. /////////////////////////////////////////////////////////////////////////////
.set previousversion "4.75"
-.set version "4.77"
+.set version "4.80"
.set ACL "access control lists (ACLs)"
.set I " "
<bookinfo>
<title>Specification of the Exim Mail Transfer Agent</title>
<titleabbrev>The Exim MTA</titleabbrev>
-<date>06 May 2011</date>
+<date>17 May 2012</date>
<author><firstname>Exim</firstname><surname>Maintainers</surname></author>
<authorinitials>EM</authorinitials>
<revhistory><revision>
- <revnumber>4.77</revnumber>
- <date>10 Oct 2011</date>
+ <revnumber>4.80</revnumber>
+ <date>17 May 2012</date>
<authorinitials>EM</authorinitials>
</revision></revhistory>
-<copyright><year>2011</year><holder>University of Cambridge</holder></copyright>
+<copyright><year>2012</year><holder>University of Cambridge</holder></copyright>
</bookinfo>
.literal off
overridden if necessary.
+.new
.section "PCRE library" "SECTpcre"
.cindex "PCRE library"
Exim no longer has an embedded PCRE library as the vast majority of
to install the PCRE or PCRE development package for your operating
system. If your system has a normal PCRE installation the Exim build
process will need no further configuration. If the library or the
-headers are in an unusual location you will need to set the PCRE_LIBS
-and INCLUDE directives appropriately. If your operating system has no
+headers are in an unusual location you will need to either set the PCRE_LIBS
+and INCLUDE directives appropriately,
+or set PCRE_CONFIG=yes to use the installed &(pcre-config)& command.
+If your operating system has no
PCRE support then you will need to obtain and build the current PCRE
from &url(ftp://ftp.csx.cam.ac.uk/pub/software/programming/pcre/).
+More information on PCRE is available at &url(http://www.pcre.org/).
+.wen
.section "DBM libraries" "SECTdb"
.cindex "DBM libraries" "discussion of"
preference to the address taken from the message. The caller of Exim must be a
trusted user for the sender of a message to be set in this way.
+.vitem &%-bmalware%&&~<&'filename'&>
+.oindex "&%-bmalware%&"
+.cindex "testing", "malware"
+.cindex "malware scan test"
+This debugging option causes Exim to scan the given file,
+using the malware scanning framework. The option of &%av_scanner%& influences
+this option, so if &%av_scanner%&'s value is dependent upon an expansion then
+the expansion should have defaults which apply to this invocation. ACLs are
+not invoked, so if &%av_scanner%& references an ACL variable then that variable
+will never be populated and &%-bmalware%& will fail.
+
+Exim will have changed working directory before resolving the filename, so
+using fully qualified pathnames is advisable. Exim will be running as the Exim
+user when it tries to open the file, rather than as the invoking user.
+This option requires admin privileges.
+
+The &%-bmalware%& option will not be extended to be more generally useful,
+there are better tools for file-scanning. This option exists to help
+administrators verify their Exim and AV scanner configuration.
+
.vitem &%-bnq%&
.oindex "&%-bnq%&"
.cindex "address qualification, suppressing"
Exim behaves in exactly the same way as it does when receiving a message via
the listening daemon.
-.vitem &%-bmalware%&&~<&'filename'&>
-.oindex "&%-bmalware%&"
-.cindex "testing", "malware"
-.cindex "malware scan test"
-This debugging option causes Exim to scan the given file,
-using the malware scanning framework. The option of &%av_scanner%& influences
-this option, so if &%av_scanner%&'s value is dependent upon an expansion then
-the expansion should have defaults which apply to this invocation. ACLs are
-not invoked, so if &%av_scanner%& references an ACL variable then that variable
-will never be populated and &%-bmalware%& will fail.
-
-Exim will have changed working directory before resolving the filename, so
-using fully qualified pathnames is advisable. Exim will be running as the Exim
-user when it tries to open the file, rather than as the invoking user.
-This option requires admin privileges.
-
-The &%-bmalware%& option will not be extended to be more generally useful,
-there are better tools for file-scanning. This option exists to help
-administrators verify their Exim and AV scanner configuration.
-
.vitem &%-bt%&
.oindex "&%-bt%&"
.cindex "testing" "addresses"
than a recipient address. This affects any rewriting and qualification that
might happen.
+.vitem &%-bw%&
+.oindex "&%-bw%&"
+.cindex "daemon"
+.cindex "inetd"
+.cindex "inetd" "wait mode"
+This option runs Exim as a daemon, awaiting incoming SMTP connections,
+similarly to the &%-bd%& option. All port specifications on the command-line
+and in the configuration file are ignored. Queue-running may not be specified.
+
+In this mode, Exim expects to be passed a socket as fd 0 (stdin) which is
+listening for connections. This permits the system to start up and have
+inetd (or equivalent) listen on the SMTP ports, starting an Exim daemon for
+each port only when the first connection is received.
+
+If the option is given as &%-bw%&<&'time'&> then the time is a timeout, after
+which the daemon will exit, which should cause inetd to listen once more.
+
.vitem &%-C%&&~<&'filelist'&>
.oindex "&%-C%&"
.cindex "configuration file" "alternate"
i.e. PLAIN or LOGIN. The &%server_advertise_condition%& setting controls
when Exim offers authentication to clients; in the examples, this is only
when TLS or SSL has been started, so to enable the authenticators you also
-need to add support for TLS as described in &<<SECTdefconfmain>>&.
+need to add support for TLS as described in section &<<SECTdefconfmain>>&.
The &%server_condition%& setting defines how to verify that the username and
password are correct. In the examples it just produces an error message.
To make the authenticators work, you can use a string expansion
-expression like one of the examples in &<<CHAPplaintext>>&.
+expression like one of the examples in chapter &<<CHAPplaintext>>&.
Beware that the sequence of the parameters to PLAIN and LOGIN differ; the
-usercode and password are in different positions. &<<CHAPplaintext>>&
-covers both.
+usercode and password are in different positions.
+Chapter &<<CHAPplaintext>>& covers both.
.ecindex IIDconfiwal
&`fail`& keyword causes a &'forced expansion failure'& &-- see section
&<<SECTforexpfai>>& for an explanation of what this means.
-The supported DNS record types are A, CNAME, MX, NS, PTR, SRV, and TXT, and,
-when Exim is compiled with IPv6 support, AAAA (and A6 if that is also
+.new
+The supported DNS record types are A, CNAME, MX, NS, PTR, SPF, SRV, and TXT,
+and, when Exim is compiled with IPv6 support, AAAA (and A6 if that is also
configured). If no type is given, TXT is assumed. When the type is PTR,
the data can be an IP address, written as normal; inversion and the addition of
&%in-addr.arpa%& or &%ip6.arpa%& happens automatically. For example:
+.wen
.code
${lookup dnsdb{ptr=192.168.4.5}{$value}fail}
.endd
white space is ignored.
.cindex "TXT record" "in &(dnsdb)& lookup"
+.cindex "SPF record" "in &(dnsdb)& lookup"
+.new
For TXT records with multiple items of data, only the first item is returned,
unless a separator for them is specified using a comma after the separator
character followed immediately by the TXT record item separator. To concatenate
-items without a separator, use a semicolon instead.
+items without a separator, use a semicolon instead. For SPF records the
+default behaviour is to concatenate multiple items without using a separator.
+.wen
.code
${lookup dnsdb{>\n,: txt=a.b.example}}
${lookup dnsdb{>\n; txt=a.b.example}}
+${lookup dnsdb{spf=example.org}}
.endd
It is permitted to specify a space as the separator character. Further
white space is ignored.
permitted. This can be useful when processing numbers extracted from dates or
times, which often do have leading zeros.
-A number may be followed by &"K"& or &"M"& to multiply it by 1024 or 1024*1024,
+A number may be followed by &"K"&, &"M"& or &"G"& to multiply it by 1024, 1024*1024
+or 1024*1024*1024,
respectively. Negative numbers are supported. The result of the computation is
-a decimal representation of the answer (without &"K"& or &"M"&). For example:
+a decimal representation of the answer (without &"K"&, &"M"& or &"G"&). For example:
.display
&`${eval:1+1} `& yields 2
supplied number and is at least 0. The quality of this randomness depends
on how Exim was built; the values are not suitable for keying material.
If Exim is linked against OpenSSL then RAND_pseudo_bytes() is used.
+.new
+If Exim is linked against GnuTLS then gnutls_rnd(GNUTLS_RND_NONCE) is used,
+for versions of GnuTLS with that function.
+.wen
Otherwise, the implementation may be arc4random(), random() seeded by
srandomdev() or srandom(), or a custom implementation even weaker than
random().
The value will be retained for the lifetime of the message. During outbound
SMTP deliveries, it reflects the value of the &%tls_sni%& option on
the transport.
-
-This is currently only available when using OpenSSL, built with support for
-SNI.
.wen
.vitem &$tod_bsdinbox$&
.vindex "&$tod_epoch$&"
The time and date as a number of seconds since the start of the Unix epoch.
+.vitem &$tod_epoch_l$&
+.vindex "&$tod_epoch_l$&"
+The time and date as a number of microseconds since the start of the Unix epoch.
+
.vitem &$tod_full$&
.vindex "&$tod_full$&"
A full version of the time and date, for example: Wed, 16 Oct 1995 09:51:40
.section "TLS" "SECID108"
.table2
-.row &%gnutls_require_kx%& "control GnuTLS key exchanges"
-.row &%gnutls_require_mac%& "control GnuTLS MAC algorithms"
-.row &%gnutls_require_protocols%& "control GnuTLS protocols"
.row &%gnutls_compat_mode%& "use GnuTLS compatibility mode"
.row &%openssl_options%& "adjust OpenSSL compatibility options"
.row &%tls_advertise_hosts%& "advertise TLS to these hosts"
.row &%tls_certificate%& "location of server certificate"
.row &%tls_crl%& "certificate revocation list"
+.row &%tls_dh_max_bits%& "clamp D-H bit count suggestion"
.row &%tls_dhparam%& "DH parameters for server"
.row &%tls_on_connect_ports%& "specify SSMTP (SMTPS) ports"
.row &%tls_privatekey%& "location of server private key"
Those options that undergo string expansion before use are marked with
†.
-.option accept_8bitmime main boolean false
+.new
+.option accept_8bitmime main boolean true
.cindex "8BITMIME"
.cindex "8-bit characters"
This option causes Exim to send 8BITMIME in its response to an SMTP
EHLO command, and to accept the BODY= parameter on MAIL commands.
However, though Exim is 8-bit clean, it is not a protocol converter, and it
takes no steps to do anything special with messages received by this route.
-Consequently, this option is turned off by default.
+
+Historically Exim kept this option off by default, but the maintainers
+feel that in today's Internet, this causes more problems than it solves.
+It now defaults to true.
+A more detailed analysis of the issues is provided by Dan Bernstein:
+.display
+&url(http://cr.yp.to/smtp/8bitmime.html)
+.endd
+.wen
.option acl_not_smtp main string&!! unset
.cindex "&ACL;" "for non-SMTP messages"
See &%gecos_name%& above.
-.option gnutls_require_kx main string unset
-This option controls the key exchange mechanisms when GnuTLS is used in an Exim
-server. For details, see section &<<SECTreqciphgnu>>&.
-
-.option gnutls_require_mac main string unset
-This option controls the MAC algorithms when GnuTLS is used in an Exim
-server. For details, see section &<<SECTreqciphgnu>>&.
-
-.option gnutls_require_protocols main string unset
-This option controls the protocols when GnuTLS is used in an Exim
-server. For details, see section &<<SECTreqciphgnu>>&.
-
.option gnutls_compat_mode main boolean unset
This option controls whether GnuTLS is used in compatibility mode in an Exim
server. This reduces security slightly, but improves interworking with older
transport driver.
- .option openssl_options main "string list" unset
+ .option openssl_options main "string list" "+no_sslv2"
.cindex "OpenSSL "compatibility options"
This option allows an administrator to adjust the SSL options applied
by OpenSSL to connections. It is given as a space-separated list of items,
invoking Exim with the &%-bV%& flag.
.new
-Historical note: prior to release 4.78, Exim defaulted this value to
+Historical note: prior to release 4.80, Exim defaulted this value to
"+dont_insert_empty_fragments", which may still be needed for compatibility
with some clients, but which lowers security by increasing exposure to
some now infamous attacks.
.wen
+.new
+.option tls_dh_max_bits main integer 2236
+.cindex "TLS" "D-H bit count"
+The number of bits used for Diffie-Hellman key-exchange may be suggested by
+the chosen TLS library. That value might prove to be too high for
+interoperability. This option provides a maximum clamp on the value
+suggested, trading off security for interoperability.
+
+The value must be at least 1024.
+
+The value 2236 was chosen because, at time of adding the option, it was the
+hard-coded maximum value supported by the NSS cryptographic library, as used
+by Thunderbird, while GnuTLS was suggesting 2432 bits as normal.
+
+If you prefer more security and are willing to break some clients, raise this
+number.
+
+Note that the value passed to GnuTLS for *generating* a new prime may be a
+little less than this figure, because GnuTLS is inexact and may produce a
+larger prime than requested.
+.wen
+
+
.option tls_dhparam main string&!! unset
.cindex "TLS" "D-H parameters for server"
-The value of this option is expanded, and must then be the absolute path to
-a file which contains the server's DH parameter values.
-This is used only for OpenSSL. When Exim is linked with GnuTLS, this option is
-ignored. See section &<<SECTopenvsgnu>>& for further details.
+.new
+The value of this option is expanded and indicates the source of DH parameters
+to be used by Exim.
+
+If it is a filename starting with a &`/`&, then it names a file from which DH
+parameters should be loaded. If the file exists, it should hold a PEM-encoded
+PKCS#3 representation of the DH prime. If the file does not exist, for
+OpenSSL it is an error. For GnuTLS, Exim will attempt to create the file and
+fill it with a generated DH prime. For OpenSSL, if the DH bit-count from
+loading the file is greater than &%tls_dh_max_bits%& then it will be ignored,
+and treated as though the &%tls_dhparam%& were set to "none".
+
+If this option expands to the string "none", then no DH parameters will be
+loaded by Exim.
+
+If this option expands to the string "historic" and Exim is using GnuTLS, then
+Exim will attempt to load a file from inside the spool directory. If the file
+does not exist, Exim will attempt to create it.
+See section &<<SECTgnutlsparam>>& for further details.
+
+If Exim is using OpenSSL and this option is empty or unset, then Exim will load
+a default DH prime; the default is the 2048 bit prime described in section
+2.2 of RFC 5114, "2048-bit MODP Group with 224-bit Prime Order Subgroup", which
+in IKE is assigned number 23.
+
+Otherwise, the option must expand to the name used by Exim for any of a number
+of DH primes specified in RFC 2409, RFC 3526 and RFC 5114. As names, Exim uses
+"ike" followed by the number used by IKE, of "default" which corresponds to
+"ike23".
+
+The available primes are:
+&`ike1`&, &`ike2`&, &`ike5`&,
+&`ike14`&, &`ike15`&, &`ike16`&, &`ike17`&, &`ike18`&,
+&`ike22`&, &`ike23`& (aka &`default`&) and &`ike24`&.
+
+Some of these will be too small to be accepted by clients.
+Some may be too large to be accepted by clients.
+.wen
.option tls_on_connect_ports main "string list" unset
case routing fails.
+.new
+.section "Declining addresses by dnslookup" "SECTdnslookupdecline"
+.cindex "&(dnslookup)& router" "declines"
+There are a few cases where a &(dnslookup)& router will decline to accept
+an address; if such a router is expected to handle "all remaining non-local
+domains", then it is important to set &%no_more%&.
+
+Reasons for a &(dnslookup)& router to decline currently include:
+.ilist
+The domain does not exist in DNS
+.next
+The domain exists but the MX record's host part is just "."; this is a common
+convention (borrowed from SRV) used to indicate that there is no such service
+for this domain and to not fall back to trying A/AAAA records.
+.next
+Ditto, but for SRV records, when &%check_srv%& is set on this router.
+.next
+MX record points to a non-existent host.
+.next
+MX record points to an IP address and the main section option
+&%allow_mx_to_ip%& is not set.
+.next
+MX records exist and point to valid hosts, but all hosts resolve only to
+addresses blocked by the &%ignore_target_hosts%& generic option on this router.
+.next
+The domain is not syntactically valid (see also &%allow_utf8_domains%& and
+&%dns_check_names_pattern%& for handling one variant of this)
+.next
+&%check_secondary_mx%& is set on this router but the local host can
+not be found in the MX records (see below)
+.endlist
+.wen
+
+
.section "Private options for dnslookup" "SECID118"
instead of using the DNS. Of course, that function may in fact use the DNS, but
it may also consult other sources of information such as &_/etc/hosts_&.
-.option gnutls_require_kx smtp string unset
-This option controls the key exchange mechanisms when GnuTLS is used in an Exim
-client. For details, see section &<<SECTreqciphgnu>>&.
-
-.option gnutls_require_mac smtp string unset
-This option controls the MAC algorithms when GnuTLS is used in an Exim
-client. For details, see section &<<SECTreqciphgnu>>&.
-
-.option gnutls_require_protocols smtp string unset
-This option controls the protocols when GnuTLS is used in an Exim
-client. For details, see section &<<SECTreqciphgnu>>&.
-
.option gnutls_compat_mode smtp boolean unset
This option controls whether GnuTLS is used in compatibility mode in an Exim
server. This reduces security slightly, but improves interworking with older
.cindex "authentication" "CRAM-MD5"
.cindex "authentication" "SCRAM-SHA-1"
The &(gsasl)& authenticator provides server integration for the GNU SASL
-library and the mechanisms it provides. This is new as of the 4.78 release
+library and the mechanisms it provides. This is new as of the 4.80 release
and there are a few areas where the library does not let Exim smoothly
scale to handle future authentication mechanisms, so no guarantee can be
made that any particular new authentication mechanism will be supported
without code changes in Exim.
-.option server_channelbinding gsasl bool false
+.option server_channelbinding gsasl boolean false
Some authentication mechanisms are able to use external context at both ends
of the session to bind the authentication to that context, and fail the
authentication process if that context differs. Specifically, some TLS
The &%tls_verify_certificates%& option must contain the name of a file, not the
name of a directory (for OpenSSL it can be either).
.next
-The &%tls_dhparam%& option is ignored, because early versions of GnuTLS had no
-facility for varying its Diffie-Hellman parameters. I understand that this has
-changed, but Exim has not been updated to provide this facility.
+.new
+The default value for &%tls_dhparam%& differs for historical reasons.
+.wen
.next
.vindex "&$tls_peerdn$&"
Distinguished Name (DN) strings reported by the OpenSSL library use a slash for
affects the value of the &$tls_peerdn$& variable.
.next
OpenSSL identifies cipher suites using hyphens as separators, for example:
-DES-CBC3-SHA. GnuTLS uses underscores, for example: RSA_ARCFOUR_SHA. What is
-more, OpenSSL complains if underscores are present in a cipher list. To make
-life simpler, Exim changes underscores to hyphens for OpenSSL and hyphens to
-underscores for GnuTLS when processing lists of cipher suites in the
+DES-CBC3-SHA. GnuTLS historically used underscores, for example:
+RSA_ARCFOUR_SHA. What is more, OpenSSL complains if underscores are present
+in a cipher list. To make life simpler, Exim changes underscores to hyphens
+for OpenSSL and passes the string unchanged to GnuTLS (expecting the library
+to handle its own older variants) when processing lists of cipher suites in the
&%tls_require_ciphers%& options (the global option and the &(smtp)& transport
option).
.next
.endlist
-.section "GnuTLS parameter computation" "SECID181"
+.section "GnuTLS parameter computation" "SECTgnutlsparam"
+.new
+This section only applies if &%tls_dhparam%& is set to &`historic`& or to
+an explicit path; if the latter, then the text about generation still applies,
+but not the chosen filename.
+By default, as of Exim 4.80 a hard-coded D-H prime is used.
+See the documentation of &%tls_dhparam%& for more information.
+
GnuTLS uses D-H parameters that may take a substantial amount of time
to compute. It is unreasonable to re-compute them for every TLS session.
Therefore, Exim keeps this data in a file in its spool directory, called
-&_gnutls-params_&. The file is owned by the Exim user and is readable only by
+&_gnutls-params-NNNN_& for some value of NNNN, corresponding to the number
+of bits requested.
+The file is owned by the Exim user and is readable only by
its owner. Every Exim process that start up GnuTLS reads the D-H
parameters from this file. If the file does not exist, the first Exim process
that needs it computes the data and writes it to a temporary file which is
For maximum security, the parameters that are stored in this file should be
recalculated periodically, the frequency depending on your paranoia level.
+If you are avoiding using the fixed D-H primes published in RFCs, then you
+are concerned about some advanced attacks and will wish to do this; if you do
+not regenerate then you might as well stick to the standard primes.
+
Arranging this is easy in principle; just delete the file when you want new
values to be computed. However, there may be a problem. The calculation of new
parameters needs random numbers, and these are obtained from &_/dev/random_&.
a substantial amount of time, causing timeouts on incoming connections.
The solution is to generate the parameters externally to Exim. They are stored
-in &_gnutls-params_& in PEM format, which means that they can be generated
-externally using the &(certtool)& command that is part of GnuTLS.
+in &_gnutls-params-N_& in PEM format, which means that they can be
+generated externally using the &(certtool)& command that is part of GnuTLS.
To replace the parameters with new ones, instead of deleting the file
and letting Exim re-create it, you can generate new parameters using
&(certtool)& and, when this has been done, replace Exim's cache file by
renaming. The relevant commands are something like this:
.code
+# ls
+[ look for file; assume gnutls-params-2236 is the most recent ]
# rm -f new-params
# touch new-params
# chown exim:exim new-params
+# chmod 0600 new-params
+# certtool --generate-dh-params --bits 2236 >>new-params
+# openssl dhparam -noout -text -in new-params | head
+[ check the first line, make sure it's not more than 2236;
+ if it is, then go back to the start ("rm") and repeat
+ until the size generated is at most the size requested ]
# chmod 0400 new-params
-# certtool --generate-privkey --bits 512 >new-params
-# echo "" >>new-params
-# certtool --generate-dh-params --bits 1024 >> new-params
-# mv new-params gnutls-params
+# mv new-params gnutls-params-2236
.endd
If Exim never has to generate the parameters itself, the possibility of
stalling is removed.
+The filename changed in Exim 4.80, to gain the -bits suffix. The value which
+Exim will choose depends upon the version of GnuTLS in use. For older GnuTLS,
+the value remains hard-coded in Exim as 1024. As of GnuTLS 2.12.x, there is
+a way for Exim to ask for the "normal" number of bits for D-H public-key usage,
+and Exim does so. This attempt to remove Exim from TLS policy decisions
+failed, as GnuTLS 2.12 returns a value higher than the current hard-coded limit
+of the NSS library. Thus Exim gains the &%tls_dh_max_bits%& global option,
+which applies to all D-H usage, client or server. If the value returned by
+GnuTLS is greater than &%tls_dh_max_bits%& then the value will be clamped down
+to &%tls_dh_max_bits%&. The default value has been set at the current NSS
+limit, which is still much higher than Exim historically used.
+
+The filename and bits used will change as the GnuTLS maintainers change the
+value for their parameter &`GNUTLS_SEC_PARAM_NORMAL`&, as clamped by
+&%tls_dh_max_bits%&. At the time of writing (mid 2012), GnuTLS 2.12 recommends
+2432 bits, while NSS is limited to 2236 bits.
+
+In fact, the requested value will be *lower* than &%tls_dh_max_bits%&, to
+increase the chance of the generated prime actually being within acceptable
+bounds, as GnuTLS has been observed to overshoot. Note the check step in the
+procedure above. There is no sane procedure available to Exim to double-check
+the size of the generated prime, so it might still be too large.
+.wen
+
.section "Requiring specific ciphers in OpenSSL" "SECTreqciphssl"
.cindex "TLS" "requiring specific ciphers (OpenSSL)"
suites before the cipher negotiation takes place. This specifies which ciphers
are acceptable. The list is colon separated and may contain names like
DES-CBC3-SHA. Exim passes the expanded value of &%tls_require_ciphers%&
-directly to this function call. The following quotation from the OpenSSL
+directly to this function call.
+Many systems will install the OpenSSL manual-pages, so you may have
+&'ciphers(1)'& available to you.
+The following quotation from the OpenSSL
documentation specifies what forms of item are allowed in the cipher string:
.ilist
not be moved to the end of the list.
.endlist
+.new
+The OpenSSL &'ciphers(1)'& command may be used to test the results of a given
+string:
+.code
+# note single-quotes to get ! past any shell history expansion
+$ openssl ciphers 'HIGH:!MD5:!SHA1'
+.endd
+
+This example will let the library defaults be permitted on the MX port, where
+there's probably no identity verification anyway, but ups the ante on the
+submission ports where the administrator might have some influence on the
+choice of clients used:
+.code
+# OpenSSL variant; see man ciphers(1)
+tls_require_ciphers = ${if =={$received_port}{25}\
+ {DEFAULT}\
+ {HIGH:!MD5:!SHA1}}
+.endd
+.wen
+
+.new
.section "Requiring specific ciphers or other parameters in GnuTLS" &&&
"SECTreqciphgnu"
.cindex "GnuTLS" "specifying parameters for"
.cindex "TLS" "specifying key exchange methods (GnuTLS)"
.cindex "TLS" "specifying MAC algorithms (GnuTLS)"
.cindex "TLS" "specifying protocols (GnuTLS)"
+.cindex "TLS" "specifying priority string (GnuTLS)"
.oindex "&%tls_require_ciphers%&" "GnuTLS"
-The GnuTLS library allows the caller to specify separate lists of permitted key
-exchange methods, main cipher algorithms, MAC algorithms, and protocols.
-Unfortunately, these lists are numerical, and the library does not have a
-function for turning names into numbers. Consequently, lists of recognized
-names have to be built into the application. The permitted key exchange
-methods, ciphers, and MAC algorithms may be used in any combination to form a
-cipher suite. This is unlike OpenSSL, where complete cipher suite names are
-passed to its control function.
-
-For compatibility with OpenSSL, the &%tls_require_ciphers%& option can be set
-to complete cipher suite names such as RSA_ARCFOUR_SHA, but for GnuTLS this
-option controls only the cipher algorithms. Exim searches each item in the
-list for the name of an available algorithm. For example, if the list
-contains RSA_AES_SHA, then AES is recognized, and the behaviour is exactly
-the same as if just AES were given.
-
-.oindex "&%gnutls_require_kx%&"
-.oindex "&%gnutls_require_mac%&"
-.oindex "&%gnutls_require_protocols%&"
-There are additional options called &%gnutls_require_kx%&,
-&%gnutls_require_mac%&, and &%gnutls_require_protocols%& that can be used to
-restrict the key exchange methods, MAC algorithms, and protocols, respectively.
-These options are ignored if OpenSSL is in use.
-
-All four options are available as global options, controlling how Exim
-behaves as a server, and also as options of the &(smtp)& transport, controlling
-how Exim behaves as a client. All the values are string expanded. After
-expansion, the values must be colon-separated lists, though the separator
-can be changed in the usual way.
-
-Each of the four lists starts out with a default set of algorithms. If the
-first item in a list does &'not'& start with an exclamation mark, all the
-default items are deleted. In this case, only those that are explicitly
-specified can be used. If the first item in a list &'does'& start with an
-exclamation mark, the defaults are left on the list.
-
-Then, any item that starts with an exclamation mark causes the relevant
-entry to be removed from the list, and any item that does not start with an
-exclamation mark causes a new entry to be added to the list. Unrecognized
-items in the list are ignored. Thus:
-.code
-tls_require_ciphers = !ARCFOUR
-.endd
-allows all the defaults except ARCFOUR, whereas
-.code
-tls_require_ciphers = AES : 3DES
-.endd
-allows only cipher suites that use AES or 3DES.
-
-For &%tls_require_ciphers%& the recognized names are AES_256, AES_128, AES
-(both of the preceding), 3DES, ARCFOUR_128, ARCFOUR_40, and ARCFOUR (both of
-the preceding). The default list does not contain all of these; it just has
-AES_256, AES_128, 3DES, and ARCFOUR_128.
-
-For &%gnutls_require_kx%&, the recognized names are DHE_RSA, RSA (which
-includes DHE_RSA), DHE_DSS, and DHE (which includes both DHE_RSA and
-DHE_DSS). The default list contains RSA, DHE_DSS, DHE_RSA.
-
-For &%gnutls_require_mac%&, the recognized names are SHA (synonym SHA1), and
-MD5. The default list contains SHA, MD5.
-
-.new
-For &%gnutls_require_protocols%&, the recognized names are TLS1.2, TLS1.1,
-TLS1.0, (TLS1) and SSL3.
-The default list contains TLS1.2, TLS1.1, TLS1.0, SSL3.
-TLS1 is an alias for TLS1.0, for backwards compatibility.
-For sufficiently old versions of the GnuTLS library, TLS1.2 or TLS1.1 might
-not be supported and will not be recognised by Exim.
+The GnuTLS library allows the caller to provide a "priority string", documented
+as part of the &[gnutls_priority_init]& function. This is very similar to the
+ciphersuite specification in OpenSSL.
+
+The &%tls_require_ciphers%& option is treated as the GnuTLS priority string.
+
+The &%tls_require_ciphers%& option is available both as an global option,
+controlling how Exim behaves as a server, and also as an option of the
+&(smtp)& transport, controlling how Exim behaves as a client. In both cases
+the value is string expanded. The resulting string is not an Exim list and
+the string is given to the GnuTLS library, so that Exim does not need to be
+aware of future feature enhancements of GnuTLS.
+
+Documentation of the strings accepted may be found in the GnuTLS manual, under
+"Priority strings". This is online as
+&url(http://www.gnu.org/software/gnutls/manual/html_node/Priority-Strings.html),
+but beware that this relates to GnuTLS 3, which may be newer than the version
+installed on your system. If you are using GnuTLS 3,
+&url(http://www.gnu.org/software/gnutls/manual/html_node/Listing-the-ciphersuites-in-a-priority-string.html, then the example code)
+on that site can be used to test a given string.
+
+Prior to Exim 4.80, an older API of GnuTLS was used, and Exim supported three
+additional options, "&%gnutls_require_kx%&", "&%gnutls_require_mac%&" and
+"&%gnutls_require_protocols%&". &%tls_require_ciphers%& was an Exim list.
+
+This example will let the library defaults be permitted on the MX port, where
+there's probably no identity verification anyway, and lowers security further
+by increasing compatibility; but this ups the ante on the submission ports
+where the administrator might have some influence on the choice of clients
+used:
+.code
+# GnuTLS variant
+tls_require_ciphers = ${if =={$received_port}{25}\
+ {NORMAL:%COMPAT}\
+ {SECURE128}}
+.endd
.wen
-In a server, the order of items in these lists is unimportant. The server
-advertises the availability of all the relevant cipher suites. However, in a
-client, the order in the &%tls_require_ciphers%& list specifies a preference
-order for the cipher algorithms. The first one in the client's list that is
-also advertised by the server is tried first. The default order is as listed
-above.
-
-
.section "Configuring an Exim server to use TLS" "SECID182"
.cindex "TLS" "configuring an Exim server"
tls_dhparam = /some/file/name
.endd
is set, the SSL library is initialized for the use of Diffie-Hellman ciphers
-with the parameters contained in the file. This increases the set of cipher
-suites that the server supports. See the command
+with the parameters contained in the file.
+.new
+Set this to &`none`& to disable use of DH entirely, by making no prime
+available:
+.code
+tls_dhparam = none
+.endd
+This may also be set to a string identifying a standard prime to be used for
+DH; if it is set to &`default`& or, for OpenSSL, is unset, then the prime
+used is &`ike23`&. There are a few standard primes available, see the
+documetnation for &%tls_dhparam%& for the complete list.
+
+See the command
.code
openssl dhparam
.endd
-for a way of generating this data. At present, &%tls_dhparam%& is used only
-when Exim is linked with OpenSSL. It is ignored if GnuTLS is being used.
+for a way of generating file data.
+.wen
The strings supplied for these three options are expanded every time a client
host connects. It is therefore possible to use different certificates and keys
The Exim developers are proceeding cautiously and so far no other TLS options
are re-expanded.
-Currently SNI support is only available if using OpenSSL, with TLS Extensions
-support enabled therein.
+When Exim is built againt OpenSSL, OpenSSL must have been built with support
+for TLS Extensions. This holds true for OpenSSL 1.0.0+ and 0.9.8+ with
+enable-tlsext in EXTRACONFIGURE. If you invoke &(openssl s_client -h)& and
+see &`-servername`& in the output, then OpenSSL has support.
+
+When Exim is built against GnuTLS, SNI support is available as of GnuTLS
+0.5.10. (Its presence predates the current API which Exim uses, so if Exim
+built, then you have SNI support).
.wen
In order to see the contents of messages on the queue, and to operate on them,
&'eximon'& must either be run as root or by an admin user.
+The command-line parameters of &'eximon'& are passed to &_eximon.bin_& and may
+contain X11 resource parameters interpreted by the X11 library. In addition,
+if the first parameter starts with the string "gdb" then it is removed and the
+binary is invoked under gdb (the parameter is used as the gdb command-name, so
+versioned variants of gdb can be invoked).
+
The monitor's window is divided into three parts. The first contains one or
more stripcharts and two action buttons, the second contains a &"tail"& of the
main log file, and the third is a display of the queue of messages awaiting
Change log file for Exim from version 4.21
-------------------------------------------
-Exim version 4.78
+Exim version 4.80
-----------------
PP/01 Handle short writes when writing local log-files.
non-compliant senders.
Analysis and variant patch by Todd Lyons.
-NM/04 Bugzilla 1237 - fix cases where printf format usage no indicated
+NM/04 Bugzilla 1237 - fix cases where printf format usage not indicated
Bug report from Lars Müller <lars@samba.org> (via SUSE),
Patch from Dirk Mueller <dmueller@suse.com>
JH/01 Bugzilla 660 - Multi-valued attributes from ldap now parseable as a
comma-sep list; embedded commas doubled.
+JH/02 Refactored ACL "verify =" logic to table-driven dispatch.
+
PP/15 LDAP: Check for errors of TLS initialisation, to give correct
diagnostics.
Report and patch from Dmitry Banschikov.
Not seeing resolver debug output on NetBSD, but suspect this is a
resolver implementation change.
-PP/20 Disable SSLv2 by default in OpenSSL support.
+PP/20 Revert part of NM/04, it broke log_path containing %D expansions.
+ Left warnings. Added "eximon gdb" invocation mode.
+
+PP/21 Defaulting "accept_8bitmime" to true, not false.
+
+PP/22 Added -bw for inetd wait mode support.
+
+PP/23 Added PCRE_CONFIG=yes support to Makefile for using pcre-config to
+ locate the relevant includes and libraries. Made this the default.
+
+PP/24 Fixed headers_only on smtp transports (was not sending trailing dot).
+ Bugzilla 1246, report and most of solution from Tomasz Kusy.
+
+JH/03 ${eval } now uses 64-bit and supports a "g" suffix (like to "k" and "m").
+ This may cause build issues on older platforms.
+
+PP/25 Revamped GnuTLS support, passing tls_require_ciphers to
+ gnutls_priority_init, ignoring Exim options gnutls_require_kx,
+ gnutls_require_mac & gnutls_require_protocols (no longer supported).
+ Added SNI support via GnuTLS too.
+ Made ${randint:..} supplier available, if using not-too-old GnuTLS.
+
+PP/26 Added EXPERIMENTAL_OCSP for OpenSSL.
+
+PP/27 Applied dnsdb SPF support patch from Janne Snabb.
+ Applied second patch from Janne, implementing suggestion to default
+ multiple-strings-in-record handling to match SPF spec.
+
+JH/04 Added expansion variable $tod_epoch_l for a higher-precision time.
+
+PP/28 Fix DCC dcc_header content corruption (stack memory referenced,
+ read-only, out of scope).
+ Patch from Wolfgang Breyha, report from Stuart Northfield.
+
+PP/29 Fix three issues highlighted by clang analyser static analysis.
+ Only crash-plausible issue would require the Cambridge-specific
+ iplookup router and a misconfiguration.
+ Report from Marcin Mirosław.
+
+PP/30 Another attempt to deal with PCRE_PRERELEASE, this one less buggy.
+
+PP/31 %D in printf continues to cause issues (-Wformat=security), so for
+ now guard some of the printf checks behind WANT_DEEPER_PRINTF_CHECKS.
+ As part of this, removing so much warning spew let me fix some minor
+ real issues in debug logging.
+
+PP/32 GnuTLS was always using default tls_require_ciphers, due to a missing
+ assignment on my part. Fixed.
+
+PP/33 Added tls_dh_max_bits option, defaulting to current hard-coded limit
+ of NSS, for GnuTLS/NSS interop. Problem root cause diagnosis by
+ Janne Snabb (who went above and beyond: thank you).
+
+PP/34 Validate tls_require_ciphers on startup, since debugging an invalid
+ string otherwise requires a connection and a bunch more work and it's
+ relatively easy to get wrong. Should also expose TLS library linkage
+ problems.
+
+PP/35 Pull in <features.h> on Linux, for some portability edge-cases of
+ 64-bit ${eval} (JH/03).
+
+PP/36 Define _GNU_SOURCE in exim.h; it's needed for some releases of
+ GNU libc to support some of the 64-bit stuff, should not lead to
+ conflicts. Defined before os.h is pulled in, so if a given platform
+ needs to override this, it can.
+
+PP/37 Unbreak Cyrus SASL auth: SSF retrieval was incorrect, Exim thought
+ protection layer was required, which is not implemented.
+ Bugzilla 1254, patch from Wolfgang Breyha.
+
+PP/38 Overhaul DH prime handling, supply RFC-specified DH primes as built
+ into Exim, default to IKE id 23 from RFC 5114 (2048 bit). Make
+ tls_dhparam take prime identifiers. Also unbreak combination of
+ OpenSSL+DH_params+TLSSNI.
+
++PP/39 Disable SSLv2 by default in OpenSSL support.
+
Exim version 4.77
-----------------
test from the snapshots or the CVS before the documentation is updated. Once
the documentation is updated, this file is reduced to a short list.
-Version 4.78
+Version 4.80
------------
1. New authenticator driver, "gsasl". Server-only (at present).
"LOOKUP_LIBS" directly. Similarly for handling the TLS library support
without adjusting "TLS_INCLUDE" and "TLS_LIBS".
+ In addition, setting PCRE_CONFIG=yes will query the pcre-config tool to
+ find the headers and libraries for PCRE.
+
4. New expansion variable $tls_bits.
5. New lookup type, "dbmjz". Key is an Exim list, the elements of which will
A new log_selector, +tls_sni, has been added, to log received SNI values
for Exim as a server.
- Currently OpenSSL only.
+ 8. The existing "accept_8bitmime" option now defaults to true. This means
+ that Exim is deliberately not strictly RFC compliant. We're following
+ Dan Bernstein's advice in http://cr.yp.to/smtp/8bitmime.html by default.
+ Those who disagree, or know that they are talking to mail servers that,
+ even today, are not 8-bit clean, need to turn off this option.
+
+ 9. Exim can now be started with -bw (with an optional timeout, given as
+ -bw<timespec>). With this, stdin at startup is a socket that is
+ already listening for connections. This has a more modern name of
+ "socket activation", but forcing the activated socket to fd 0. We're
+ interested in adding more support for modern variants.
+
+10. ${eval } now uses 64-bit values on supporting platforms. A new "G" suffix
+ for numbers indicates multiplication by 1024^3.
+
+11. The GnuTLS support has been revamped; the three options gnutls_require_kx,
+ gnutls_require_mac & gnutls_require_protocols are no longer supported.
+ tls_require_ciphers is now parsed by gnutls_priority_init(3) as a priority
+ string, documentation for which is at:
+ http://www.gnu.org/software/gnutls/manual/html_node/Priority-Strings.html
+
+ SNI support has been added to Exim's GnuTLS integration too.
+
+ For sufficiently recent GnuTLS libraries, ${randint:..} will now use
+ gnutls_rnd(), asking for GNUTLS_RND_NONCE level randomness.
+
+12. With OpenSSL, if built with EXPERIMENTAL_OCSP, a new option tls_ocsp_file
+ is now available. If the contents of the file are valid, then Exim will
+ send that back in response to a TLS status request; this is OCSP Stapling.
+ Exim will not maintain the contents of the file in any way: administrators
+ are responsible for ensuring that it is up-to-date.
+
+ See "experimental-spec.txt" for more details.
+
+13. ${lookup dnsdb{ }} supports now SPF record types. They are handled
+ identically to TXT record lookups.
+
+14. New expansion variable $tod_epoch_l for higher-precision time.
+
+15. New global option tls_dh_max_bits, defaulting to current value of NSS
+ hard-coded limit of DH ephemeral bits, to fix interop problems caused by
+ GnuTLS 2.12 library recommending a bit count higher than NSS supports.
+
+16. tls_dhparam now used by both OpenSSL and GnuTLS, can be path or identifier.
+ Option can now be a path or an identifier for a standard prime.
+ If unset, we use the DH prime from section 2.2 of RFC 5114, "ike23".
+ Set to "historic" to get the old GnuTLS behaviour of auto-generated DH
+ primes.
- 8. SSLv2 now disabled by default in OpenSSL. (Never supported by GnuTLS).
++17. SSLv2 now disabled by default in OpenSSL. (Never supported by GnuTLS).
+ Use "openssl_options -no_sslv2" to re-enable support, if your OpenSSL
+ install was not built with OPENSSL_NO_SSL2 ("no-ssl2").
+
Version 4.77
------------
in fact some of them were inherited from earlier versions.
-----------------------------------------------------------------------------------------
-accept_8bitmime boolean false main 1.60
+accept_8bitmime boolean true main 1.60 changed to true in 4.80
acl_not_smtp string* unset main 4.11
acl_not_smtp_mime string* unset main 4.51 with content scan
acl_smtp_auth string* unset main 4.00
gecos_pattern string unset main
gethostbyname boolean false smtp
gnutls_compat_mode boolean unset main 4.70
-gnutls_require_kx string* unset main 4.67
- string* unset smtp 4.67
-gnutls_require_mac string* unset main 4.67
- string* unset smtp 4.67
-gnutls_require_protocols string* unset main 4.67
- string* unset smtp 4.67
+gnutls_require_kx string* unset main 4.67 deprecated, warns
+ string* unset smtp 4.67 deprecated, warns
+gnutls_require_mac string* unset main 4.67 deprecated, warns
+ string* unset smtp 4.67 deprecated, warns
+gnutls_require_protocols string* unset main 4.67 deprecated, warns
+ string* unset smtp 4.67 deprecated, warns
group string + routers 4.00
unset transports 4.00 replaces local option in some transports
header_line_maxsize integer 0 (unset) main 4.14
once_file_size integer 0 autoreply 3.20
once_repeat time 0s autoreply 2.95
one_time boolean false redirect 4.00
- openssl_options string unset main 4.73 default to unset in 4.80
-openssl_options string +no_sslv2 main 4.73 default changed in 4.78
++openssl_options string +no_sslv2 main 4.73 default changed in 4.80
optional boolean false iplookup 4.00
oracle_servers string unset main 4.00
owners string list unset redirect 4.00
senders address list unset routers 4.00
serialize_hosts host list unset smtp 1.60
server_advertise_condition string* unset authenticators 4.14
-server_channelbinding bool false gsasl 4.78
+server_channelbinding bool false gsasl 4.80
server_condition string* unset authenticators 3.10 (plaintext) 4.64 (others)
-server_hostname string* "$primary_hostname" cyrus_sasl,gsasl,heimdal_gssapi (cyrus-only) 4.78 (others)
-server_keytab string* unset heimdal_gssapi 4.78
+server_hostname string* "$primary_hostname" cyrus_sasl,gsasl,heimdal_gssapi (cyrus-only) 4.80 (others)
+server_keytab string* unset heimdal_gssapi 4.80
server_mail_auth_condition string* unset authenticators 3.22
-server_mech string public_name cyrus_sasl,gsasl 4.43 (cyrus-only) 4.78 (others)
-server_password string unset gsasl 4.78
+server_mech string public_name cyrus_sasl,gsasl 4.43 (cyrus-only) 4.80 (others)
+server_password string unset gsasl 4.80
server_prompts string* unset plaintext 3.10
-server_realm string unset cyrus_sasl,gsasl 4.43 (cyrus-only) 4.78 (others)
-server_scram_iter string* unset gsasl 4.78
-server_scram_salt string* unset gsasl 4.78
+server_realm string unset cyrus_sasl,gsasl 4.43 (cyrus-only) 4.80 (others)
+server_scram_iter string* unset gsasl 4.80
+server_scram_salt string* unset gsasl 4.80
server_secret string* unset cram_md5 3.10
-server_service string "smtp" cyrus_sasl,gsasl,heimdal_gssapi (cyrus-only) 4.78 (others)
+server_service string "smtp" cyrus_sasl,gsasl,heimdal_gssapi (cyrus-only) 4.80 (others)
server_set_id string* unset authenticators 3.10
shadow_condition string* unset transports
shadow_transport string unset transports
tls_advertise_hosts host list * main 3.20
tls_certificate string* unset main 3.20
unset smtp 3.20
+tls_dh_max_bits integer 2236 main 4.80
tls_dhparam string* unset main 3.20
tls_on_connect_ports string unset main 4.43
tls_privatekey string* unset main 3.20
tls_remember_emstp boolean false main 4.21
tls_require_ciphers string* unset smtp 4.00 replaces tls_verify_ciphers
string* unset main 4.33
-tls_sni string* unset main 4.78
+tls_sni string* unset main 4.80
tls_tempfail_tryclear boolean true smtp 4.05
tls_try_verify_hosts host list unset main 4.00
tls_verify_certificates string* unset main 3.20
-bV Verify version number
-bv Test recipient address verification
-bvs Test sender address verification
+-bw + Inetd wait mode
-C + Use alternate configuration file
-D + Define macro for configuration file
-d + Turn on debugging output
MSGLOG_DIRECTORY_MODE optional* mode for message log directory
MV_COMMAND system path to mv command
NO_SYMLINK optional install doesn't make 'exim" symlink
-PCRE_CFLAGS system compile flags for PCRE library
+PCRE_CONFIG system* use pcre-config for PCRE support
+PCRE_LIBS system* library for using PCRE
PERL_CC system* compiler for Perl interface code
PERL_CCOPTS system* flags for same
PERL_COMMAND system path to Perl
that might affect a running system.
-Exim version 4.78
+Exim version 4.80
-----------------
+ * BEWARE backwards-incompatible changes in SSL libraries, thus the version
+ bump. See points below for details.
+ Also an LDAP data returned format change.
+
* The value of $tls_peerdn is now print-escaped when written to the spool file
in a -tls_peerdn line, and unescaped when read back in. We received reports
of values with embedded newlines, which caused spool file corruption.
the message. No tool has been provided as we believe this is a rare
occurence.
+ * For OpenSSL, SSLv2 is now disabled by default. (GnuTLS does not support
+ SSLv2). RFC 6176 prohibits SSLv2 and some informal surveys suggest no
+ actual usage. You can re-enable with the "openssl_options" Exim option,
+ in the main configuration section. Note that supporting SSLv2 exposes
+ you to ciphersuite downgrade attacks.
+
* With OpenSSL 1.0.1+, Exim now supports TLS 1.1 and TLS 1.2. If built
against 1.0.1a then you will get a warning message and the
"openssl_options" value will not parse "no_tlsv1_1": the value changes
"openssl_options" gains "no_tlsv1_1", "no_tlsv1_2" and "no_compression".
COMPATIBILITY WARNING: The default value of "openssl_options" is no longer
- "+dont_insert_empty_fragments". We default to unset. That old default was
- grandfathered in from before openssl_options became a configuration option.
+ "+dont_insert_empty_fragments". We default to "+no_sslv2".
+ That old default was grandfathered in from before openssl_options became a
+ configuration option.
Empty fragments are inserted by default through TLS1.0, to partially defend
against certain attacks; TLS1.1+ change the protocol so that this is not
needed. The DIEF SSL option was required for some old releases of mail
security for compatibility. Exim is now defaulting to higher security and
rewarding more modern clients.
+ If the option tls_dhparams is set and the parameters loaded from the file
+ have a bit-count greater than the new option tls_dh_max_bits, then the file
+ will now be ignored. If this affects you, raise the tls_dh_max_bits limit.
+ We suspect that most folks are using dated defaults and will not be affected.
+
* Ldap lookups returning multi-valued attributes now separate the attributes
with only a comma, not a comma-space sequence. Also, an actual comma within
a returned attribute is doubled. This makes it possible to parse the
attribute as a comma-separated list. Note the distinction from multiple
attributes being returned, where each one is a name=value pair.
+ If you are currently splitting the results from LDAP upon a comma, then you
+ should check carefully to see if adjustments are needed.
+
+ This change lets cautious folks distinguish "comma used as separator for
+ joining values" from "comma inside the data".
+
+ * accept_8bitmime now defaults on, which is not RFC compliant but is better
+ suited to today's Internet. See http://cr.yp.to/smtp/8bitmime.html for a
+ sane rationale. Those who wish to be strictly RFC compliant, or know that
+ they need to talk to servers that are not 8-bit-clean, now need to take
+ explicit configuration action to default this option off. This is not a
+ new option, you can safely force it off before upgrading, to decouple
+ configuration changes from the binary upgrade while remaining RFC compliant.
+
+ * The GnuTLS support has been mostly rewritten, to use APIs which don't cause
+ deprecation warnings in GnuTLS 2.12.x. As part of this, these three options
+ are no longer supported:
+
+ gnutls_require_kx
+ gnutls_require_mac
+ gnutls_require_protocols
+
+ Their functionality is entirely subsumed into tls_require_ciphers. In turn,
+ tls_require_ciphers is no longer an Exim list and is not parsed by Exim, but
+ is instead given to gnutls_priority_init(3), which expects a priority string;
+ this behaviour is much closer to the OpenSSL behaviour. See:
+
+ http://www.gnu.org/software/gnutls/manual/html_node/Priority-Strings.html
+
+ for fuller documentation of the strings parsed. The three gnutls_require_*
+ options are still parsed by Exim and, for this release, silently ignored.
+ A future release will add warnings, before a later still release removes
+ parsing entirely and the presence of the options will be a configuration
+ error.
+
+ Note that by default, GnuTLS will not accept RSA-MD5 signatures in chains.
+ A tls_require_ciphers value of NORMAL:%VERIFY_ALLOW_SIGN_RSA_MD5 may
+ re-enable support, but this is not supported by the Exim maintainers.
+ Our test suite no longer includes MD5-based certificates.
+
+ This rewrite means that Exim will continue to build against GnuTLS in the
+ future, brings Exim closer to other GnuTLS applications and lets us add
+ support for SNI and other features more readily. We regret that it wasn't
+ feasible to retain the three dropped options.
+
+ * If built with TLS support, then Exim will now validate the value of
+ the main section tls_require_ciphers option at start-up. Before, this
+ would cause a STARTTLS 4xx failure, now it causes a failure to start.
+ Running with a broken configuration which causes failures that may only
+ be left in the logs has been traded off for something more visible. This
+ change makes an existing problem more prominent, but we do not believe
+ anyone would deliberately be running with an invalid tls_require_ciphers
+ option.
+
+ This also means that library linkage issues caused by conflicts of some
+ kind might take out the main daemon, not just the delivery or receiving
+ process. Conceivably some folks might prefer to continue delivering
+ mail plaintext when their binary is broken in this way, if there is a
+ server that is a candidate to receive such mails that does not advertise
+ STARTTLS. Note that Exim is typically a setuid root binary and given
+ broken linkage problems that cause segfaults, we feel it is safer to
+ fail completely. (The check is not done as root, to ensure that problems
+ here are not made worse by the check).
+
+ * The "tls_dhparam" option has been updated, so that it can now specify a
+ path or an identifier for a standard DH prime from one of a few RFCs.
+ The default for OpenSSL is no longer to not use DH but instead to use
+ one of these standard primes. The default for GnuTLS is no longer to use
+ a file in the spool directory, but to use that same standard prime.
+ The option is now used by GnuTLS too. If it points to a path, then
+ GnuTLS will use that path, instead of a file in the spool directory;
+ GnuTLS will attempt to create it if it does not exist.
+
+ To preserve the previous behaviour of generating files in the spool
+ directory, set "tls_dhparam = historic". Since prior releases of Exim
+ ignored tls_dhparam when using GnuTLS, this can safely be done before
+ the upgrade.
+
+
Exim version 4.77
-----------------
problem. Prior to this release, supported values were "TLS1" and "SSL3",
so you should be able to update configuration prior to update.
+ [nb: gnutls_require_protocols removed in Exim 4.80, instead use
+ tls_require_ciphers to provide a priority string; see notes above]
+
* The match_<type>{string1}{string2} expansion conditions no longer subject
string2 to string expansion, unless Exim was built with the new
"EXPAND_LISTMATCH_RHS" option. Too many people have inadvertently created
* Exim - an Internet mail transport agent *
*************************************************/
-/* Copyright (c) University of Cambridge 1995 - 2009 */
+/* Copyright (c) University of Cambridge 1995 - 2012 */
/* See the file NOTICE for conditions of use and distribution. */
/* This module provides the TLS (aka SSL) support for Exim using the OpenSSL
#include <openssl/ssl.h>
#include <openssl/err.h>
#include <openssl/rand.h>
+#ifdef EXPERIMENTAL_OCSP
+#include <openssl/ocsp.h>
+#endif
+
+#ifdef EXPERIMENTAL_OCSP
+#define EXIM_OCSP_SKEW_SECONDS (300L)
+#define EXIM_OCSP_MAX_AGE (-1L)
+#endif
+
+#if OPENSSL_VERSION_NUMBER >= 0x0090806fL && !defined(OPENSSL_NO_TLSEXT)
+#define EXIM_HAVE_OPENSSL_TLSEXT
+#endif
/* Structure for collecting random data for seeding. */
static const uschar *sid_ctx = US"exim";
static SSL_CTX *ctx = NULL;
+#ifdef EXIM_HAVE_OPENSSL_TLSEXT
static SSL_CTX *ctx_sni = NULL;
+#endif
static SSL *ssl = NULL;
static char ssl_errstring[256];
typedef struct tls_ext_ctx_cb {
uschar *certificate;
uschar *privatekey;
+#ifdef EXPERIMENTAL_OCSP
+ uschar *ocsp_file;
+ uschar *ocsp_file_expanded;
+ OCSP_RESPONSE *ocsp_response;
+#endif
uschar *dhparam;
/* these are cached from first expand */
uschar *server_cipher_list;
static int
setup_certs(SSL_CTX *sctx, uschar *certs, uschar *crl, host_item *host, BOOL optional);
+/* Callbacks */
+#ifdef EXIM_HAVE_OPENSSL_TLSEXT
+static int tls_servername_cb(SSL *s, int *ad ARG_UNUSED, void *arg);
+#endif
+#ifdef EXPERIMENTAL_OCSP
+static int tls_stapling_cb(SSL *s, void *arg);
+#endif
+
/*************************************************
* Handle TLS error *
/* If dhparam is set, expand it, and load up the parameters for DH encryption.
Arguments:
- dhparam DH parameter file
+ dhparam DH parameter file or fixed parameter identity string
host connected host, if client; NULL if server
Returns: TRUE if OK (nothing to set up, or setup worked)
*/
static BOOL
-init_dh(uschar *dhparam, host_item *host)
+init_dh(SSL_CTX *sctx, uschar *dhparam, host_item *host)
{
-BOOL yield = TRUE;
BIO *bio;
DH *dh;
uschar *dhexpanded;
+const char *pem;
if (!expand_check(dhparam, US"tls_dhparam", &dhexpanded))
return FALSE;
-if (dhexpanded == NULL) return TRUE;
-
-if ((bio = BIO_new_file(CS dhexpanded, "r")) == NULL)
+if (dhexpanded == NULL || *dhexpanded == '\0')
{
- tls_error(string_sprintf("could not read dhparams file %s", dhexpanded),
- host, (uschar *)strerror(errno));
- yield = FALSE;
+ bio = BIO_new_mem_buf(CS std_dh_prime_default(), -1);
}
-else
+else if (dhexpanded[0] == '/')
{
- if ((dh = PEM_read_bio_DHparams(bio, NULL, NULL, NULL)) == NULL)
+ bio = BIO_new_file(CS dhexpanded, "r");
+ if (bio == NULL)
{
tls_error(string_sprintf("could not read dhparams file %s", dhexpanded),
- host, NULL);
- yield = FALSE;
+ host, US strerror(errno));
+ return FALSE;
}
- else
+ }
+else
+ {
+ if (Ustrcmp(dhexpanded, "none") == 0)
{
- SSL_CTX_set_tmp_dh(ctx, dh);
- DEBUG(D_tls)
- debug_printf("Diffie-Hellman initialized from %s with %d-bit key\n",
- dhexpanded, 8*DH_size(dh));
- DH_free(dh);
+ DEBUG(D_tls) debug_printf("Requested no DH parameters.\n");
+ return TRUE;
+ }
+
+ pem = std_dh_prime_named(dhexpanded);
+ if (!pem)
+ {
+ tls_error(string_sprintf("Unknown standard DH prime \"%s\"", dhexpanded),
+ host, US strerror(errno));
+ return FALSE;
}
+ bio = BIO_new_mem_buf(CS pem, -1);
+ }
+
+dh = PEM_read_bio_DHparams(bio, NULL, NULL, NULL);
+if (dh == NULL)
+ {
BIO_free(bio);
+ tls_error(string_sprintf("Could not read tls_dhparams \"%s\"", dhexpanded),
+ host, NULL);
+ return FALSE;
+ }
+
+/* Even if it is larger, we silently return success rather than cause things
+ * to fail out, so that a too-large DH will not knock out all TLS; it's a
+ * debatable choice. */
+if ((8*DH_size(dh)) > tls_dh_max_bits)
+ {
+ DEBUG(D_tls)
+ debug_printf("dhparams file %d bits, is > tls_dh_max_bits limit of %d",
+ 8*DH_size(dh), tls_dh_max_bits);
+ }
+else
+ {
+ SSL_CTX_set_tmp_dh(sctx, dh);
+ DEBUG(D_tls)
+ debug_printf("Diffie-Hellman initialized from %s with %d-bit prime\n",
+ dhexpanded ? dhexpanded : US"default", 8*DH_size(dh));
}
-return yield;
+DH_free(dh);
+BIO_free(bio);
+
+return TRUE;
}
+#ifdef EXPERIMENTAL_OCSP
+/*************************************************
+* Load OCSP information into state *
+*************************************************/
+
+/* Called to load the OCSP response from the given file into memory, once
+caller has determined this is needed. Checks validity. Debugs a message
+if invalid.
+
+ASSUMES: single response, for single cert.
+
+Arguments:
+ sctx the SSL_CTX* to update
+ cbinfo various parts of session state
+ expanded the filename putatively holding an OCSP response
+
+*/
+
+static void
+ocsp_load_response(SSL_CTX *sctx,
+ tls_ext_ctx_cb *cbinfo,
+ const uschar *expanded)
+{
+BIO *bio;
+OCSP_RESPONSE *resp;
+OCSP_BASICRESP *basic_response;
+OCSP_SINGLERESP *single_response;
+ASN1_GENERALIZEDTIME *rev, *thisupd, *nextupd;
+X509_STORE *store;
+unsigned long verify_flags;
+int status, reason, i;
+
+cbinfo->ocsp_file_expanded = string_copy(expanded);
+if (cbinfo->ocsp_response)
+ {
+ OCSP_RESPONSE_free(cbinfo->ocsp_response);
+ cbinfo->ocsp_response = NULL;
+ }
+
+bio = BIO_new_file(CS cbinfo->ocsp_file_expanded, "rb");
+if (!bio)
+ {
+ DEBUG(D_tls) debug_printf("Failed to open OCSP response file \"%s\"\n",
+ cbinfo->ocsp_file_expanded);
+ return;
+ }
+
+resp = d2i_OCSP_RESPONSE_bio(bio, NULL);
+BIO_free(bio);
+if (!resp)
+ {
+ DEBUG(D_tls) debug_printf("Error reading OCSP response.\n");
+ return;
+ }
+
+status = OCSP_response_status(resp);
+if (status != OCSP_RESPONSE_STATUS_SUCCESSFUL)
+ {
+ DEBUG(D_tls) debug_printf("OCSP response not valid: %s (%d)\n",
+ OCSP_response_status_str(status), status);
+ return;
+ }
+
+basic_response = OCSP_response_get1_basic(resp);
+if (!basic_response)
+ {
+ DEBUG(D_tls)
+ debug_printf("OCSP response parse error: unable to extract basic response.\n");
+ return;
+ }
+
+store = SSL_CTX_get_cert_store(sctx);
+verify_flags = OCSP_NOVERIFY; /* check sigs, but not purpose */
+
+/* May need to expose ability to adjust those flags?
+OCSP_NOSIGS OCSP_NOVERIFY OCSP_NOCHAIN OCSP_NOCHECKS OCSP_NOEXPLICIT
+OCSP_TRUSTOTHER OCSP_NOINTERN */
+
+i = OCSP_basic_verify(basic_response, NULL, store, verify_flags);
+if (i <= 0)
+ {
+ DEBUG(D_tls) {
+ ERR_error_string(ERR_get_error(), ssl_errstring);
+ debug_printf("OCSP response verify failure: %s\n", US ssl_errstring);
+ }
+ return;
+ }
+
+/* Here's the simplifying assumption: there's only one response, for the
+one certificate we use, and nothing for anything else in a chain. If this
+proves false, we need to extract a cert id from our issued cert
+(tls_certificate) and use that for OCSP_resp_find_status() (which finds the
+right cert in the stack and then calls OCSP_single_get0_status()).
+
+I'm hoping to avoid reworking a bunch more of how we handle state here. */
+single_response = OCSP_resp_get0(basic_response, 0);
+if (!single_response)
+ {
+ DEBUG(D_tls)
+ debug_printf("Unable to get first response from OCSP basic response.\n");
+ return;
+ }
+
+status = OCSP_single_get0_status(single_response, &reason, &rev, &thisupd, &nextupd);
+/* how does this status differ from the one above? */
+if (status != OCSP_RESPONSE_STATUS_SUCCESSFUL)
+ {
+ DEBUG(D_tls) debug_printf("OCSP response not valid (take 2): %s (%d)\n",
+ OCSP_response_status_str(status), status);
+ return;
+ }
+
+if (!OCSP_check_validity(thisupd, nextupd, EXIM_OCSP_SKEW_SECONDS, EXIM_OCSP_MAX_AGE))
+ {
+ DEBUG(D_tls) debug_printf("OCSP status invalid times.\n");
+ return;
+ }
+
+cbinfo->ocsp_response = resp;
+}
+#endif
+
+
+
+
/*************************************************
* Expand key and cert file specs *
*************************************************/
*/
static int
-tls_expand_session_files(SSL_CTX *sctx, const tls_ext_ctx_cb *cbinfo)
+tls_expand_session_files(SSL_CTX *sctx, tls_ext_ctx_cb *cbinfo)
{
uschar *expanded;
"SSL_CTX_use_PrivateKey_file file=%s", expanded), cbinfo->host, NULL);
}
+#ifdef EXPERIMENTAL_OCSP
+if (cbinfo->ocsp_file != NULL)
+ {
+ if (!expand_check(cbinfo->ocsp_file, US"tls_ocsp_file", &expanded))
+ return DEFER;
+
+ if (expanded != NULL && *expanded != 0)
+ {
+ DEBUG(D_tls) debug_printf("tls_ocsp_file %s\n", expanded);
+ if (cbinfo->ocsp_file_expanded &&
+ (Ustrcmp(expanded, cbinfo->ocsp_file_expanded) == 0))
+ {
+ DEBUG(D_tls)
+ debug_printf("tls_ocsp_file value unchanged, using existing values.\n");
+ } else {
+ ocsp_load_response(sctx, cbinfo, expanded);
+ }
+ }
+ }
+#endif
+
return OK;
}
Returns: SSL_TLSEXT_ERR_{OK,ALERT_WARNING,ALERT_FATAL,NOACK}
*/
-static int
-tls_servername_cb(SSL *s, int *ad ARG_UNUSED, void *arg);
-/* pre-declared for SSL_CTX_set_tlsext_servername_callback call within func */
-
+#ifdef EXIM_HAVE_OPENSSL_TLSEXT
static int
tls_servername_cb(SSL *s, int *ad ARG_UNUSED, void *arg)
{
const char *servername = SSL_get_servername(s, TLSEXT_NAMETYPE_host_name);
-const tls_ext_ctx_cb *cbinfo = (tls_ext_ctx_cb *) arg;
+tls_ext_ctx_cb *cbinfo = (tls_ext_ctx_cb *) arg;
int rc;
int old_pool = store_pool;
SSL_CTX_set_tlsext_servername_arg(ctx_sni, cbinfo);
if (cbinfo->server_cipher_list)
SSL_CTX_set_cipher_list(ctx_sni, CS cbinfo->server_cipher_list);
+#ifdef EXPERIMENTAL_OCSP
+if (cbinfo->ocsp_file)
+ {
+ SSL_CTX_set_tlsext_status_cb(ctx_sni, tls_stapling_cb);
+ SSL_CTX_set_tlsext_status_arg(ctx, cbinfo);
+ }
+#endif
+rc = setup_certs(ctx_sni, tls_verify_certificates, tls_crl, NULL, FALSE);
+if (rc != OK) return SSL_TLSEXT_ERR_NOACK;
+
+/* do this after setup_certs, because this can require the certs for verifying
+OCSP information. */
rc = tls_expand_session_files(ctx_sni, cbinfo);
if (rc != OK) return SSL_TLSEXT_ERR_NOACK;
-rc = setup_certs(ctx_sni, tls_verify_certificates, tls_crl, NULL, FALSE);
+rc = init_dh(ctx_sni, cbinfo->dhparam, NULL);
if (rc != OK) return SSL_TLSEXT_ERR_NOACK;
DEBUG(D_tls) debug_printf("Switching SSL context.\n");
SSL_set_SSL_CTX(s, ctx_sni);
+return SSL_TLSEXT_ERR_OK;
+}
+#endif /* EXIM_HAVE_OPENSSL_TLSEXT */
+
+
+
+
+#ifdef EXPERIMENTAL_OCSP
+/*************************************************
+* Callback to handle OCSP Stapling *
+*************************************************/
+
+/* Called when acting as server during the TLS session setup if the client
+requests OCSP information with a Certificate Status Request.
+
+Documentation via openssl s_server.c and the Apache patch from the OpenSSL
+project.
+
+*/
+
+static int
+tls_stapling_cb(SSL *s, void *arg)
+{
+const tls_ext_ctx_cb *cbinfo = (tls_ext_ctx_cb *) arg;
+uschar *response_der;
+int response_der_len;
+
+DEBUG(D_tls) debug_printf("Received TLS status request (OCSP stapling); %s response.\n",
+ cbinfo->ocsp_response ? "have" : "lack");
+if (!cbinfo->ocsp_response)
+ return SSL_TLSEXT_ERR_NOACK;
+
+response_der = NULL;
+response_der_len = i2d_OCSP_RESPONSE(cbinfo->ocsp_response, &response_der);
+if (response_der_len <= 0)
+ return SSL_TLSEXT_ERR_NOACK;
+
+SSL_set_tlsext_status_ocsp_resp(ssl, response_der, response_der_len);
return SSL_TLSEXT_ERR_OK;
}
+#endif /* EXPERIMENTAL_OCSP */
+
static int
tls_init(host_item *host, uschar *dhparam, uschar *certificate,
- uschar *privatekey, address_item *addr)
+ uschar *privatekey,
+#ifdef EXPERIMENTAL_OCSP
+ uschar *ocsp_file,
+#endif
+ address_item *addr)
{
long init_options;
int rc;
cbinfo = store_malloc(sizeof(tls_ext_ctx_cb));
cbinfo->certificate = certificate;
cbinfo->privatekey = privatekey;
+#ifdef EXPERIMENTAL_OCSP
+cbinfo->ocsp_file = ocsp_file;
+#endif
cbinfo->dhparam = dhparam;
cbinfo->host = host;
EVP_add_digest(EVP_sha256());
#endif
- /* Create a context */
+ /* Create a context.
+ The OpenSSL docs in 1.0.1b have not been updated to clarify TLS variant
+ negotiation in the different methods; as far as I can tell, the only
+ *_{server,client}_method which allows negotiation is SSLv23, which exists even
+ when OpenSSL is built without SSLv2 support.
+ By disabling with openssl_options, we can let admins re-enable with the
+ existing knob. */
ctx = SSL_CTX_new((host == NULL)?
SSLv23_server_method() : SSLv23_client_method());
/* Initialize with DH parameters if supplied */
-if (!init_dh(dhparam, host)) return DEFER;
+if (!init_dh(ctx, dhparam, host)) return DEFER;
-/* Set up certificate and key */
+/* Set up certificate and key (and perhaps OCSP info) */
rc = tls_expand_session_files(ctx, cbinfo);
if (rc != OK) return rc;
/* If we need to handle SNI, do so */
-#if OPENSSL_VERSION_NUMBER >= 0x0090806fL && !defined(OPENSSL_NO_TLSEXT)
+#ifdef EXIM_HAVE_OPENSSL_TLSEXT
if (host == NULL)
{
+#ifdef EXPERIMENTAL_OCSP
+ /* We check ocsp_file, not ocsp_response, because we care about if
+ the option exists, not what the current expansion might be, as SNI might
+ change the certificate and OCSP file in use between now and the time the
+ callback is invoked. */
+ if (cbinfo->ocsp_file)
+ {
+ SSL_CTX_set_tlsext_status_cb(ctx, tls_stapling_cb);
+ SSL_CTX_set_tlsext_status_arg(ctx, cbinfo);
+ }
+#endif
/* We always do this, so that $tls_sni is available even if not used in
tls_certificate */
SSL_CTX_set_tlsext_servername_callback(ctx, tls_servername_cb);
Arguments:
require_ciphers allowed ciphers
- ------------------------------------------------------
- require_mac list of allowed MACs ) Not used
- require_kx list of allowed key_exchange methods ) for
- require_proto list of allowed protocols ) OpenSSL
- ------------------------------------------------------
Returns: OK on success
DEFER for errors before the start of the negotiation
*/
int
-tls_server_start(uschar *require_ciphers, uschar *require_mac,
- uschar *require_kx, uschar *require_proto)
+tls_server_start(const uschar *require_ciphers)
{
int rc;
uschar *expciphers;
/* Initialize the SSL library. If it fails, it will already have logged
the error. */
-rc = tls_init(NULL, tls_dhparam, tls_certificate, tls_privatekey, NULL);
+rc = tls_init(NULL, tls_dhparam, tls_certificate, tls_privatekey,
+#ifdef EXPERIMENTAL_OCSP
+ tls_ocsp_file,
+#endif
+ NULL);
if (rc != OK) return rc;
cbinfo = static_cbinfo;
return FAIL;
/* In OpenSSL, cipher components are separated by hyphens. In GnuTLS, they
-are separated by underscores. So that I can use either form in my tests, and
-also for general convenience, we turn underscores into hyphens here. */
+were historically separated by underscores. So that I can use either form in my
+tests, and also for general convenience, we turn underscores into hyphens here.
+*/
if (expciphers != NULL)
{
verify_certs file for certificate verify
crl file containing CRL
require_ciphers list of allowed ciphers
- ------------------------------------------------------
- require_mac list of allowed MACs ) Not used
- require_kx list of allowed key_exchange methods ) for
- require_proto list of allowed protocols ) OpenSSL
- ------------------------------------------------------
timeout startup timeout
Returns: OK on success
tls_client_start(int fd, host_item *host, address_item *addr, uschar *dhparam,
uschar *certificate, uschar *privatekey, uschar *sni,
uschar *verify_certs, uschar *crl,
- uschar *require_ciphers, uschar *require_mac, uschar *require_kx,
- uschar *require_proto, int timeout)
+ uschar *require_ciphers, int timeout)
{
static uschar txt[256];
uschar *expciphers;
X509* server_cert;
int rc;
-rc = tls_init(host, dhparam, certificate, privatekey, addr);
+rc = tls_init(host, dhparam, certificate, privatekey,
+#ifdef EXPERIMENTAL_OCSP
+ NULL,
+#endif
+ addr);
if (rc != OK) return rc;
tls_certificate_verified = FALSE;
tls_sni = NULL;
else
{
+#ifdef EXIM_HAVE_OPENSSL_TLSEXT
DEBUG(D_tls) debug_printf("Setting TLS SNI \"%s\"\n", tls_sni);
SSL_set_tlsext_host_name(ssl, tls_sni);
+#else
+ DEBUG(D_tls)
+ debug_printf("OpenSSL at build-time lacked SNI support, ignoring \"%s\"\n",
+ tls_sni);
+#endif
}
}
+/*************************************************
+* Let tls_require_ciphers be checked at startup *
+*************************************************/
+
+/* The tls_require_ciphers option, if set, must be something which the
+library can parse.
+
+Returns: NULL on success, or error message
+*/
+
+uschar *
+tls_validate_require_cipher(void)
+{
+SSL_CTX *ctx;
+uschar *s, *expciphers, *err;
+
+/* this duplicates from tls_init(), we need a better "init just global
+state, for no specific purpose" singleton function of our own */
+
+SSL_load_error_strings();
+OpenSSL_add_ssl_algorithms();
+#if (OPENSSL_VERSION_NUMBER >= 0x0090800fL) && !defined(OPENSSL_NO_SHA256)
+/* SHA256 is becoming ever more popular. This makes sure it gets added to the
+list of available digests. */
+EVP_add_digest(EVP_sha256());
+#endif
+
+if (!(tls_require_ciphers && *tls_require_ciphers))
+ return NULL;
+
+if (!expand_check(tls_require_ciphers, US"tls_require_ciphers", &expciphers))
+ return US"failed to expand tls_require_ciphers";
+
+if (!(expciphers && *expciphers))
+ return NULL;
+
+/* normalisation ripped from above */
+s = expciphers;
+while (*s != 0) { if (*s == '_') *s = '-'; s++; }
+
+err = NULL;
+
+ctx = SSL_CTX_new(SSLv23_server_method());
+if (!ctx)
+ {
+ ERR_error_string(ERR_get_error(), ssl_errstring);
+ return string_sprintf("SSL_CTX_new() failed: %s", ssl_errstring);
+ }
+
+DEBUG(D_tls)
+ debug_printf("tls_require_ciphers expands to \"%s\"\n", expciphers);
+
+if (!SSL_CTX_set_cipher_list(ctx, CS expciphers))
+ {
+ ERR_error_string(ERR_get_error(), ssl_errstring);
+ err = string_sprintf("SSL_CTX_set_cipher_list(%s) failed", expciphers);
+ }
+
+SSL_CTX_free(ctx);
+
+return err;
+}
+
+
+
+
/*************************************************
* Report the library versions. *
*************************************************/
/*************************************************
-* Pseudo-random number generation *
+* Random number generation *
*************************************************/
/* Pseudo-random number generation. The result is not expected to be
*/
int
-pseudo_random_number(int max)
+vaguely_random_number(int max)
{
unsigned int r;
int i, needed_len;
needed_len = i;
/* We do not care if crypto-strong */
-(void) RAND_pseudo_bytes(smallbuf, needed_len);
+i = RAND_pseudo_bytes(smallbuf, needed_len);
+if (i < 0)
+ {
+ DEBUG(D_all)
+ debug_printf("OpenSSL RAND_pseudo_bytes() not supported by RAND method, using fallback.\n");
+ return vaguely_random_number_fallback(max);
+ }
+
r = 0;
for (p = smallbuf; needed_len; --needed_len, ++p)
{
BOOL adding, item_parsed;
result = 0L;
-/* Prior to 4.78 we or'd in SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS; removed
+/* Prior to 4.80 we or'd in SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS; removed
* from default because it increases BEAST susceptibility. */
+ #ifdef SSL_OP_NO_SSLv2
+ result |= SSL_OP_NO_SSLv2;
+ #endif
if (option_spec == NULL)
{
git://git.exim.org / exim.git / commitdiff