FIPS provider correctly supports no-des build time option and doesn't
advertise DES related algorithms. However KAT test for DES is still
attempted to be executed and fails.
This prevents configuring FIPS provider without legacy behaviour as
defined in SP 800-131Arev2. Also see #25761 internal docs.
Fix `enable-fips no-des` build option, and add a daily checker for
"legacy-free" (as much as currently feasible) FIPS configuration.
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/25762)
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/25640)
fix https://github.com/openssl/openssl/issues/25112
As defined in the C standard:
In all cases the argument is an int, the value of which shall
be representable as an unsigned char or shall equal the value
of the macro EOF. If the argument has any other value, the
behavior is undefined.
This is because they're designed to work with the int values returned
by getc or fgetc; they need extra work to handle a char value.
If EOF is -1 (as it almost always is), with 8-bit bytes, the allowed
inputs to the ctype.h functions are:
{-1, 0, 1, 2, 3, ..., 255}.
However, on platforms where char is signed, such as x86 with the
usual ABI, code like
char *p = ...;
... isspace(*p) ...
may pass in values in the range:
{-128, -127, -126, ..., -2, -1, 0, 1, ..., 127}.
This has two problems:
1. Inputs in the set {-128, -127, -126, ..., -2} are forbidden.
2. The non-EOF byte 0xff is conflated with the value EOF = -1, so
even though the input is not forbidden, it may give the wrong
answer.
Casting char inputs to unsigned char first works around this, by
mapping the (non-EOF character) range {-128, -127, ..., -1} to {128,
129, ..., 255}, leaving no collisions with EOF. So the above
fragment needs to be:
char *p = ...;
... isspace((unsigned char)*p) ...
This patch inserts unsigned char casts where necessary. Most of the
cases I changed, I compile-tested using -Wchar-subscripts -Werror on
NetBSD, which defines the ctype.h functions as macros so that they
trigger the warning when the argument has type char. The exceptions
are under #ifdef __VMS or #ifdef _WIN32. I left alone calls where
the input is int where the cast would obviously be wrong; and I left
alone calls where the input is already unsigned char so the cast is
unnecessary.
Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Kurt Roeckx <kurt@roeckx.be>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/25113)
Changed all provider implementations that have a set_ctx_params()
to call this function instead of just testing (params == NULL).This
detects the case wherean OSSL_PARAM array contains just a terminator
entry.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Tom Cosgrove <tom.cosgrove@arm.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/25499)
Document the fact that we now require unwrappedlen/wrappedlen to be set
to the size of the unwrapped/wrapped buffers
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/25522)
Outlen was never validated in this function prior to use, nor is it set
to the decrypted value on sucess. Add both of those operations
Fixes#25509
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/25522)
Signed-off-by: lan1120 <lanming@huawei.com>
Reviewed-by: Hugo Landau <hlandau@devever.net>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23094)
This keeps the code consistent with the changes done for other
algorithms that support sigalg_set_ctx_params().
set_ctx_params() should always return 1 if the parameter is unknown.
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/25570)
Self tests no longer use the EVP_DigestSign/Verify API's.
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/25570)
After rudimentary analysis, it appears the below functions can
potentially produce output, whilst the provider is in error state.
These functions were detected using this method:
```
CFLAGS='-save-temps' ./Configure enable-fips --debug
make -j10
find . -name '*.i' | xargs git add -f
git grep --cached -p ossl_prov_is_running | grep libfips-lib > ossl_prov_is_running.txt
git grep --cached -p 'return' | grep libfips-lib > return.txt
grep '\.i=' return.txt > func-with_return.txt
grep '\.i=' ossl_prov_is_running.txt > func-with-ossl_prov_is_running.txt
grep --fixed-strings --line-regexp --file=func-with-ossl_prov_is_running.txt return.txt > func-without-ossl_prov_is_running.txt
grep -e newctx -e initctx -e dupctx func-without-ossl_prov_is_running.txt | grep -v ossl_prov_is_running
```
And from there doing manual inspection, as the list was short at that
point.
As in compile with keeping pre-processed source code; and use `git
grep --cached -p` to find these preprocessed files, and scan for calls
to return or opssl_prov_is_running, with function name printed. And
then exclude one from the other, to hopefully get a list of all the
functions that do not check for ossl_prov_is_running.
As number of functions without "func-without-ossl_prov_is_running"
check is large, I do wonder which other functions are "interesting" to
check for. I think I'm not scanning for _update functions
correctly. Any tips on improving above analysis will help with
maintaining such checks going forward.
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
Reviewed-by: Tom Cosgrove <tom.cosgrove@arm.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/25580)
In this function the salt can be either a zero buffer of exactly mdlen
length, or an arbitrary salt of prevsecretlen length.
Although in practice OpenSSL will always pass in a salt of mdlen size
bytes in the current TLS 1.3 code, the openssl kdf command can pass in
arbitrary values (I did it for testing), and a future change in the
higher layer code could also result in unmatched lengths.
If prevsecretlen is > mdlen this will cause incorrect salt expansion, if
prevsecretlen < mdlen this could cause a crash or reading random
information. Inboth case the generated output would be incorrect.
Signed-off-by: Simo Sorce <simo@redhat.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/25579)
The code was not detecting that the cofactor was set up correctly
if OSSL_PKEY_PARAM_USE_COFACTOR_ECDH was set, resulting in an incorrect
FIPS indicator error being triggered.
Added a test for all possible combinations of a EVP_PKEY setting
OSSL_PKEY_PARAM_USE_COFACTOR_ECDH and the derive context setting
OSSL_EXCHANGE_PARAM_EC_ECDH_COFACTOR_MODE.
This only affects the B & K curves (which have a cofactor that is not 1).
Bug reported by @abkarcher
Testing this properly, also detected a memory leak of privk when the
FIPS indicator error was triggered (in the case where mode = 0 and
use_cofactor was 1).
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/25548)
Similar to other KDFs, the input key should be 112 bits long.
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/25529)
See Section 5 Key Agreement Using Diffie-Hellman and MQV of
[NIST SP 800-131Ar2](https://nvlpubs.nist.gov/nistpubs/SpecialPublications/NIST.SP.800-131Ar2.pdf).
Strengths less than 112bits is disallowed, thus eliminating SHA1.
Skip cms test case that requires use of SHA1 with X9.42 DH.
Rename ossl_fips_ind_digest_check to ossl_fips_ind_digest_exch_check
Add myself to Changes for fips indicator work
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/25517)
For FIPS 140-3 the continuous tests specified in SP 800-90B need to be
included on the output of any entropy source.
They are implemented here as a replacement for the primary DRBG in the FIPS
provider. This results in a setup that looks like this:
+-------------+
| |
| Seed Source |
| |
+------+------+
|
|
v
+-------------+
| |
| CRNG Test |
| |
++----------+-+
| |
| |
v v
+--------------+ +--------------+
| | | |
| Public DRBG | | Private DRBG |
| | | |
+--------------+ +--------------+
An additional benefit, that of avoiding DRBG chains, is also gained.
The current standards do not permit the output of one DRBG to be used
as the input for a second (i.e. a chain).
This also leaves open the future possibility of incorporating a seed
source inside the FIPS boundary.
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/25415)
Related to #8331
Addressing found issues by adding specific error messages to improve
feedback when tag length checks fail for the `EVP_CTRL_AEAD_SET_TAG`
parameter in the AES-OCB algorithm.
- Added PROV_R_INVALID_TAG_LENGTH error to indicate when the current tag
length exceeds the maximum tag length of the algorithm.
- Added `PROV_R_INVALID_TAG_LENGTH` error to indicate when the current tag
length in the context does not match a custom tag length provided as
a parameter.
- Added `ERR_R_PASSED_INVALID_ARGUMENT` error to handle cases where an
invalid pointer is passed in encryption mode.
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/25425)
The CPACF instruction KM provides support for accelerating the full
AES-XTS algorithm on newer machines for AES_XTS_128 and AES_XTS_256.
Preliminary measurements showed performance improvements of up to 50%,
dependent on the message size.
Signed-off-by: Holger Dengler <dengler@linux.ibm.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/25414)
We just allow all possible settables all the time. Some things like the
digest name can't actually be changed in some circumstances - but we already
have checks for those things. It's still possible to pass a digest of the
same name to one that's already been set for example.
Fixes#25012
Reviewed-by: Dmitry Belyavskiy <beldmit@gmail.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/25057)
We need a digest for the none when doing deterministic ECDSA. Give a
better error message if one hasn't been supplied.
See openssl/openssl#25012
Reviewed-by: Dmitry Belyavskiy <beldmit@gmail.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/25057)
On the first squeeze call, when finishing the absorb process, also set
the NIP flag, if we are still in XOF_STATE_INIT state. When MSA 12 is
available, the state buffer A has not been zeroed during initialization,
thus we must also pass the NIP flag here. This situation can happen
when a squeeze is performed without a preceding absorb (i.e. a SHAKE
of the empty message).
Add a test that performs a squeeze without a preceding absorb and check
if the result is correct.
Fixes: 25f5d7b85f
Signed-off-by: Ingo Franzki <ifranzki@linux.ibm.com>
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/25388)
If the data to absorb is less than a block, then the KIMD instruction is
called with zero bytes. This is superfluous, and causes incorrect hash
output later on if this is the very first absorb call, i.e. when the
xof_state is still XOF_STATE_INIT and MSA 12 is available. In this case
the NIP flag is set in the function code for KIMD, but KIMD ignores the
NIP flag when it is called with zero bytes to process.
Skip any KIMD calls for zero length data. Also do not set the xof_state
to XOF_STATE_ABSORB until the first call to KIMD with data. That way,
the next KIMD (with non-zero length data) or KLMD call will get the NIP
flag set and will then honor it to produce correct output.
Fixes: 25f5d7b85f
Signed-off-by: Ingo Franzki <ifranzki@linux.ibm.com>
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/25388)
Correctly display the number of requested threads and the number
of available threads.
CLA: trivial
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/25375)
In case of prehash or prehash-by-caller is set skip the s390x specific
acceleration an fallback to the non-accelerated code path.
Fixes: 6696682774
Signed-off-by: Ingo Franzki <ifranzki@linux.ibm.com>
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/25351)
We checked using 'md_nid < 0', which is faulty.
Impact: DSA and ECDSA signature provider implementations
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/24992)
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/24992)
(in the code, "sigalg" is used to refer to these composite algorithms,
which is a nod to libcrypto and libssl, where that term is commonly used
for composite algorithms)
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/24992)
On newer machines the SHA3/SHAKE performance of CPACF instructions KIMD and KLMD
can be enhanced by using additional modifier bits. This allows the application
to omit initializing the ICV, but also affects the internal processing of the
instructions. Performance is mostly gained when processing short messages.
The new CPACF feature is backwards compatible with older machines, i.e. the new
modifier bits are ignored on older machines. However, to save the ICV
initialization, the application must detect the MSA level and omit the ICV
initialization only if this feature is supported.
Signed-off-by: Joerg Schmidbauer <jschmidb@de.ibm.com>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/25235)
In this mode, only the ph instances are supported, and must be set
explicitly through a parameter. The caller is assumed to pass a
prehash to EVP_PKEY_{sign,verify}().
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/24975)
Add EVP_PKEY_{sign,verify}_message support for our Ed25519 and Ed448
implementations, including ph and ctx variants.
Tests are added with test_evp stanzas.
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/24975)
Added sm2 testcases to endecode_test.c.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/25266)
For SHAKE algorithms we now return 0 from EVP_MD_size().
So all the places that check for < 0 needed to change to <= 0
(Otherwise the behaviour will be to digest nothing in most cases).
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/25285)
Added the function EVP_MD_CTX_get_size_ex() which checks for XOF and
does a ctx get rather than just returning EVP_MD_size().
SHAKE did not have a get_ctx_params() so that had to be added to return the xoflen.
Added a helper function EVP_MD_xof()
EVP_MD_CTX_size() was just an aliased macro for EVP_MD_size(), so to
keep it the same I added an extra function.
EVP_MD_size() always returns 0 for SHAKE now, since it caches the value
of md_size at the time of an EVP_MD_fetch(). This is probably better
than returning the incorrect initial value it was before e.g (16 for
SHAKE128) and returning tht always instead of the set xoflen.
Note BLAKE2B uses "size" instead of "xoflen" to do a similar thing.
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/25285)
This issue has been discovered by osss-fuzzer [1]. The test function decodes
RSA key created by fuzzer and calls EVP_PKEY_pairwise_check() which
proceeds to ossl_bn_miller_rabin_is_prime() check which takes too long
exceeding timeout (45secs).
The idea is to fix OSSL_DECODER_from_data() code path so invalid
RSA keys will be refused.
[1] https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=69134
Test case generated by the fuzzer is added.
Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/25190)
The options in fipsprov.c are now generated using macros with fips_indicator_params.inc.
This should keep the naming consistent.
Some FIPS related headers have moved to providers/fips/include so that
they can use fips_indicator_params.inc.
securitycheck.h now includes fipsindicator.h, and fipsindicator.h includes
fipscommon.h.
fipsinstall.c uses OSSL_PROV_PARAM_ for the configurable FIPS options rather than
using OSSL_PROV_FIPS_PARAM_* as this was confusing as to which one should be used.
fips_names.h just uses aliases now for existing public names.
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/25162)
X25519 and X448 are unapproved in FIPS 140-3
So always trigger the indicator callback if these Keys are used,
and add "fips-indicator" getters that return 0.
This has been added to keygen and key exchange.
(KEM will also require it if ever becomes a FIPS algorithm).
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/25246)
Cleanup + remove a few tests that are not required.
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/25222)
SSKDF KMAC tests added.
Added FIPS indicator tests for SSKDF Hash, HMAC, and KMAC cases.
Added short salt length tests for SSKDF HMAC and KMAC.
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/25049)
