This test was accidentally passing the value 2048 into the keytype
parameter of PKCS12_create, not the mac_iter parameter (because it had
one too many `nil`s in the call). This value is invalid, and will make
OpenSSL perform an out-of-bounds read which is caught when compiling
with ASAN.
This commit fixes the tests, and also adds some validation to
PKCS12.create to make sure any keytype passed is actually valid. Since
there only two valid keytype constants, and the whole feature is an
export-grade crypto era thing only ever supported by old MSIE, it seems
far more likely that code in the whild is using keytype similarly by
mistake rather than as intended. So this validation might catch that.
47028686d2
[pkuzco: expanded the fix for other content types]
[ky: adjusted formatting and the exception type]
07eceb7f63
Co-authored-by: pkuzco <b.naamneh@gmail.com>
Co-authored-by: Kazuki Yamaguchi <k@rhe.jp>
OpenSSL::Cipher#update currently allocates the output buffer with size
(input data length)+(the block size of the cipher). This is insufficient
for the id-aes{128,192,256}-wrap-pad (AES keywrap with padding) ciphers.
They have a block size of 8 bytes, but the output may be up to 15 bytes
larger than the input.
Use (input data length)+EVP_MAX_BLOCK_LENGTH (== 32) as the output
buffer size, instead. OpenSSL doesn't provide a generic way to tell the
maximum required buffer size for ciphers, but this is large enough for
all algorithms implemented in current versions of OpenSSL.
Fixes: https://bugs.ruby-lang.org/issues/202363035559f54
RFC 2986, section 4.1 only defines version 1 for CSRs. This version
is encoded as a 0. Starting with OpenSSL 3.3, setting the CSR version
to anything but 1 fails.
Do not attempt to generate a CSR with invalid version (which now fails)
and invalidate the CSR in test_sign_and_verify_rsa_sha1 by changing its
subject rather than using an invalid version.
This commit fixes the following error.
```
2) Error: test_version(OpenSSL::TestX509Request): OpenSSL::X509::RequestError:
X509_REQ_set_version: passed invalid argument
/home/runner/work/openssl/openssl/test/openssl/test_x509req.rb:18:in `version='
/home/runner/work/openssl/openssl/test/openssl/test_x509req.rb:18:in `issue_csr'
/home/runner/work/openssl/openssl/test/openssl/test_x509req.rb:43:in
`test_version'
40: req = OpenSSL::X509::Request.new(req.to_der)
41: assert_equal(0, req.version)
42:
=> 43: req = issue_csr(1, @dn, @rsa1024, OpenSSL::Digest.new('SHA256'))
44: assert_equal(1, req.version)
45: req = OpenSSL::X509::Request.new(req.to_der)
46: assert_equal(1, req.version)
```
c06fdeb091
Note that I created the `dsa2048.pem` and signature text
(`signature_encoded.txt`), that is used as a text to create the `signature0` in
the `test_sign_verify` by the following steps with the `openssl` CLI on FIPS
module.
```
$ OPENSSL_DIR="${HOME}/.local/openssl-3.3.0-dev-fips-debug-1f03d33ef5"
$ export OPENSSL_CONF="${OPENSSL_DIR}/ssl/openssl_fips.cnf"
$ "${OPENSSL_DIR}/bin/openssl" dsaparam -out dsaparam2048.pem 2048
$ "${OPENSSL_DIR}/bin/openssl" gendsa -out dsa2048.pem dsaparam2048.pem
$ echo -n "Sign me!" > data.txt
$ "${OPENSSL_DIR}/bin/openssl" dgst -sha256 -sign dsa2048.pem data.txt > signature.txt
$ cat signature.txt | base64 > signature_encoded.txt
```
Skip the `test_DSAPrivateKey_encrypted` on FIPS because AES-128-CBC, the
password based encryption used in the PEM format uses MD5 for deriving the
encryption key from the password, and MD5 is not FIPS-approved.
See also the comment on the `test/openssl/utils.rb#omit_on_fips`.
4bdcb419a9
In some cases such as OpenSSL package in FreeBSD[1], the legacy provider is not
installed intentionally. So, we omit a test depending the legacy provider if the
legacy provider is not loadable.
For the test_openssl_provider_names test, we use base provider[2] instead of
legacy provider, because we would expect the base provider is always loadable
in OpenSSL 3 for now.
* [1] https://www.freshports.org/security/openssl/
* [2] https://wiki.openssl.org/index.php/OpenSSL_3.0#Providers7223da7730
This reverts commit 9b7543d588.
The case not providing the legacy provider can happen not only in FreeBSD but
also in other platforms. The fix is addressed in the next commit.
8c26fc26a1
This commit fixes the following errors in the tests.
Because the OpenSSL project changed the code to make the time string format
without second invalid. So, we drop the assertions.
```
1) Error: test_generalizedtime(OpenSSL::TestASN1): OpenSSL::ASN1::ASN1Error: generalizedtime is too short
/home/runner/work/ruby-openssl/ruby-openssl/test/openssl/test_asn1.rb:698:in `decode'
/home/runner/work/ruby-openssl/ruby-openssl/test/openssl/test_asn1.rb:698:in `decode_test'
/home/runner/work/ruby-openssl/ruby-openssl/test/openssl/test_asn1.rb:433:in `test_generalizedtime'
430: OpenSSL::ASN1::GeneralizedTime.new(Time.utc(9999, 9, 8, 23, 43, 39))
431: # LibreSSL 3.6.0 requires the seconds element
432: return if libressl?
=> 433: decode_test B(%w{ 18 0D }) + "201612081934Z".b,
434: OpenSSL::ASN1::GeneralizedTime.new(Time.utc(2016, 12, 8, 19, 34, 0))
435: # not implemented
436: # decode_test B(%w{ 18 13 }) + "2016120819+0930".b,
2) Error: test_utctime(OpenSSL::TestASN1): OpenSSL::ASN1::ASN1Error: utctime is too short
/home/runner/work/ruby-openssl/ruby-openssl/test/openssl/test_asn1.rb:698:in `decode'
/home/runner/work/ruby-openssl/ruby-openssl/test/openssl/test_asn1.rb:698:in `decode_test'
/home/runner/work/ruby-openssl/ruby-openssl/test/openssl/test_asn1.rb:411:in `test_utctime'
408: end
409: # Seconds is omitted. LibreSSL 3.6.0 requires it
410: return if libressl?
=> 411: decode_test B(%w{ 17 0B }) + "1609082343Z".b,
412: OpenSSL::ASN1::UTCTime.new(Time.utc(2016, 9, 8, 23, 43, 0))
413: # not implemented
414: # decode_test B(%w{ 17 11 }) + "5009082343+0930".b,
```
2e826d5715
We use dh2048_ffdhe2048.pem file (DH 2048 bits) instead of dh1024.pem file in
both non-FIPS and FIPS cases. Because the following command fails to generate
the pem file with 1024 bits. And the OpenSSL FIPS 140-2 security policy
document explains the DH public keys are allowed from 2048 bits.[1]
```
$ OPENSSL_CONF=/home/jaruga/.local/openssl-3.3.0-dev-fips-debug-1aa08644ec/ssl/openssl_fips.cnf \
/home/jaruga/.local/openssl-3.3.0-dev-fips-debug-1aa08644ec/bin/openssl \
dhparam -out dh1024.pem 1024
Generating DH parameters, 1024 bit long safe prime
dhparam: Generating DH key parameters failed
```
The dh2048_ffdhe2048.pem file was created by the following command with the
OpenSSL FIPS configuration file. The logic to generate the DH pem file is
different between non-FIPS and FIPS cases. In FIPS, it seems that the command
always returns the text defined as ffdhe2048 in the FFDHE groups in RFC 7919
unlike non-FIPS.[2]
As the generated pem file is a normal and valid PKCS#3-style group parameter, we
use the file for the non-FIPS case too.
```
$ OPENSSL_CONF=/home/jaruga/.local/openssl-3.3.0-dev-fips-debug-1aa08644ec/ssl/openssl_fips.cnf \
/home/jaruga/.local/openssl-3.3.0-dev-fips-debug-1aa08644ec/bin/openssl \
dhparam -out dh2048_ffdhe2048.pem 2048
```
Note that the hard-coded PEM-encoded string in the `test_DHparams` is
intentional to avoid modifying the content unintentionally.
* [1] https://www.openssl.org/source/ - OpenSSL 3.0.8 FIPS 140-2 security
policy document page 25, Table 10 – Public Keys - DH Public
- DH (2048/3072/4096/6144/8192) public key agreement key
* [2] RFC7919 - Appendix A.1: ffdhe2048
https://www.rfc-editor.org/rfc/rfc7919#appendix-A.1
Check that OpenSSL::PKey::EC.builtin_curves returns an array in the
expected format.
Similarly to OpenSSL::Cipher.ciphers, OpenSSL::PKey::EC.builtin_curves
returns a list of known named curves rather than actually usable ones.
https://github.com/ruby/openssl/issues/671 found that the list may
include unapproved (and thus unusable) curves when the FIPS module is
loaded.
c53cbabe00
The test case test_error_data utilizes the error message generated by
X509V3_EXT_nconf_nid(). The next commit will use X509V3_EXT_nconf(),
which generates a slightly different error message. Let's adapt the
check to it.
9cdfa3a4d1
instead of looking of NIDs and then using X509V3_EXT_nconf_nid,
instead just pass strings to X509V3_EXT_nconf, which has all the logic for
processing dealing with generic extensions
also process the oid through ln2nid() to retain compatibility.
[rhe: tweaked commit message and added a test case]
9f15741331
OpenSSL::X509::ExtensionFactory#create_ext and #create_extensions
accepts both sn (short names) and ln (long names) for registered OIDs.
This is different from the behavior of the openssl command-line utility
which accepts only sn in openssl.cnf keys.
Add a test case to check this.
91ae46c8d7
(https://github.com/ruby/openssl/pull/586)
String#unpack1 avoids the intermediate array created by String#unpack
for single elements, while also making a call to Array#first/[0]
unnecessary.
8eb0715a42
Remove the OSSL_DEBUG flag and OpenSSL.mem_check_start which is only
compiled when the flag is given. They are meant purely for development
of Ruby/OpenSSL.
OpenSSL.mem_check_start helped us find memory leak bugs in past, but
it is no longer working with the recent OpenSSL versions. Let's just
remove it now.
8c7a6a17e2
Error entries in the OpenSSL error queue may contain additional
contextual information associated with the error, which can be helpful
when debugging.
This "additional data" is currently only printed to stderr when
OpenSSL.debug is enabled. Let's include this in the exception messages
raised with ossl_raise(), too.
$ ruby -Ilib -ropenssl -e'OpenSSL.debug=true; OpenSSL::SSL::SSLContext.new.ecdh_curves="P-256:not-a-curve"'
-e:1: warning: error on stack: error:0A080106:SSL routines:gid_cb:passed invalid argument (group 'not-a-curve' cannot be set)
-e:1:in `ecdh_curves=': passed invalid argument (group 'not-a-curve' cannot be set) (OpenSSL::SSL::SSLError)
from -e:1:in `<main>'
1c5bbdd68e
Test that the error message contains the necessary text instead of the
exact match. Exceptions raised by ossl_raise() may carry additional
context information in the message.
fd6f76c0d3
When compiled with OpenSSL <= 1.1.1, OpenSSL::SSL::SSLContext#setup
does not raise an exception on an error return from
SSL_CTX_load_verify_locations(), but instead only prints a verbose-mode
warning. This is not helpful since it very likely indicates an actual
error, such as the specified file not being readable.
Also, OpenSSL's error queue is not correctly cleared:
$ ruby -w -ropenssl -e'OpenSSL.debug=true; ctx=OpenSSL::SSL::SSLContext.new; ctx.ca_file="bad-path"; ctx.setup; pp OpenSSL.errors'
-e:1: warning: can't set verify locations
["error:02001002:system library:fopen:No such file or directory",
"error:2006D080:BIO routines:BIO_new_file:no such file",
"error:0B084002:x509 certificate routines:X509_load_cert_crl_file: system lib"]
The behavior is currently different when compiled with OpenSSL >= 3.0:
SSLError is raised if SSL_CTX_load_verify_file() or
SSL_CTX_load_verify_dir() fails.
This inconsistency was unintentionally introduced by commit 5375a55ffc
("ssl: use SSL_CTX_load_verify_{file,dir}() if available", 2020-02-22).
However, raising SSLError seems more appropriate in this situation.
Let's adjust the OpenSSL <= 1.1.1 code so that it behaves the same way
as the OpenSSL >= 3.0 code currently does.
Fixes: https://github.com/ruby/openssl/issues/6497eb10f7b75
This reverts commit fc4629d246.
The test case "test_connect_certificate_verify_failed_exception_message"
does want to check the reason behind a certificate verification failure
to be included in the exception message.
c309745eb8
(https://github.com/ruby/openssl/pull/646)
Add OpenSSL::PKey.new_raw_private_key, #raw_private_key and public
equivalents. These methods are useful for importing and exporting keys
that support "raw private/public key". Currently, OpenSSL implements
X25519/X448 and Ed25519/Ed448 keys.
[rhe: rewrote commit message]
3f29525618
Co-authored-by: Bart de Water <bartdewater@gmail.com>
If x is a modular square root of a (mod p) then so is (p - x). Both
answers are valid. In particular, both 2 and 3 are valid square roots
of 4 (mod 5). Do not assume that a particular square root is chosen by
the algorithm. Indeed, the algorithm in OpenSSL and LibreSSL <= 3.7
returns a non-deterministic answer in many cases. LibreSSL 3.8 and
later will always return the smaller of the two possible answers. This
breaks the current test case.
Instead of checking for a particular square root, check that the square
of the claimed square root is the given value. This is always true. Add
the simplest test case where the answer is indeed non-deterministic.
93548ae959
It's to test the `OpenSSL::PKey.read` in the `test/openssl/test_pkey.rb`.
I added the pending status to the following tests failing on the FIPS mode
case in the `test/openssl/test_pkey.rb`.
* `test_ed25519`
* `test_x25519`
* `test_compare?`
8149cdf6e8
