openssl/doc/man5/config.pod
Benjamin Kaduk 3472082b4b openssl-config: add example libssl system-defaults
Provide a "simple" example for affecting the systemwide default behavior
of libssl.  The large number of mandatory nested sections makes this
less simple than the main description might suggest.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/10937)
2020-01-24 20:52:38 -08:00

587 lines
19 KiB
Plaintext

=pod
=head1 NAME
config - OpenSSL CONF library configuration files
=head1 DESCRIPTION
The OpenSSL CONF library can be used to read configuration files.
It is used for the OpenSSL master configuration file B<openssl.cnf>
and in a few other places like B<SPKAC> files and certificate extension
files for the B<x509> utility. OpenSSL applications can also use the
CONF library for their own purposes.
A configuration file is divided into a number of sections. Each section
starts with a line C<[section_name]> and ends when a new section is
started or end of file is reached. A section name can consist of
alphanumeric characters and underscores. The brackets are required.
The first section of a configuration file is special and is referred
to as the B<default> section. This section is usually unnamed and spans from the
start of file until the first named section. When a name is being looked up
it is first looked up in a named section (if any) and then the
default section.
The environment is mapped onto a section called B<ENV>.
Comments can be included by preceding them with the B<#> character
Other files can be included using the B<.include> directive followed
by a path. If the path points to a directory all files with
names ending with B<.cnf> or B<.conf> are included from the directory.
Recursive inclusion of directories from files in such directory is not
supported. That means the files in the included directory can also contain
B<.include> directives but only inclusion of regular files is supported
there. The inclusion of directories is not supported on systems without
POSIX IO support.
It is strongly recommended to use absolute paths with the B<.include>
directive. Relative paths are evaluated based on the application current
working directory so unless the configuration file containing the
B<.include> directive is application specific the inclusion will not
work as expected. The environment variable B<OPENSSL_CONF_INCLUDE> can also be
used to specify the path to prepend to all .include paths.
There can be optional B<=> character and whitespace characters between
B<.include> directive and the path which can be useful in cases the
configuration file needs to be loaded by old OpenSSL versions which do
not support the B<.include> syntax. They would bail out with error
if the B<=> character is not present but with it they just ignore
the include.
Pragmas can be specified with the B<.pragma> directive.
See L</PRAGMAS> for mor information.
Each section in a configuration file consists of a number of name and
value pairs of the form B<name=value>
The B<name> string can contain any alphanumeric characters as well as
a few punctuation symbols such as B<.> B<,> B<;> and B<_>.
The B<value> string consists of the string following the B<=> character
until end of line with any leading and trailing white space removed.
The value string undergoes variable expansion. This can be done by
including the form B<$var> or B<${var}>: this will substitute the value
of the named variable in the current section. It is also possible to
substitute a value from another section using the syntax B<$section::name>
or B<${section::name}>. By using the form B<$ENV::name> environment
variables can be substituted. It is also possible to assign values to
environment variables by using the name B<ENV::name>, this will work
if the program looks up environment variables using the B<CONF> library
instead of calling getenv() directly. The value string must not exceed 64k in
length after variable expansion. Otherwise an error will occur.
It is possible to escape certain characters by using any kind of quote
or the B<\> character. By making the last character of a line a B<\>
a B<value> string can be spread across multiple lines. In addition
the sequences B<\n>, B<\r>, B<\b> and B<\t> are recognized.
All expansion and escape rules as described above that apply to B<value>
also apply to the path of the B<.include> directive.
=head1 PRAGMAS
Pragmas can be used to change the behavior of the configuration file
parser, among others. Currently supported pragmas are:
=over 4
=item B<.pragma> B<dollarid>:I<value>
I<value> can be one of:
=over 4
=item B<"on"> or B<"true">
this signifies that dollar signs are considered an identity character
from this point on and that variable expansion requires the use of
braces or parentheses. In other words, C<foo$bar> will be considered
a name instead of C<foo> followed by the expansion of the variable
C<bar>.
This is suitable for platforms where the dollar sign is commonly used
as part of names.
=item B<"off"> or B<"false">
Turns this pragma off, i.e. C<foo$bar> will be interpreted as C<foo>
followed by the expansion of the variable C<bar>.
=back
By default, this pragma is turned off.
=back
=head1 OPENSSL LIBRARY CONFIGURATION
Applications can automatically configure certain
aspects of OpenSSL using the master OpenSSL configuration file, or optionally
an alternative configuration file. The B<openssl> utility includes this
functionality: any sub command uses the master OpenSSL configuration file
unless an option is used in the sub command to use an alternative configuration
file.
To enable library configuration the default section needs to contain an
appropriate line which points to the main configuration section. The default
name is B<openssl_conf> which is used by the B<openssl> utility. Other
applications may use an alternative name such as B<myapplication_conf>.
All library configuration lines appear in the default section at the start
of the configuration file.
The configuration section should consist of a set of name value pairs which
contain specific module configuration information. The B<name> represents
the name of the I<configuration module>. The meaning of the B<value> is
module specific: it may, for example, represent a further configuration
section containing configuration module specific information. E.g.:
# This must be in the default section
openssl_conf = openssl_init
[openssl_init]
oid_section = new_oids
engines = engine_section
providers = provider_section
[new_oids]
... new oids here ...
[engine_section]
... engine stuff here ...
[provider_section]
... provider stuff here ...
The features of each configuration module are described below.
=head2 ASN1 Object Configuration Module
This module has the name B<oid_section>. The value of this variable points
to a section containing name value pairs of OIDs: the name is the OID short
and long name, the value is the numerical form of the OID. Although some of
the B<openssl> utility sub commands already have their own ASN1 OBJECT section
functionality not all do. By using the ASN1 OBJECT configuration module
B<all> the B<openssl> utility sub commands can see the new objects as well
as any compliant applications. For example:
[new_oids]
some_new_oid = 1.2.3.4
some_other_oid = 1.2.3.5
It is also possible to set the value to the long name followed
by a comma and the numerical OID form. For example:
shortName = some object long name, 1.2.3.4
=head2 Engine Configuration Module
This ENGINE configuration module has the name B<engines>. The value of this
variable points to a section containing further ENGINE configuration
information.
The section pointed to by B<engines> is a table of engine names (though see
B<engine_id> below) and further sections containing configuration information
specific to each ENGINE.
Each ENGINE specific section is used to set default algorithms, load
dynamic, perform initialization and send ctrls. The actual operation performed
depends on the I<command> name which is the name of the name value pair. The
currently supported commands are listed below.
For example:
[engine_section]
# Configure ENGINE named "foo"
foo = foo_section
# Configure ENGINE named "bar"
bar = bar_section
[foo_section]
... foo ENGINE specific commands ...
[bar_section]
... "bar" ENGINE specific commands ...
The command B<engine_id> is used to give the ENGINE name. If used this
command must be first. For example:
[engine_section]
# This would normally handle an ENGINE named "foo"
foo = foo_section
[foo_section]
# Override default name and use "myfoo" instead.
engine_id = myfoo
The command B<dynamic_path> loads and adds an ENGINE from the given path. It
is equivalent to sending the ctrls B<SO_PATH> with the path argument followed
by B<LIST_ADD> with value 2 and B<LOAD> to the dynamic ENGINE. If this is
not the required behaviour then alternative ctrls can be sent directly
to the dynamic ENGINE using ctrl commands.
The command B<init> determines whether to initialize the ENGINE. If the value
is B<0> the ENGINE will not be initialized, if B<1> and attempt it made to
initialized the ENGINE immediately. If the B<init> command is not present
then an attempt will be made to initialize the ENGINE after all commands in
its section have been processed.
The command B<default_algorithms> sets the default algorithms an ENGINE will
supply using the functions ENGINE_set_default_string().
If the name matches none of the above command names it is assumed to be a
ctrl command which is sent to the ENGINE. The value of the command is the
argument to the ctrl command. If the value is the string B<EMPTY> then no
value is sent to the command.
For example:
[engine_section]
# Configure ENGINE named "foo"
foo = foo_section
[foo_section]
# Load engine from DSO
dynamic_path = /some/path/fooengine.so
# A foo specific ctrl.
some_ctrl = some_value
# Another ctrl that doesn't take a value.
other_ctrl = EMPTY
# Supply all default algorithms
default_algorithms = ALL
=head2 Provider Configuration Module
This provider configuration module has the name B<providers>. The
value of this variable points to a section containing further provider
configuration information.
The section pointed to by B<providers> is a table of provider names
(though see B<identity> below) and further sections containing
configuration information specific to each provider module.
Each provider specific section is used to load its module, perform
activation and set parameters to pass to the provider on demand. The
actual operation performed depends on the name of the name value pair.
The currently supported commands are listed below.
For example:
[provider_section]
# Configure provider named "foo"
foo = foo_section
# Configure provider named "bar"
bar = bar_section
[foo_section]
... "foo" provider specific parameters ...
[bar_section]
... "bar" provider specific parameters ...
The command B<identity> is used to give the provider name. For example:
[provider_section]
# This would normally handle a provider named "foo"
foo = foo_section
[foo_section]
# Override default name and use "myfoo" instead.
identity = myfoo
The parameter B<module> loads and adds a provider module from the
given module path. That path may be a simple filename, a relative
path or an absolute path.
The parameter B<activate> determines whether to activate the
provider. The value has no importance, the presence of the parameter
is enough for activation to take place.
All parameters in the section as well as sub-sections are made
available to the provider.
=head2 EVP Configuration Module
This module has the name B<alg_section> which points to a section containing
algorithm commands.
The supported algorithm commands are:
=over 4
=item B<default_properties>
The value may be anything that is acceptable as a property query
string for EVP_set_default_properties().
=item B<fips_mode> (deprecated)
The value is a boolean that can be B<yes> or B<no>. If the value is
B<yes>, this is exactly equivalent to:
default_properties = fips=yes
If the value is B<no>, nothing happens.
=back
These two commands should not be used together, as there is no control
over how they affect each other.
The use of B<fips_mode> is strongly discouraged and is only present
for backward compatibility with earlier OpenSSL FIPS modules.
=head2 SSL Configuration Module
This module has the name B<ssl_conf> which points to a section containing
SSL configurations.
Each line in the SSL configuration section contains the name of the
configuration and the section containing it.
Each configuration section consists of command value pairs for B<SSL_CONF>.
Each pair will be passed to a B<SSL_CTX> or B<SSL> structure if it calls
SSL_CTX_config() or SSL_config() with the appropriate configuration name.
Note: any characters before an initial dot in the configuration section are
ignored so the same command can be used multiple times.
For example:
ssl_conf = ssl_sect
[ssl_sect]
server = server_section
[server_section]
RSA.Certificate = server-rsa.pem
ECDSA.Certificate = server-ecdsa.pem
Ciphers = ALL:!RC4
The system default configuration with name B<system_default> if present will
be applied during any creation of the B<SSL_CTX> structure.
Example of a configuration with the system default:
ssl_conf = ssl_sect
[ssl_sect]
system_default = system_default_sect
[system_default_sect]
MinProtocol = TLSv1.2
=head1 NOTES
If a configuration file attempts to expand a variable that doesn't exist
then an error is flagged and the file will not load. This can happen
if an attempt is made to expand an environment variable that doesn't
exist. For example in a previous version of OpenSSL the default OpenSSL
master configuration file used the value of B<HOME> which may not be
defined on non Unix systems and would cause an error.
This can be worked around by including a B<default> section to provide
a default value: then if the environment lookup fails the default value
will be used instead. For this to work properly the default value must
be defined earlier in the configuration file than the expansion. See
the B<EXAMPLES> section for an example of how to do this.
If the same variable exists in the same section then all but the last
value will be silently ignored. In certain circumstances such as with
DNs the same field may occur multiple times. This is usually worked
around by ignoring any characters before an initial B<.> e.g.
1.OU="My first OU"
2.OU="My Second OU"
=head1 EXAMPLES
Here is a sample configuration file using some of the features
mentioned above.
# This is the default section.
HOME=/temp
configdir=$ENV::HOME/config
[ section_one ]
# We are now in section one.
# Quotes permit leading and trailing whitespace
any = " any variable name "
other = A string that can \
cover several lines \
by including \\ characters
message = Hello World\n
[ section_two ]
greeting = $section_one::message
This next example shows how to expand environment variables safely.
Suppose you want a variable called B<tmpfile> to refer to a
temporary filename. The directory it is placed in can determined by
the B<TEMP> or B<TMP> environment variables but they may not be
set to any value at all. If you just include the environment variable
names and the variable doesn't exist then this will cause an error when
an attempt is made to load the configuration file. By making use of the
default section both values can be looked up with B<TEMP> taking
priority and B</tmp> used if neither is defined:
TMP=/tmp
# The above value is used if TMP isn't in the environment
TEMP=$ENV::TMP
# The above value is used if TEMP isn't in the environment
tmpfile=${ENV::TEMP}/tmp.filename
Simple OpenSSL library configuration example to enter FIPS mode:
# Default appname: should match "appname" parameter (if any)
# supplied to CONF_modules_load_file et al.
openssl_conf = openssl_conf_section
[openssl_conf_section]
# Configuration module list
alg_section = evp_sect
[evp_sect]
# Set to "yes" to enter FIPS mode if supported
fips_mode = yes
Note: in the above example you will get an error in non FIPS capable versions
of OpenSSL.
Simple OpenSSL library configuration to make TLS 1.3 the system-default
minimum TLS version:
# Toplevel section for openssl (including libssl)
openssl_conf = default_conf_section
[default_conf_section]
# We only specify configuration for the "ssl module"
ssl_conf = ssl_section
[ssl_section]
system_default = system_default_section
[system_default_section]
MinProtocol = TLSv1.3
More complex OpenSSL library configuration. Add OID and don't enter FIPS mode:
# Default appname: should match "appname" parameter (if any)
# supplied to CONF_modules_load_file et al.
openssl_conf = openssl_conf_section
[openssl_conf_section]
# Configuration module list
alg_section = evp_sect
oid_section = new_oids
[evp_sect]
# This will have no effect as FIPS mode is off by default.
# Set to "yes" to enter FIPS mode, if supported
fips_mode = no
[new_oids]
# New OID, just short name
newoid1 = 1.2.3.4.1
# New OID shortname and long name
newoid2 = New OID 2 long name, 1.2.3.4.2
The above examples can be used with any application supporting library
configuration if "openssl_conf" is modified to match the appropriate "appname".
For example if the second sample file above is saved to "example.cnf" then
the command line:
OPENSSL_CONF=example.cnf openssl asn1parse -genstr OID:1.2.3.4.1
will output:
0:d=0 hl=2 l= 4 prim: OBJECT :newoid1
showing that the OID "newoid1" has been added as "1.2.3.4.1".
=head1 ENVIRONMENT
=over 4
=item B<OPENSSL_CONF>
The path to the config file.
Ignored in set-user-ID and set-group-ID programs.
=item B<OPENSSL_ENGINES>
The path to the engines directory.
Ignored in set-user-ID and set-group-ID programs.
=item B<OPENSSL_MODULES>
The path to the directory with OpenSSL modules, such as providers.
Ignored in set-user-ID and set-group-ID programs.
=item B<OPENSSL_CONF_INCLUDE>
The optional path to prepend to all .include paths.
=back
=head1 BUGS
Currently there is no way to include characters using the octal B<\nnn>
form. Strings are all null terminated so nulls cannot form part of
the value.
The escaping isn't quite right: if you want to use sequences like B<\n>
you can't use any quote escaping on the same line.
Files are loaded in a single pass. This means that an variable expansion
will only work if the variables referenced are defined earlier in the
file.
=head1 HISTORY
An undocumented API, NCONF_WIN32(), used a slightly different set
of parsing rules there were intended to be tailored to
the Microsoft Windows platform.
Specifically, the backslash character was not an escape character and
could be used in pathnames, only the double-quote character was recognized,
and comments began with a semi-colon.
This function was deprecated in OpenSSL 3.0; applications with
configuration files using that syntax will have to be modified.
=head1 SEE ALSO
L<openssl-x509(1)>, L<openssl-req(1)>, L<openssl-ca(1)>, L<fips_config(5)>
=head1 COPYRIGHT
Copyright 2000-2018 The OpenSSL Project Authors. All Rights Reserved.
Licensed under the Apache License 2.0 (the "License"). You may not use
this file except in compliance with the License. You can obtain a copy
in the file LICENSE in the source distribution or at
L<https://www.openssl.org/source/license.html>.
=cut