diff --git a/bin/tools/named-rrchecker.rst b/bin/tools/named-rrchecker.rst index 28da236b72..ec80946027 100644 --- a/bin/tools/named-rrchecker.rst +++ b/bin/tools/named-rrchecker.rst @@ -26,37 +26,210 @@ Synopsis Description ~~~~~~~~~~~ -:program:`named-rrchecker` reads a individual DNS resource record from standard +:program:`named-rrchecker` reads a single DNS resource record (RR) from standard input and checks whether it is syntactically correct. +The input format is a minimal subset of the DNS zone file format. The entire input must be: + CLASS TYPE RDATA + +* Input must not start with an owner (domain) name +* The `CLASS` field is mandatory (typically ``IN``). +* The `TTL` field **must not** be present. +* RDATA format is specific to each RRTYPE. +* Leading and trailing whitespace in each field is ignored. + +Format details can be found in :rfc:`1035#section-5.1` under ```` +specification. :rfc:`3597` format is also accepted in any of the input fields. +See :ref:`Examples`. + + Options ~~~~~~~ +.. option:: -o origin + + This option specifies the origin to be used when interpreting names in the record: + it defaults to root (`.`). The specified origin is always taken as an absolute name. + +.. option:: -p + + This option prints out the resulting record in canonical form. If there + is no canonical form defined, the record is printed in :rfc:`3597` unknown + record format. + +.. option:: -u + + This option prints out the resulting record in :rfc:`3597` unknown record + format. + +.. option:: -C, -T, -P + + These options do not read input. They print out known classes, standard types, + and private type mnemonics. Each item is printed on a separate line. + The resulting list of private types may be empty + .. option:: -h This option prints out the help menu. -.. option:: -o origin - This option specifies the origin to be used when interpreting - the record. +.. _examples: -.. option:: -p +Examples +~~~~~~~~ +Pay close attention to the :manpage:`echo` command line options `-e` and `-n`, as they affect whitespace in the input to ``named-rrchecker``. - This option prints out the resulting record in canonical form. If there - is no canonical form defined, the record is printed in unknown - record format. +echo -n 'IN A 192.0.2.1' | named-rrchecker + * Valid input is in :rfc:`1035` format with no newline at the end of the input. + * Return code 0. -.. option:: -u +echo -e '\\n \\n IN\\tA 192.0.2.1 \\t \\n\\n ' | named-rrchecker -p + * Valid input with leading and trailing whitespace. + * Output: ``IN A 192.0.2.1`` + * Leading and trailing whitespace is not part of the output. - This option prints out the resulting record in unknown record form. -.. option:: -C, -T, -P +Relative names and origin +^^^^^^^^^^^^^^^^^^^^^^^^^ +echo 'IN CNAME target' | named-rrchecker -p + * Valid input with a relative name as the CNAME target. + * Output: ``IN CNAME target.`` + * Relative name `target` from the input is converted to an absolute name using the default origin ``.`` (root). + +echo 'IN CNAME target' | named-rrchecker -p -o origin.test + * Valid input with a relative name as the CNAME target. + * Output: ``IN CNAME target.origin.test.`` + * Relative name `target` from the input is converted to an absolute name using the specified origin ``origin.test`` +echo 'IN CNAME target.' | named-rrchecker -p -o origin.test + * Valid input with an absolute name as the CNAME target. + * Output: ``IN CNAME target.`` + * The specified origin has no influence if `target` from the input is already absolute. + + +Special characters +^^^^^^^^^^^^^^^^^^ +Special characters allowed in zone files by :rfc:`1035#section-5.1` are accepted. + +echo 'IN CNAME t\\097r\\get\\.' | named-rrchecker -p -o origin.test + * Valid input with backslash escapes. + * Output: ``IN CNAME target\..origin.test.`` + * ``\097`` denotes an ASCII value in decimal, which, in this example, is the character ``a``. + * ``\g`` is converted to a plain ``g`` because the ``g`` character does not have a special meaning and so the ``\`` prefix does nothing in this case. + * ``\.`` denotes a literal ASCII dot (here as a part of the CNAME target name). Special meaning of ``.`` as the DNS label separator was disabled by the preceding ``\`` prefix. + +echo 'IN CNAME @' | named-rrchecker -p -o origin.test + * Valid input with ``@`` used as a reference to the specified origin. + * Output: ``IN CNAME origin.test.`` + +echo 'IN CNAME \\@' | named-rrchecker -p -o origin.test + * Valid input with a literal ``@`` character (escaped). + * Output: ``IN CNAME \@.origin.test.`` + +echo 'IN CNAME prefix.@' | named-rrchecker -p -o origin.test + * Valid input with ``@`` used as a reference to the specifed origin. + * Output: ``IN CNAME prefix.\@.origin.test.`` + * ``@`` has special meaning only if it is free-standing. + +echo 'IN A 192.0.2.1; comment' | named-rrchecker -p + * Valid input with a trailing comment. Note the lack of whitespace before the start of the comment. + * Output: ``IN A 192.0.2.1`` + +For multi-line examples see the next section. + +Multi-token records +^^^^^^^^^^^^^^^^^^^ +echo -e 'IN TXT two words \\n' | named-rrchecker -p + * Valid TXT RR with two unquoted words and trailing whitespace. + * Output: ``IN TXT "two" "words"`` + * Two unquoted words in the input are treated as two ``\ s per :rfc:`1035#section-3.3.14`. + * Trailing whitespace is omitted from the last ``. + +echo -e 'IN TXT "two words" \\n' | named-rrchecker -p + * Valid TXT RR with one `character-string` and trailing whitespace. + * Output: ``IN TXT "two words"`` + +echo -e 'IN TXT "problematic newline\\n"' | named-rrchecker -p + * Invalid input - the closing ``"`` is not detected before the end of the line. + +echo 'IN TXT "with newline\\010"' | named-rrchecker -p + * Valid input with an escaped newline character inside `character-string`. + * Output: ``IN TXT "with newline\010"`` + +echo -e 'IN TXT ( two\\nwords )' | named-rrchecker -p + * Valid multi-line input with line continuation allowed inside optional parentheses in the RDATA field. + * Output: ``IN TXT "two" "words"`` + +echo -e 'IN TXT ( two\\nwords ; misplaced comment )' | named-rrchecker -p + * Invalid input - comments, starting with ";", are ignored by the parser, so the closing parenthesis should be before the semicolon. + +echo -e 'IN TXT ( two\\nwords ; a working comment\\n )' | named-rrchecker -p + * Valid input - the comment is terminated with a newline. + * Output: ``IN TXT "two" "words"`` + +echo 'IN HTTPS 1 . alpn="h2,h3"' | named-rrchecker -p + * Valid HTTPS record + * Output: ``IN HTTPS 1 . alpn="h2,h3"`` + +echo -e 'IN HTTPS ( 1 \\n . \\n alpn="dot")port=853' | named-rrchecker -p + * Valid HTTPS record with individual sub-fields split across multiple lines + using :rfc:`1035#section-5.1` parentheses syntax to group data that crosses + a line boundary. + * Note the missing whitespace between the closing parenthesis and adjacent tokens. + * Output: ``IN HTTPS 1 . alpn="dot" port=853`` + + +Unknown type handling +^^^^^^^^^^^^^^^^^^^^^ + +echo 'IN A 192.0.2.1' | named-rrchecker -u + * Valid input in :rfc:`1035` format. + * Output in :rfc:`3957` format: ``CLASS1 TYPE1 \# 4 C0000201`` + +echo 'CLASS1 TYPE1 \\# 4 C0000201' | named-rrchecker -p + * Valid input in :rfc:`3597` format. + * Output in :rfc:`1035` format: ``IN A 192.0.2.1`` + +echo 'IN A \\# 4 C0000201' | named-rrchecker -p + * Valid input with class and type in :rfc:`1035` format and rdata in :rfc:`3597` format. + * Output in :rfc:`1035` format: ``IN A 192.0.2.1`` + +echo 'IN HTTPS 1 . key3=\\001\\000' | named-rrchecker -p + * Valid input with :rfc:`9460` syntax for an unknown `key3` field. Syntax ``\001\000`` produces two octets with values 1 and 0, respectively. + * Output: ``IN HTTPS 1 . port=256`` + * `key3` matches the standardized key name `port`. + * Octets 1 and 0 were decoded as integer values in big-endian encoding. + +echo 'IN HTTPS 1 . key3=\\001' | named-rrchecker -p + * Invalid input - the length of the value for `key3` (i.e. port) does not match the known standard format for that parameter in the SVCB RRTYPE. + +echo 'IN HTTPS 1 . port=\\001\\000' | named-rrchecker -p + * Invalid input - the key `port`, when specified using its standard mnemonic name, **must** use standard key-specific syntax. + +Meta values +^^^^^^^^^^^ + +echo 'IN AXFR' | named-rrchecker + * Invalid input - AXFR is a meta type, not a genuine RRTYPE. + +echo 'ANY A 192.0.2.1' | named-rrchecker + * Invalid input - ANY is meta class, not a true class. + +echo 'A 192.0.2.1' | named-rrchecker + * Invalid input - the class field is missing, so the parser would try and fail to interpret the RRTYPE A as the class. + + +Return Codes +~~~~~~~~~~~~ + +0 + The whole input was parsed as one syntactically valid resource record. + +1 + The input is not a syntactically valid resource record, or the given type is not + supported, or either/both class and type are meta-values, which should not appear in zone files. - These options print out the known class, standard type, - and private type mnemonics, respectively. See Also ~~~~~~~~ -:rfc:`1034`, :rfc:`1035`, :iscman:`named(8) `. +:rfc:`1034`, :rfc:`1035`, :rfc:`3957`, :iscman:`named(8) `.