-
OPENSSL INSTALLATION
--------------------
This document describes installation on all supported operating
- systems (the Linux/Unix family, OpenVMS and Windows)
+ systems (the Unix/Linux family (which includes Mac OS/X), OpenVMS,
+ and Windows).
To install OpenSSL, you will need:
For additional platform specific requirements, solutions to specific
issues and other details, please read one of these:
+ * NOTES.UNIX (any supported Unix like system)
* NOTES.VMS (OpenVMS)
* NOTES.WIN (any supported Windows)
* NOTES.DJGPP (DOS platform with DJGPP)
+ * NOTES.ANDROID (obviously Android [NDK])
Notational conventions in this document
---------------------------------------
If you want to just get on with it, do:
- on Unix:
+ on Unix (again, this includes Mac OS/X):
$ ./config
$ make
$ @config --prefix=PROGRAM:[INSTALLS] --openssldir=SYS$MANAGER:[OPENSSL]
+ (Note: if you do add options to the configuration command, please make sure
+ you've read more than just this Quick Start, such as relevant NOTES.* files,
+ the options outline below, as configuration options may change the outcome
+ in otherwise unexpected ways)
+
Configuration Options
---------------------
put together one-size-fits-all instructions. You might
have to pass more flags or set up environment variables
to actually make it work. Android and iOS cases are
- discussed in corresponding Configurations/10-main.cf
- sections. But there are cases when this option alone is
+ discussed in corresponding Configurations/15-*.conf
+ files. But there are cases when this option alone is
sufficient. For example to build the mingw64 target on
Linux "--cross-compile-prefix=x86_64-w64-mingw32-"
works. Naturally provided that mingw packages are
"--cross-compile-prefix=mipsel-linux-gnu-" suffices
in such case. Needless to mention that you have to
invoke ./Configure, not ./config, and pass your target
- name explicitly.
+ name explicitly. Also, note that --openssldir refers
+ to target's file system, not one you are building on.
--debug
- Build OpenSSL with debugging symbols.
+ Build OpenSSL with debugging symbols and zero optimization
+ level.
--libdir=DIR
The name of the directory under the top of the installation
without a path). This flag must be provided if the
zlib-dynamic option is not also used. If zlib-dynamic is used
then this flag is optional and a default value ("ZLIB1") is
- used if not provided.
+ used if not provided.
On VMS: this is the filename of the zlib library (with or
without a path). This flag is optional and if not provided
then "GNV$LIBZSHR", "GNV$LIBZSHR32" or "GNV$LIBZSHR64" is
used by default depending on the pointer size chosen.
+
+ --with-rand-seed=seed1[,seed2,...]
+ A comma separated list of seeding methods which will be tried
+ by OpenSSL in order to obtain random input (a.k.a "entropy")
+ for seeding its cryptographically secure random number
+ generator (CSPRNG). The current seeding methods are:
+
+ os: Use a trusted operating system entropy source.
+ This is the default method if such an entropy
+ source exists.
+ getrandom: Use the L<getrandom(2)> or equivalent system
+ call.
+ devrandom: Use the the first device from the DEVRANDOM list
+ which can be opened to read random bytes. The
+ DEVRANDOM preprocessor constant expands to
+ "/dev/urandom","/dev/random","/dev/srandom" on
+ most unix-ish operating systems.
+ egd: Check for an entropy generating daemon.
+ rdcpu: Use the RDSEED or RDRAND command if provided by
+ the CPU.
+ librandom: Use librandom (not implemented yet).
+ none: Disable automatic seeding. This is the default
+ on some operating systems where no suitable
+ entropy source exists, or no support for it is
+ implemented yet.
+
+ For more information, see the section 'Note on random number
+ generation' at the end of this document.
+
no-afalgeng
Don't build the AFALG engine. This option will be forced if
on a platform that does not support AFALG.
no-shared option.
no-asm
- Do not use assembler code. On some platforms a small amount
- of assembler code may still be used.
+ Do not use assembler code. This should be viewed as
+ debugging/trouble-shooting option rather than production.
+ On some platforms a small amount of assembler code may
+ still be used even with this option.
no-async
Do not build support for async operations.
error strings. For a statically linked application this may
be undesirable if small executable size is an objective.
+ no-autoload-config
+ Don't automatically load the default openssl.cnf file.
+ Typically OpenSSL will automatically load a system config
+ file which configures default ssl options.
no-capieng
Don't build the CAPI engine. This option will be forced if
enable-ec_nistp_64_gcc_128
Enable support for optimised implementations of some commonly
- used NIST elliptic curves. This is only supported on some
- platforms.
+ used NIST elliptic curves.
+ This is only supported on platforms:
+ - with little-endian storage of non-byte types
+ - that tolerate misaligned memory references
+ - where the compiler:
+ - supports the non-standard type __uint128_t
+ - defines the built-in macro __SIZEOF_INT128__
enable-egd
Build support for gathering entropy from EGD (Entropy
no-err
Don't compile in any error strings.
+ enable-external-tests
+ Enable building of integration with external test suites.
+ This is a developer option and may not work on all platforms.
+ The only supported external test suite at the current time is
+ the BoringSSL test suite. See the file test/README.external
+ for further details.
+
no-filenames
Don't compile in filename and line number information (e.g.
for errors and memory allocation).
available if the GOST algorithms are also available through
loading an externally supplied engine.
- enable-heartbeats
- Build support for DTLS heartbeats.
-
no-hw-padlock
Don't build the padlock engine.
Don't build SRTP support
no-sse2
- Exclude SSE2 code paths. Normally SSE2 extension is
- detected at run-time, but the decision whether or not the
- machine code will be executed is taken solely on CPU
- capability vector. This means that if you happen to run OS
- kernel which does not support SSE2 extension on Intel P4
- processor, then your application might be exposed to
- "illegal instruction" exception. There might be a way
- to enable support in kernel, e.g. FreeBSD kernel can be
- compiled with CPU_ENABLE_SSE, and there is a way to
- disengage SSE2 code paths upon application start-up,
- but if you aim for wider "audience" running such kernel,
- consider no-sse2. Both the 386 and no-asm options imply
- no-sse2.
+ Exclude SSE2 code paths from 32-bit x86 assembly modules.
+ Normally SSE2 extension is detected at run-time, but the
+ decision whether or not the machine code will be executed
+ is taken solely on CPU capability vector. This means that
+ if you happen to run OS kernel which does not support SSE2
+ extension on Intel P4 processor, then your application
+ might be exposed to "illegal instruction" exception.
+ There might be a way to enable support in kernel, e.g.
+ FreeBSD kernel can be compiled with CPU_ENABLE_SSE, and
+ there is a way to disengage SSE2 code paths upon application
+ start-up, but if you aim for wider "audience" running
+ such kernel, consider no-sse2. Both the 386 and
+ no-asm options imply no-sse2.
enable-ssl-trace
Build with the SSL Trace capabilities (adds the "-trace"
has an impact when not built "shared".
no-stdio
- Don't use any C "stdio" features. Only libcrypto and libssl
- can be built in this way. Using this option will suppress
+ Don't use anything from the C header file "stdio.h" that
+ makes use of the "FILE" type. Only libcrypto and libssl can
+ be built in this way. Using this option will suppress
building the command line applications. Additionally since
the OpenSSL tests also use the command line applications the
tests will also be skipped.
+ no-tests
+ Don't build test programs or run any test.
+
no-threads
Don't try to build with support for multi-threaded
applications.
where loading of shared libraries is supported.
386
- On Intel hardware, use the 80386 instruction set only
- (the default x86 code is more efficient, but requires at
- least a 486). Note: Use compiler flags for any other CPU
- specific configuration, e.g. "-m32" to build x86 code on
- an x64 system.
+ In 32-bit x86 builds, when generating assembly modules,
+ use the 80386 instruction set only (the default x86 code
+ is more efficient, but requires at least a 486). Note:
+ This doesn't affect code generated by compiler, you're
+ likely to complement configuration command line with
+ suitable compiler-specific option.
no-<prot>
Don't build support for negotiating the specified SSL/TLS
- protocol (one of ssl, ssl3, tls, tls1, tls1_1, tls1_2, dtls,
- dtls1 or dtls1_2). If "no-tls" is selected then all of tls1,
- tls1_1 and tls1_2 are disabled. Similarly "no-dtls" will
- disable dtls1 and dtls1_2. The "no-ssl" option is synonymous
- with "no-ssl3". Note this only affects version negotiation.
- OpenSSL will still provide the methods for applications to
- explicitly select the individual protocol versions.
+ protocol (one of ssl, ssl3, tls, tls1, tls1_1, tls1_2,
+ tls1_3, dtls, dtls1 or dtls1_2). If "no-tls" is selected then
+ all of tls1, tls1_1, tls1_2 and tls1_3 are disabled.
+ Similarly "no-dtls" will disable dtls1 and dtls1_2. The
+ "no-ssl" option is synonymous with "no-ssl3". Note this only
+ affects version negotiation. OpenSSL will still provide the
+ methods for applications to explicitly select the individual
+ protocol versions.
no-<prot>-method
As for no-<prot> but in addition do not build the methods for
applications to explicitly select individual protocol
- versions.
+ versions. Note that there is no "no-tls1_3-method" option
+ because there is no application method for TLSv1.3. Using
+ individual protocol methods directly is deprecated.
+ Applications should use TLS_method() instead.
enable-<alg>
Build with support for the specified algorithm, where <alg>
no-<alg>
Build without support for the specified algorithm, where
- <alg> is one of: bf, blake2, camellia, cast, chacha, cmac,
- des, dh, dsa, ecdh, ecdsa, idea, md4, mdc2, ocb, poly1305,
- rc2, rc4, rmd160, scrypt, seed or whirlpool. The "ripemd"
- algorithm is deprecated and if used is synonymous with rmd160.
+ <alg> is one of: aria, bf, blake2, camellia, cast, chacha,
+ cmac, des, dh, dsa, ecdh, ecdsa, idea, md4, mdc2, ocb,
+ poly1305, rc2, rc4, rmd160, scrypt, seed, siphash, sm2, sm3,
+ sm4 or whirlpool. The "ripemd" algorithm is deprecated and
+ if used is synonymous with rmd160.
+
+ -Dxxx, -Ixxx, -Wp, -lxxx, -Lxxx, -Wl, -rpath, -R, -framework, -static
+ These system specific options will be recognised and
+ passed through to the compiler to allow you to define
+ preprocessor symbols, specify additional libraries, library
+ directories or other compiler options. It might be worth
+ noting that some compilers generate code specifically for
+ processor the compiler currently executes on. This is not
+ necessarily what you might have in mind, since it might be
+ unsuitable for execution on other, typically older,
+ processor. Consult your compiler documentation.
+
+ Take note of the VAR=value documentation below and how
+ these flags interact with those variables.
+
+ -xxx, +xxx
+ Additional options that are not otherwise recognised are
+ passed through as they are to the compiler as well. Again,
+ consult your compiler documentation.
+
+ Take note of the VAR=value documentation below and how
+ these flags interact with those variables.
+
+ VAR=value
+ Assignment of environment variable for Configure. These
+ work just like normal environment variable assignments,
+ but are supported on all platforms and are confined to
+ the configuration scripts only. These assignments override
+ the corresponding value in the inherited environment, if
+ there is one.
+
+ The following variables are used as "make variables" and
+ can be used as an alternative to giving preprocessor,
+ compiler and linker options directly as configuration.
+ The following variables are supported:
+
+ AR The static library archiver.
+ ARFLAGS Flags for the static library archiver.
+ AS The assembler compiler.
+ ASFLAGS Flags for the assembler compiler.
+ CC The C compiler.
+ CFLAGS Flags for the C compiler.
+ CXX The C++ compiler.
+ CXXFLAGS Flags for the C++ compiler.
+ CPP The C/C++ preprocessor.
+ CPPFLAGS Flags for the C/C++ preprocessor.
+ CPPDEFINES List of CPP macro definitions, separated
+ by a platform specific character (':' or
+ space for Unix, ';' for Windows, ',' for
+ VMS). This can be used instead of using
+ -D (or what corresponds to that on your
+ compiler) in CPPFLAGS.
+ CPPINCLUDES List of CPP inclusion directories, separated
+ the same way as for CPPDEFINES. This can
+ be used instead of -I (or what corresponds
+ to that on your compiler) in CPPFLAGS.
+ HASHBANGPERL Perl invocation to be inserted after '#!'
+ in public perl scripts (only relevant on
+ Unix).
+ LD The program linker (not used on Unix, $(CC)
+ is used there).
+ LDFLAGS Flags for the shared library, DSO and
+ program linker.
+ LDLIBS Extra libraries to use when linking.
+ Takes the form of a space separated list
+ of library specifications on Unix and
+ Windows, and as a comma separated list of
+ libraries on VMS.
+ RANLIB The library archive indexer.
+ RC The Windows resources manipulator.
+ RCFLAGS Flags for the Windows reources manipulator.
+ RM The command to remove files and directories.
+
+ These cannot be mixed with compiling / linking flags given
+ on the command line. In other words, something like this
+ isn't permitted.
+
+ ./config -DFOO CPPFLAGS=-DBAR -DCOOKIE
+
+ Backward compatibility note:
+
+ To be compatible with older configuration scripts, the
+ environment variables are ignored if compiling / linking
+ flags are given on the command line, except for these:
+
+ AR, CC, CXX, CROSS_COMPILE, HASHBANGPERL, PERL, RANLIB, RC
+ and WINDRES
+
+ For example, the following command will not see -DBAR:
+
+ CPPFLAGS=-DBAR ./config -DCOOKIE
+
+ However, the following will see both set variables:
+
+ CC=gcc CROSS_COMPILE=x86_64-w64-mingw32- \
+ ./config -DCOOKIE
+
+ reconf
+ reconfigure
+ Reconfigure from earlier data. This fetches the previous
+ command line options and environment from data saved in
+ "configdata.pm", and runs the configuration process again,
+ using these options and environment.
+ Note: NO other option is permitted together with "reconf".
+ This means that you also MUST use "./Configure" (or
+ what corresponds to that on non-Unix platforms) directly
+ to invoke this option.
+ Note: The original configuration saves away values for ALL
+ environment variables that were used, and if they weren't
+ defined, they are still saved away with information that
+ they weren't originally defined. This information takes
+ precedence over environment variables that are defined
+ when reconfiguring.
+
+ Displaying configuration data
+ -----------------------------
+
+ The configuration script itself will say very little, and finishes by
+ creating "configdata.pm". This perl module can be loaded by other scripts
+ to find all the configuration data, and it can also be used as a script to
+ display all sorts of configuration data in a human readable form.
+
+ For more information, please do:
+
+ $ ./configdata.pm --help # Unix
- -Dxxx, -lxxx, -Lxxx, -fxxx, -mXXX, -Kxxx
- These system specific options will be passed through to the
- compiler to allow you to define preprocessor symbols, specify
- additional libraries, library directories or other compiler
- options.
+ or
+ $ perl configdata.pm --help # Windows and VMS
Installation in Detail
----------------------
("openssl"). The libraries will be built in the top-level directory,
and the binary will be in the "apps" subdirectory.
- If the build fails, look at the output. There may be reasons for
- the failure that aren't problems in OpenSSL itself (like missing
- standard headers). If you are having problems you can get help by
- sending an email to the openssl-users email list (see
- https://www.openssl.org/community/mailinglists.html for details). If it
- is a bug with OpenSSL itself, please report the problem to
- <rt@openssl.org> (note that your message will be recorded in the request
- tracker publicly readable at
- https://www.openssl.org/community/index.html#bugs and will be
- forwarded to a public mailing list). Please check out the request
- tracker. Maybe the bug was already reported or has already been
- fixed.
+ Troubleshooting:
+
+ If the build fails, look at the output. There may be reasons
+ for the failure that aren't problems in OpenSSL itself (like
+ missing standard headers).
+
+ If the build succeeded previously, but fails after a source or
+ configuration change, it might be helpful to clean the build tree
+ before attempting another build. Use this command:
+
+ $ make clean # Unix
+ $ mms clean ! (or mmk) OpenVMS
+ $ nmake clean # Windows
- (If you encounter assembler error messages, try the "no-asm"
- configuration option as an immediate fix.)
+ Assembler error messages can sometimes be sidestepped by using the
+ "no-asm" configuration option.
Compiling parts of OpenSSL with gcc and others with the system
compiler will result in unresolved symbols on some systems.
+ If you are still having problems you can get help by sending an email
+ to the openssl-users email list (see
+ https://www.openssl.org/community/mailinglists.html for details). If
+ it is a bug with OpenSSL itself, please open an issue on GitHub, at
+ https://github.com/openssl/openssl/issues. Please review the existing
+ ones first; maybe the bug was already reported or has already been
+ fixed.
+
3. After a successful build, the libraries should be tested. Run:
$ make test # Unix
$ nmake TESTS='test_rsa test_dsa' test # Windows
And of course, you can combine (Unix example shown):
-
+
$ make VERBOSE=1 TESTS='test_rsa test_dsa' test
You can find the list of available tests like this:
compiler optimization flags from the CFLAGS line in Makefile and
run "make clean; make" or corresponding.
- Please send bug reports to <rt@openssl.org>.
+ To report a bug please open an issue on GitHub, at
+ https://github.com/openssl/openssl/issues.
For more details on how the make variables TESTS can be used,
see section TESTS in Detail below.
command symbols.
[.SYSTEST] Contains the installation verification procedure.
[.HTML] Contains the HTML rendition of the manual pages.
-
+
Additionally, install will add the following directories under
OPENSSLDIR (the directory given with --openssldir or its default)
uninstall
Uninstall all OpenSSL components.
+ reconfigure
+ reconf
+ Re-run the configuration process, as exactly as the last time
+ as possible.
+
update
This is a developer option. If you are developing a patch for
OpenSSL you may need to use this if you want to update
-xxx Removes 'xxx' from the current set of tests. If this is the
first token in the list, the current set of tests is first
assigned the whole set of available tests, effectively making
- this token equivalent to TESTS="alltests -xxx"
+ this token equivalent to TESTS="alltests -xxx".
+ nn Adds the test group 'nn' (which is a number) to the current
+ set of tests.
+ -nn Removes the test group 'nn' from the current set of tests.
+ If this is the first token in the list, the current set of
+ tests is first assigned the whole set of available tests,
+ effectively making this token equivalent to
+ TESTS="alltests -xxx".
Also, all tokens except for "alltests" may have wildcards, such as *.
(on Unix and Windows, BSD style wildcards are supported, while on VMS,
$ make TESTS='test_ssl* -test_ssl_*' test
+ Example: Only test group 10:
+
+ $ make TESTS='10'
+
+ Example: All tests except the slow group (group 99):
+
+ $ make TESTS='-99'
+
+ Example: All tests in test groups 80 to 99 except for tests in group 90:
+
+ $ make TESTS='[89]? -90'
+
Note on multi-threading
-----------------------
and libssl-1_1-x64.dll for 64-bit x86_64 Windows, and libcrypto-1_1-ia64.dll
and libssl-1_1-ia64.dll for IA64 Windows. With MSVC, the import libraries
are named libcrypto.lib and libssl.lib, while with MingW, they are named
- libcrypto.dll.a and libddl.dll.a.
+ libcrypto.dll.a and libssl.dll.a.
On VMS, shareable images (VMS speak for shared libraries) are named
ossl$libcrypto0101_shr.exe and ossl$libssl0101_shr.exe. However, when
Availability of cryptographically secure random numbers is required for
secret key generation. OpenSSL provides several options to seed the
- internal PRNG. If not properly seeded, the internal PRNG will refuse
+ internal CSPRNG. If not properly seeded, the internal CSPRNG will refuse
to deliver random bytes and a "PRNG not seeded error" will occur.
- On systems without /dev/urandom (or similar) device, it may be necessary
- to install additional support software to obtain a random seed.
- Please check out the manual pages for RAND_add(), RAND_bytes(), RAND_egd(),
- and the FAQ for more information.
+ The seeding method can be configured using the --with-rand-seed option,
+ which can be used to specify a comma separated list of seed methods.
+ However in most cases OpenSSL will choose a suitable default method,
+ so it is not necessary to explicitely provide this option. Note also
+ that not all methods are available on all platforms.
+
+ I) On operating systems which provide a suitable randomness source (in
+ form of a system call or system device), OpenSSL will use the optimal
+ available method to seed the CSPRNG from the operating system's
+ randomness sources. This corresponds to the option --with-rand-seed=os.
+
+ II) On systems without such a suitable randomness source, automatic seeding
+ and reseeding is disabled (--with-rand-seed=none) and it may be necessary
+ to install additional support software to obtain a random seed and reseed
+ the CSPRNG manually. Please check out the manual pages for RAND_add(),
+ RAND_bytes(), RAND_egd(), and the FAQ for more information.
projects / openssl.git / blobdiff