X-Git-Url: https://vcs.fsf.org/?a=blobdiff_plain;f=doc%2Fdoc-txt%2Fexperimental-spec.txt;h=aa7046e58f9b8bb1398eb1795f7a44d0b62cf134;hb=4a1bd6b935ca5c5b70408a60036312d4825fd24e;hp=43f14237bc502097d9dae4ab7ae7ddf6e3494c8d;hpb=8ac90765750f87c573300b9e953af3d8090cab8b;p=exim.git diff --git a/doc/doc-txt/experimental-spec.txt b/doc/doc-txt/experimental-spec.txt index 43f14237b..aa7046e58 100644 --- a/doc/doc-txt/experimental-spec.txt +++ b/doc/doc-txt/experimental-spec.txt @@ -428,7 +428,7 @@ need to uncomment them if an rpm (or you) installed them in the package controlled locations (/usr/include and /usr/lib). -2. Use the following global settings to configure DMARC: +2. Use the following global options to configure DMARC: Required: dmarc_tld_file Defines the location of a text file of valid @@ -437,6 +437,8 @@ dmarc_tld_file Defines the location of a text file of valid the most current version can be downloaded from a link at http://publicsuffix.org/list/. See also util/renew-opendmarc-tlds.sh script. + The default for the option is currently + /etc/exim/opendmarc.tlds Optional: dmarc_history_file Defines the location of a file to log results @@ -447,11 +449,19 @@ dmarc_history_file Defines the location of a file to log results directory of this file is writable by the user exim runs as. -dmarc_forensic_sender The email address to use when sending a +dmarc_forensic_sender Alternate email address to use when sending a forensic report detailing alignment failures if a sender domain's dmarc record specifies it and you have configured Exim to send them. - Default: do-not-reply@$default_hostname + + If set, this is expanded and used for the + From: header line; the address is extracted + from it and used for the envelope from. + If not set, the From: header is expanded from + the dsn_from option, and <> is used for the + envelope from. + + Default: unset. 3. By default, the DMARC processing will run for any remote, @@ -709,6 +719,8 @@ an external directory retaining the exim spool format. The spool files can then be processed by external processes and then requeued into exim spool directories for final delivery. +However, note carefully the warnings in the main documentation on +qpool file formats. The motivation/inspiration for the transport is to allow external processes to access email queued by exim and have access to all the @@ -793,7 +805,7 @@ standard header. Note that it would be wise to strip incoming messages of A-R headers that claim to be from our own . -There are three new variables: $arc_state, $arc_state_reason, $arc_domains: +There are four new variables: $arc_state One of pass, fail, none $arc_state_reason (if fail, why) @@ -861,34 +873,138 @@ used via the transport in question. -REQUIRETLS support ------------------- -Ref: https://tools.ietf.org/html/draft-ietf-uta-smtp-require-tls-03 +Early pipelining support +------------------------ +Ref: https://datatracker.ietf.org/doc/draft-harris-early-pipe/ + +If compiled with EXPERIMENTAL_PIPE_CONNECT support is included for this feature. +The server advertises the feature in its EHLO response, currently using the name +"X_PIPE_CONNECT" (this will change, some time in the future). +A client may cache this information, along with the rest of the EHLO response, +and use it for later connections. Those later ones can send esmtp commands before +a banner is received. + +Up to 1.5 roundtrip times can be taken out of cleartext connections, 2.5 on +STARTTLS connections. + +In combination with the traditional PIPELINING feature the following example +sequences are possible (among others): -If compiled with EXPERIMENTAL_REQUIRETLS support is included for this -feature, where a REQUIRETLS option is added to the MAIL command. -The client may not retry in clear if the MAIL+REQUIRETLS fails (or was never -offered), and the server accepts an obligation that any onward transmission -by SMTP of the messages accepted will also use REQUIRETLS - or generate a -fail DSN. +(client) (server) -The Exim implementation includes -- a main-part option tls_advertise_requiretls; host list, default "*" -- an observability variable $requiretls returning yes/no -- an ACL "control = requiretls" modifier for setting the requirement -- Log lines and Received: headers capitalise the S in the protocol - element: "P=esmtpS" +EHLO,MAIL,RCPT,DATA -> + <- banner,EHLO-resp,MAIL-ack,RCPT-ack,DATA-goahead +message-data -> +------ -Differences from spec: -- we support upgrading the requirement for REQUIRETLS, including adding - it from cold, withing an MTA. The spec only define the sourcing MUA - as being able to source the requirement, and makes no mention of upgrade. -- No support is coded for the RequireTLS header (which can be used - to annul DANE and/or STS policiy). [can this be done in ACL?] +EHLO,MAIL,RCPT,BDAT -> + <- banner,EHLO-resp,MAIL-ack,RCPT-ack +message-data -> +------ -Note that REQUIRETLS is only advertised once a TLS connection is acheived -(in contrast to STARTTLS). If you want to check the advertising, do something -like "swaks -s 127.0.0.1 -tls -q HELO". +EHLO,STARTTLS -> + <- banner,EHLO-resp,TLS-goahead +TLS1.2-client-hello -> + <- TLS-server-hello,cert,hello-done +client-Kex,change-cipher,finished -> + <- change-cipher,finished +EHLO,MAIL,RCPT,DATA -> + <- EHLO-resp,MAIL-ack,RCPT-ack,DATA-goahead + +------ +(tls-on-connect) +TLS1.2-client-hello -> + <- TLS-server-hello,cert,hello-done +client-Kex,change-cipher,finished -> + <- change-cipher,finshed + <- banner +EHLO,MAIL,RCPT,DATA -> + <- EHLO-resp,MAIL-ack,RCPT-ack,DATA-goahead + +Where the initial client packet is SMTP, it can combine with the TCP Fast Open +feature and be sent in the TCP SYN. + + +A main-section option "pipelining_connect_advertise_hosts" (default: *) +and an smtp transport option "hosts_pipe_connect" (default: unset) +control the feature. + +If the "pipelining" log_selector is enabled, the "L" field in server <= +log lines has a period appended if the feature was advertised but not used; +or has an asterisk appended if the feature was used. In client => lines +the "L" field has an asterisk appended if the feature was used. + +The "retry_data_expire" option controls cache invalidation. +Entries are also rewritten (or cleared) if the adverised features +change. + + +NOTE: since the EHLO command must be constructed before the connection is +made it cannot depend on the interface IP address that will be used. +Transport configurations should be checked for this. An example avoidance: + + helo_data = ${if def:sending_ip_address \ + {${lookup dnsdb{>! ptr=$sending_ip_address} \ + {${sg{$value} {^([^!]*).*\$} {\$1}}} fail}} \ + {$primary_hostname}} + + + + +TLS Session Resumption +---------------------- +TLS Session Resumption for TLS 1.2 and TLS 1.3 connections can be used (defined +in RFC 5077 for 1.2). The support for this can be included by building with +EXPERIMENTAL_TLS_RESUME defined. This requires GnuTLS 3.6.3 or OpenSSL 1.1.1 +(or later). + +Session resumption (this is the "stateless" variant) involves the server sending +a "session ticket" to the client on one connection, which can be stored by the +client and used for a later session. The ticket contains sufficient state for +the server to reconstruct the TLS session, avoiding some expensive crypto +calculation and one full packet roundtrip time. + +Operational cost/benefit: + The extra data being transmitted costs a minor amount, and the client has + extra costs in storing and retrieving the data. + + In the Exim/Gnutls implementation the extra cost on an initial connection + which is TLS1.2 over a loopback path is about 6ms on 2017-laptop class hardware. + The saved cost on a subsequent connection is about 4ms; three or more + connections become a net win. On longer network paths, two or more + connections will have an average lower startup time thanks to the one + saved packet roundtrip. TLS1.3 will save the crypto cpu costs but not any + packet roundtrips. + + Since a new hints DB is used, the hints DB maintenance should be updated + to additionally handle "tls". + +Security aspects: + The session ticket is encrypted, but is obviously an additional security + vulnarability surface. An attacker able to decrypt it would have access + all connections using the resumed session. + The session ticket encryption key is not committed to storage by the server + and is rotated regularly. Tickets have limited lifetime. + + There is a question-mark over the security of the Diffie-Helman parameters + used for session negotiation. TBD. q-value; cf bug 1895 + +Observability: + New log_selector "tls_resumption", appends an asterisk to the tls_cipher "X=" + element. + + Variables $tls_{in,out}_resumption have bits 0-4 indicating respectively + support built, client requested ticket, client offered session, + server issued ticket, resume used. A suitable decode list is provided + in the builtin macro _RESUME_DECODE for ${listextract {}{}}. + +Issues: + In a resumed session: + $tls_{in,out}_certificate_verified will be set, and verify = certificate + will be true, when verify failed but tls_try_verify_hosts allowed the + connection (under OpenSSL) + $tls_{in,out}_cipher will have values different to the original (under GnuTLS) + $tls_{in,out}_ocsp will be "not requested" or "no response" --------------------------------------------------------------