recognizes DESTDIR first; if it is not set, ROOT is used.
514 7. If DESTDIR is set when installing Exim, it no longer prepends its value to
515 the path of the system aliases file that appears in the default
516 configuration (when a default configuration is installed). If an aliases
517 file is actually created, its name *does* use the prefix.
523 1. The default for the maximum number of unknown SMTP commands that Exim will
524 accept before dropping a connection has been reduced from 5 to 3. However, you
525 can now change the value by setting smtp_max_unknown_commands.
527 2. The ${quote: operator has been changed so that it turns newline and carriage
528 return characters into \n and \r, respectively.
530 3. The file names used for maildir messages now include the microsecond time
531 fraction as well as the time in seconds, to cope with systems where the process
532 id can be re-used within the same second. The format is now
534 <time>.H<microsec>P<pid>.<host>
536 This should be a compatible change, but is noted here just in case.
538 4. The rules for creating message ids have changed, to cope with systems where
539 the process id can be re-used within the same second. The format, however, is
540 unchanged, so this should not cause any problems, except as noted in the next
543 5. The maximum value for localhost_number has been reduced from 255 to 16, in
544 order to implement the new message id rules. For operating systems that have
545 case-insensitive file systems (Cygwin and Darwin), the limit is 10.
547 6. verify = header_syntax was allowing unqualified addresses in all cases. Now
548 it allows them only for locally generated messages and from hosts that match
549 sender_unqualified_hosts or recipient_unqualified_hosts, respectively.
551 7. For reasons lost in the mists of time, when a pipe transport was run, the
552 environment variable MESSAGE_ID was set to the message ID preceded by 'E' (the
553 form used in Message-ID: header lines). The 'E' has been removed.
559 1. The handling of lines in the configuration file has changed. Previously,
560 macro expansion was applied to logical lines, after continuations had been
561 joined on. This meant that it could not be used in .include lines, which are
562 handled as physical rather than logical lines. Macro expansion is now done on
563 physical lines rather than logical lines. This means there are two
566 (a) A macro that expands to # to turn a line into a comment now applies only
567 to the physical line where it appears. Previously, it would have caused
568 any following continuations also to be ignored.
570 (b) A macro name can no longer be split over the boundary between a line and
571 its continuation. Actually, this is more of a bug fix. :-)
573 2. The -D command line option must now all be within one command line item.
574 This makes it possible to use -D to set a macro to the empty string by commands
580 Previously, these items would have moved on to the next item on the command
581 line. To include spaces in a macro definition item, quotes must be used, in
582 which case you can also have spaces after -D and surrounding the equals. For
585 exim '-D ABC = something' ...
587 3. The way that addresses that redirect to themselves are handled has been
588 changed, in order to fix an obscure bug. This should not cause any problems
589 except in the case of wanting to go back from a 4.11 (or later) release to an
590 earlier release. If there are undelivered messages on the spool that contain
591 addresses which redirect to themselves, and the redirected addresses have
592 already been delivered, you might get a duplicate delivery if you revert to an
595 4. The default way of looking up IP addresses for hosts in the manualroute and
596 queryprogram routers has been changed. If "byname" or "bydns" is explicitly
597 specified, there is no change, but if no method is specified, Exim now behaves
600 First, a DNS lookup is done. If this yields anything other than
601 HOST_NOT_FOUND, that result is used. Otherwise, Exim goes on to try a call to
602 getipnodebyname() (or gethostbyname() on older systems) and the result of the
603 lookup is the result of that call.
605 This change has been made because it has been discovered that on some systems,
606 if a DNS lookup called via getipnodebyname() times out, HOST_NOT_FOUND is
607 returned instead of TRY_AGAIN. Thus, it is safest to try a DNS lookup directly
608 first, and only if that gives a definite "no such host" to try the local
611 5. In fixing the minor security problem with pid_file_path, I have removed some
612 backwards-compatible (undocumented) code which was present to ease conversion
613 from Exim 3. In Exim 4, pid_file_path is a literal; in Exim 3 it was allowed to
614 contain "%s", which was replaced by the port number for daemons listening on
615 non-standard ports. In Exim 4, such daemons do not write a pid file. The
616 backwards compatibility feature was to replace "%s" by nothing if it occurred
617 in an Exim 4 setting of pid_file_path. The bug was in this code. I have solved
618 the problem by removing the backwards compatibility feature. Thus, if you still
619 have "%s" somewhere in a setting of pid_file_path, you should remove it.
621 6. There has been an extension to lsearch files. The keys in these files may
622 now be quoted in order to allow for whitespace and colons in them. This means
623 that if you were previously using keys that began with a doublequote, you will
624 now have to wrap them with extra quotes and escape the internal quotes. The
625 possibility that anybody is actually doing this seems extremely remote, but it
626 is documented just in case.
632 The build-time parameter EXIWHAT_KILL_ARG has been renamed EXIWHAT_KILL_SIGNAL
633 to better reflect its function. The OS-specific files have been updated. Only
634 if you have explicitly set this in your Makefile (highly unlikely) do you need
git://git.exim.org / exim.git / blob