An attacker that can get Dovecot to relay a message, for example through Sieve redirect or submission relay, can use a crafted line ending in the message body to bypass the outbound protection that prevents message content from being interpreted as SMTP commands. A downstream mail server that hasn't yet fixed the SMTP smuggling vulnerability can be tricked into treating part of the message body as new SMTP commands, allowing injection of spoofed email. This is the same vulnerability class as CVE-2023-51764 and CVE-2023-51766. Where you control the receiving mail servers, ensure they reject bare carriage returns in message data. Update to non-vulnerable version. No publicly available exploits are known.

Dovecot accepts dot LF DOT LF symbol as end of DATA command. RFC requires that it should always be CR LF DOT CR LF. This causes Dovecot to convert single mail with LF DOT LF in middle, into two emails when relaying to SMTP. Dovecot will split mail with LF DOT LF into two mails. Upgrade to latest released version. No publicly available exploits are known.

Vulnerability Details: Having a large number of address headers (From, To, Cc, Bcc, etc.) becomes excessively CPU intensive. With 100k header lines CPU usage is already 12 seconds, and in a production environment we observed 500k header lines taking 18 minutes to parse. Since this can be triggered by external actors sending emails to a victim, this is a security issue.

The main problem is that each header line's address is added to the end of a linked list. This is done by walking the whole linked list, which becomes more inefficient the more addresses there are.

Workaround: One can implement restrictions on address headers on MTA component preceding Dovecot.

Fix: Install non-vulnerable version of Dovecot. Patch can be found at https://github.com/dovecot/core/compare/8e4c42d%5E...1481c04.patch

An issue was discovered in the auth component in Dovecot 2.2 and 2.3 before 2.3.20. When two passdb configuration entries exist with the same driver and args settings, incorrect usernamefilter and mechanism settings can be applied to passdb definitions. These incorrectly applied settings can lead to an unintended security configuration and can permit privilege escalation in certain configurations. The documentation does not advise against the use of passdb definitions that have the same driver and args settings. One such configuration would be where an administrator wishes to use the same PAM configuration or passwd file for both normal and master users but use the usernamefilter setting to restrict which of the users is able to be a master user.

The Sieve engine in Dovecot before 2.3.15 allows Uncontrolled Resource Consumption, as demonstrated by a situation with a complex regular expression for the regex extension.

The submission service in Dovecot before 2.3.15 allows STARTTLS command injection in lib-smtp. Sensitive information can be redirected to an attacker-controlled address.

Dovecot before 2.3.15 allows ../ Path Traversal. An attacker with access to the local filesystem can trick OAuth2 authentication into using an HS256 validation key from an attacker-controlled location. This occurs during use of local JWT validation with the posix fs driver.

An issue was discovered in Dovecot before 2.3.13. By using IMAP IDLE, an authenticated attacker can trigger unhibernation via attacker-controlled parameters, leading to access to other users' email messages (and path disclosure).

Dovecot before 2.3.13 has Improper Input Validation in lda, lmtp, and imap, leading to an application crash via a crafted email message with certain choices for ten thousand MIME parts.

In Dovecot before 2.3.9.2, an attacker can crash a push-notification driver with a crafted email when push notifications are used, because of a NULL Pointer Dereference. The email must use a group address as either the sender or the recipient.

IMAP and ManageSieve protocol parsers do not properly handle NUL byte when scanning data in quoted strings, leading to out of bounds heap memory writes.

In the IMAP Server in Dovecot 2.3.3 through 2.3.5.2, the submission-login service crashes when the client disconnects prematurely during the AUTH command.

In the IMAP Server in Dovecot 2.3.3 through 2.3.5.2, the submission-login component crashes if AUTH PLAIN is attempted over a TLS secured channel with an unacceptable authentication message.

The JSON encoder in Dovecot before 2.3.5.2 allows attackers to repeatedly crash the authentication service by attempting to authenticate with an invalid UTF-8 sequence as the username.

Dovecot before version 2.2.29 is vulnerable to a denial of service. When 'dict' passdb and userdb were used for user authentication, the username sent by the IMAP/POP3 client was sent through varexpand() to perform %variable expansion. Sending specially crafted %variable fields could result in excessive memory usage causing the process to crash (and restart), or excessive CPU usage causing all authentications to hang.

A postinstall script in the dovecot rpm allows local users to read the contents of newly created SSL/TLS key files.

A flaw was found in the way Dovecot handled SSL handshake failures. A remote attacker could use this flaw to crash the imap-login and pop3-login processes.

Note that only Dovecot installations accepting SSL/TLS connections that have SSLv3 disabled are vulnerable.

Additional details:

http://dovecot.org/pipermail/dovecot/2015-April/100618.html http://seclists.org/oss-sec/2015/q2/288

Upstream patch:

http://hg.dovecot.org/dovecot-2.2/rev/86f535375750

A security flaw was found in the way Dovecot, an IMAP and POP3 email server, performed remote server identity verification (x509 certificate's Common Name field was not checked to match provided remote server host name), when Dovecot was configured to proxy IMAP and POP3 connections to remote hosts and TLS/SSL protocols were requested (ssl=yes or starttls=yes) in the configuration to secure these connections to the destination server. A remote attacker could use this flaw to conduct man-in-the-middle (MITM) attacks via specially-crafted x509v3 certificate.

References: [1] http://www.dovecot.org/list/dovecot-news/2011-November/000200.html [2] https://secunia.com/advisories/46886/ [3] https://bugs.gentoo.org/showbug.cgi?id=390887 [4] http://wiki.dovecot.org/PasswordDatabase/ExtraFields/Proxy

Relevant upstream patch: [5] http://hg.dovecot.org/dovecot-2.0/rev/5e9eaf63a6b1

script-login in Dovecot 2.0.x before 2.0.13 does not follow the user and group configuration settings, which might allow remote authenticated users to bypass intended access restrictions by leveraging a script.

script-login in Dovecot 2.0.x before 2.0.13 does not follow the chroot configuration setting, which might allow remote authenticated users to conduct directory traversal attacks by leveraging a script.

Dovecot has released version 1.2.17 [1] and 2.0.13 [2] to address a potential crash, and possibly mailbox corruption, when dovecot parsed header names that contained NUL characters. This was due to a pointer possibly pointing past allocated memory. An upstream patch [3] is available.

[1] http://dovecot.org/pipermail/dovecot/2011-May/059086.html [2] http://dovecot.org/pipermail/dovecot/2011-May/059085.html [3] http://hg.dovecot.org/dovecot-1.1/rev/3698dfe0f21c

Dovecot has released version 1.2.17 [1] and 2.0.13 [2] to address a potential crash, and possibly mailbox corruption, when dovecot parsed header names that contained NUL characters. This was due to a pointer possibly pointing past allocated memory. An upstream patch [3] is available.

[1] http://dovecot.org/pipermail/dovecot/2011-May/059086.html [2] http://dovecot.org/pipermail/dovecot/2011-May/059085.html [3] http://hg.dovecot.org/dovecot-1.1/rev/3698dfe0f21c

The ACL plugin in Dovecot 1.2.x before 1.2.13 propagates INBOX ACLs to newly created mailboxes in certain configurations, which might allow remote attackers to read mailboxes that have unintended weak ACLs.

Dovecot 1.2.x before 1.2.8 sets 0777 permissions during creation of certain directories at installation time, which allows local users to access arbitrary user accounts by replacing the auth socket, related to the parent directories of the basedir directory, and possibly the basedir directory itself.

Multiple stack-based buffer overflows in the Sieve plugin in Dovecot 1.0 before 1.0.4 and 1.1 before 1.1.7, as derived from Cyrus libsieve, allow context-dependent attackers to cause a denial of service (crash) and possibly execute arbitrary code via a crafted SIEVE script, as demonstrated by forwarding an e-mail message to a large number of recipients, a different vulnerability than CVE-2009-2632.

Directory traversal vulnerability in the ManageSieve implementation in Dovecot 1.0.15, 1.1, and 1.2 allows remote attackers to read and modify arbitrary .sieve files via a ".." (dot dot) in a script name.

The message parsing feature in Dovecot 1.1.4 and 1.1.5, when using the FETCH ENVELOPE command in the IMAP client, allows remote attackers to cause a denial of service (persistent crash) via an email with a malformed From address, which triggers an assertion error, aka "invalid message address parsing bug."

dovecot 1.0.7 in Red Hat Enterprise Linux (RHEL) 5, and possibly Fedora, uses world-readable permissions for dovecot.conf, which allows local users to obtain the sslkeypassword parameter value.

The ACL plugin in Dovecot before 1.1.4 treats negative access rights as if they are positive access rights, which allows attackers to bypass intended access restrictions.

The ACL plugin in Dovecot before 1.1.4 allows attackers to bypass intended access restrictions by using the "k" right to create unauthorized "parent/child/child" mailboxes.