Commit graph

16138 commits

Author SHA1 Message Date
gabor-mezei-arm 043192d209 Move mbedtls_cf_mpi_uint_cond_assign function to the constant-time module
Signed-off-by: Gabor Mezei <gabor.mezei@arm.com>
2021-11-11 10:55:56 +01:00
gabor-mezei-arm 5cec8b44a8
Move mbedtls_cf_cond_select_sign function to the constant-time module
Signed-off-by: Gabor Mezei <gabor.mezei@arm.com>
2021-11-11 10:54:20 +01:00
gabor-mezei-arm 7533253125
Move mbedtls_cf_uint_if function to the constant-time module
Signed-off-by: Gabor Mezei <gabor.mezei@arm.com>
2021-11-11 10:52:42 +01:00
gabor-mezei-arm 097d4f555e
Move mbedtls_cf_mpi_uint_lt function to the constant-time module
Signed-off-by: Gabor Mezei <gabor.mezei@arm.com>
2021-11-11 10:52:01 +01:00
gabor-mezei-arm 17da4f2a4e
Remove module dependency
Elinimate macros defined by modules locally in the functions that are
moving to the new constant-time module.

Signed-off-by: Gabor Mezei <gabor.mezei@arm.com>
2021-11-11 10:04:59 +01:00
gabor-mezei-arm 9d7bf09333
Move mbedtls_cf_size_gt function to the constant-time module
Signed-off-by: Gabor Mezei <gabor.mezei@arm.com>
2021-11-11 10:04:49 +01:00
gabor-mezei-arm 96584ddd4b
Move mbedtls_cf_size_bool_eq function to the constant-time module
There were multiple functions called mbedtls_cf_size_bool_eq. They had exactly
the same behavior, so move the one in bignum.c and remove the other.

Signed-off-by: Gabor Mezei <gabor.mezei@arm.com>
2021-11-11 10:03:55 +01:00
gabor-mezei-arm a2bcabceb2
Move mbedtls_cf_size_mask_ge function to the constant-time module
Signed-off-by: Gabor Mezei <gabor.mezei@arm.com>
2021-11-11 10:00:17 +01:00
gabor-mezei-arm 4d6b14624e
Move mbedtls_cf_size_mask_lt function to the constant-time module
Signed-off-by: Gabor Mezei <gabor.mezei@arm.com>
2021-11-11 10:00:17 +01:00
gabor-mezei-arm d361ccd663
Move mbedtls_cf_size_mask function to the constant-time module
Signed-off-by: Gabor Mezei <gabor.mezei@arm.com>
2021-11-11 10:00:17 +01:00
gabor-mezei-arm c11cac9f1b
Move mbedtls_cf_uint_mask function to the constant-time module
Signed-off-by: Gabor Mezei <gabor.mezei@arm.com>
2021-11-11 10:00:17 +01:00
gabor-mezei-arm 944c107744
Move contatnt-time memcmp functions to the contant-time module
Signed-off-by: Gabor Mezei <gabor.mezei@arm.com>
2021-11-11 10:00:12 +01:00
gabor-mezei-arm e41e3e8a8b Rename function to have suitable name
Signed-off-by: Gabor Mezei <gabor.mezei@arm.com>
2021-11-11 09:57:28 +01:00
gabor-mezei-arm 9055972227
Add a new file for constant-time functions
Signed-off-by: Gabor Mezei <gabor.mezei@arm.com>
2021-11-11 09:39:12 +01:00
Gilles Peskine 95c3971c81
Merge pull request #5133 from haampie/fix/DT_NEEDED_for_shared_libraries-2.x
Backport 2.x: DT_NEEDED for shared builds in makefile
2021-11-05 12:04:33 +01:00
Manuel Pégourié-Gonnard 2c4f032bcf
Merge pull request #5050 from gilles-peskine-arm/missing-psa-macros-2.x
Backport 2.x: Add missing PSA macros
2021-11-05 10:09:17 +01:00
Harmen Stoppels 3ed4263ad7 DT_NEEDED for shared builds in makefile
The makefile build specifies -L. -lmbedx509 -lmbedcrypto flags first,
and only then object files referencing symbols from those libraries.

In this order the linker will not add the linked libraries to the
DT_NEEDED section because they are not referenced yet (at least that
happens for me on ubuntu 20.04 with the default gnu compiler tools).

By first specifying the object files and then the linked libraries, we
do end up with libmbedx509 and libmbedcrypto in the DT_NEEDED sections.

This way running dlopen(...) on libmedtls.so just works.

Note that the CMake build does this by default.

Signed-off-by: Harmen Stoppels <harmenstoppels@gmail.com>
2021-11-05 09:31:22 +01:00
paul-elliott-arm 1aa7ad7c0f
Merge pull request #5129 from gilles-peskine-arm/base64_invasive_h-2.x
Backport 2.x: Fix copypasta in #endif comment
2021-11-04 10:06:12 +00:00
Gilles Peskine 16c2102de2 Fix copypasta in #endif comment
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
2021-11-03 18:28:40 +01:00
Gilles Peskine adcfdbf2c6 Fix test bug: some classification flags were not tested
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
2021-11-03 14:29:20 +01:00
Gilles Peskine 31b95155ba Ensure that all flags are actually tested
At least twice, we added a classification flag but forgot to test it in the
relevant test functions. Add some protection so that this doesn't happen
again. In each classification category, put a macro xxx_FLAG_MASK_PLUS_ONE
at the end. In the corresponding test function, keep track of the flags that
are tested, and check that their mask is xxx_FLAG_MASK_PLUS_ONE - 1 which is
all the bits of the previous flags set.

Now, if we add a flag without testing it, the test
TEST_EQUAL( classification_flags_tested, xxx_FLAG_MASK_PLUS_ONE - 1 )
will fail. It will also fail if we make the set of flag numbers
non-consecutive, which is ok.

This reveals that three algorithm flags had been added but not tested (in
two separate occasions). Also, one key type flag that is no longer used by
the library was still defined but not tested, which is not a test gap but is
inconsistent. It's for DSA, which is relevant to the PSA encoding even if
Mbed TLS doesn't implement it, so keep the flag and do test it.

Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
2021-11-03 14:25:41 +01:00
Gilles Peskine e65be27eea Correct block size for MD2
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
2021-11-03 13:19:02 +01:00
Gilles Peskine 19191039f9 Note the change to PSA_ALG_IS_HASH_AND_SIGN in the changelog
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
2021-11-03 13:19:02 +01:00
Gilles Peskine cc14ce08c2 Add PSA_ALG_IS_HASH_AND_SIGN to the metadata tests
The status of signature wildcards with respect to PSA_ALG_IS_HASH_AND_SIGN
is unclear in the specification. A wildcard is usually instantiated with a
specific hash, making the implementation hash-and-sign, but it could also be
instantiated with a non-hash-and-sign algorithm. For the time being, go with
what's currently implemented, which is that they are considered
hash-and-sign.

Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
2021-11-03 13:19:02 +01:00
Gilles Peskine 4bdcf9a35a Reorder macro definitions
Definition before mention

Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
2021-11-03 12:44:08 +01:00
Gilles Peskine 8cb22c8d87 Untangle PSA_ALG_IS_HASH_AND_SIGN and PSA_ALG_IS_SIGN_HASH
The current definition of PSA_ALG_IS_HASH_AND_SIGN includes
PSA_ALG_RSA_PKCS1V15_SIGN_RAW and PSA_ALG_ECDSA_ANY, which don't strictly
follow the hash-and-sign paradigm: the algorithm does not encode a hash
algorithm that is applied prior to the signature step. The definition in
fact encompasses what can be used with psa_sign_hash/psa_verify_hash, so
it's the correct definition for PSA_ALG_IS_SIGN_HASH. Therefore this commit
moves definition of PSA_ALG_IS_HASH_AND_SIGN to PSA_ALG_IS_SIGN_HASH, and
replace the definition of PSA_ALG_IS_HASH_AND_SIGN by a correct one (based
on PSA_ALG_IS_SIGN_HASH, excluding the algorithms where the pre-signature
step isn't to apply the hash encoded in the algorithm).

In the definition of PSA_ALG_SIGN_GET_HASH, keep the condition for a nonzero
output to be PSA_ALG_IS_HASH_AND_SIGN.

Everywhere else in the code base (definition of PSA_ALG_IS_SIGN_MESSAGE, and
every use of PSA_ALG_IS_HASH_AND_SIGN outside of crypto_values.h), we meant
PSA_ALG_IS_SIGN_HASH where we wrote PSA_ALG_IS_HASH_AND_SIGN, so do a
global replacement.
```
git grep -l IS_HASH_AND_SIGN ':!include/psa/crypto_values.h' | xargs perl -i -pe 's/ALG_IS_HASH_AND_SIGN/ALG_IS_SIGN_HASH/g'
```

Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
2021-11-03 12:44:08 +01:00
Gilles Peskine 1b06d09fc6 Test PSA_HASH_BLOCK_LENGTH
Only tested for algorithms for which we support HMAC, since that's all we
use PSA_HASH_BLOCK_LENGTH for at the moment.

Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
2021-11-03 12:38:57 +01:00
Gilles Peskine 285f2133f5 Use the new macro PSA_HASH_BLOCK_LENGTH
Replace an equivalent internal function.

Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
2021-11-03 12:38:57 +01:00
Mateusz Starzyk 21cac07626 Add changelog entry for new PSA Crypto API macros.
Signed-off-by: Mateusz Starzyk <mateusz.starzyk@mobica.com>
2021-11-03 12:38:57 +01:00
Mateusz Starzyk 64010dc544 Add missing PSA_KEY_ID_NULL macro.
Signed-off-by: Mateusz Starzyk <mateusz.starzyk@mobica.com>
2021-11-03 12:38:57 +01:00
Mateusz Starzyk 272a3d4dd8 Add missing PSA_HASH_BLOCK_LENGTH macro.
Signed-off-by: Mateusz Starzyk <mateusz.starzyk@mobica.com>
2021-11-03 12:38:57 +01:00
Mateusz Starzyk 294ca30120 Add missing PSA_ALG_NONE macro.
Signed-off-by: Mateusz Starzyk <mateusz.starzyk@mobica.com>
2021-11-03 12:38:57 +01:00
Mateusz Starzyk d22362c647 Add missing PSA_ALG_IS_SIGN_HASH macro.
Signed-off-by: Mateusz Starzyk <mateusz.starzyk@mobica.com>
2021-11-03 12:38:57 +01:00
Gilles Peskine 9fa1d57d6a
Merge pull request #5125 from AndrzejKurek/add-missing-test-name-backport-2x
Backport 2.x: Add a missing psa_crypto test suite test name
2021-11-03 10:37:33 +01:00
Andrzej Kurek b4206b146d Add a missing psa_crypto test suite test name
Signed-off-by: Andrzej Kurek <andrzej.kurek@arm.com>
2021-11-02 20:06:08 +01:00
Manuel Pégourié-Gonnard 1f1bc2df7c
Merge pull request #5117 from gilles-peskine-arm/psa-rsa-pss_any_salt-2.x
Backport 2.x: PSA: fix salt length for PSS verification
2021-10-29 16:36:46 +02:00
Manuel Pégourié-Gonnard 6295dcb600
Merge pull request #5051 from gilles-peskine-arm/psa-add-aria-2.x
Backport 2.x: Add ARIA to the PSA API
2021-10-29 09:38:13 +02:00
Manuel Pégourié-Gonnard 300c081940
Merge pull request #5046 from gilles-peskine-arm/rm-PSACryptoDriverModelSpec-2.x
Backport 2.x: Remove the old driver model specification draft
2021-10-29 09:36:27 +02:00
Manuel Pégourié-Gonnard 0c81ba7856
Merge pull request #4944 from gilles-peskine-arm/remove-greentea-2.x
Backport 2.x: Remove on-target testing
2021-10-29 09:33:26 +02:00
Manuel Pégourié-Gonnard fa586dbbcf
Merge pull request #5087 from gilles-peskine-arm/test_ssl_o2-2.x
Backport 2.x: Build with -O2 when running ssl-opt
2021-10-29 09:25:32 +02:00
Gilles Peskine ff30bd0111 Always set a build type for cmake when building for testing
Set the build type to Release (-O2) when running CPU-intensive tests (ssl-opt,
or unit tests with debug features). A build type of Check (-Os) would be best
when the main objective of the build is to check for build errors or warnings
and there aren't many tests to run; in this commit there are no such test
cases to change. Only use cmake with no build type (which results in not
passing a -O option, and thus missing some GCC warnings) when exercising cmake
features.

Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
2021-10-28 14:28:53 +02:00
Gilles Peskine 31fdda1262 Fix cmake invocation syntax
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
2021-10-28 14:28:53 +02:00
Gilles Peskine 2531970772 Switch cmake -O2 builds around to where we test a lot
Use Release mode (-O2) for component_test_full_cmake_clang which runs SSL
tests.

To have some coverage with Check mode (which enables more compiler warnings
but compiles with -Os), change a few other builds that only run unit tests
at most to Check mode.

Don't add any new builds, to keep the total build volume down. We don't need
extensive coverage of all combinations, just a reasonable set.

Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
2021-10-28 14:28:53 +02:00
Gilles Peskine a0c51fb53e Clarify a comment
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
2021-10-28 14:28:53 +02:00
Gilles Peskine 3daa83e99a Correct support function name
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
2021-10-28 14:28:53 +02:00
Gilles Peskine 6fa69864a2 Build with -O2 when running ssl-opt
SSL testing benefits from faster executables, so use -O2 rather than -O1.
Some builds use -O1, but that's intended for jobs that only run unit tests,
where the build takes longer than the tests.

Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
2021-10-28 14:28:53 +02:00
Gilles Peskine f8362ca847 Readability improvements
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
2021-10-28 10:07:37 +02:00
Gilles Peskine 44fa40cd56 Fix PSA_ALG_RSA_PSS verification accepting an arbitrary salt length
PSA_ALG_RSA_PSS algorithm now accepts only the same salt length for
verification that it produces when signing, as documented.

Fixes #4946.

Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
2021-10-28 10:07:27 +02:00
Gilles Peskine d79af3a522 Test PSS verification with different salt lengths
Test the following combinations:
* 1024-bit key, SHA-256, salt=0
* 1024-bit key, SHA-256, salt=31 (1 byte shorter than standard)
* 1024-bit key, SHA-256, salt=32 (standard length)
* 1024-bit key, SHA-256, salt=94 (maximum possible length)
* 1024-bit key, SHA-512, salt=61 (1 byte shorter than standard)
* 1024-bit key, SHA-512, salt=62 (standard = maximum possible length)
* 528-bit key, SHA-512, salt=0 (only possible length)

Test psa_verify_hash() for both PSA_ALG_RSA_PSS and PSA_ALG_RSA_PSS_ANY_SALT
with all of these combinations. For psa_verify_message(), just test once
with the standard length and once with a different length.

Note that as of this commit, both PSA_ALG_RSA_PSS and
PSA_ALG_RSA_PSS_ANY_SALT accept any salt length during verification, hence
all the new test cases are positive.

The verify test cases were generated using the Python script below.

```
from Cryptodome import Hash
from Cryptodome.Hash import SHA512
from Cryptodome import PublicKey
from Cryptodome.PublicKey import RSA
from Cryptodome.Signature import pss

key = {
    528: RSA.import_key(bytes.fromhex("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")),
    1024: RSA.import_key(bytes.fromhex("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")),
}
hash_module = {
    256: Hash.SHA256,
    512: Hash.SHA512,
}

def print_test_case(remark, pub, kbits, hbits, input, output):
    key_hex = pub.hex()
    input_hex = input.hex()
    output_hex = output.hex()
    print(f"""\
PSA verify hash: RSA-{kbits} PSS SHA-{hbits}, {remark}
depends_on:PSA_WANT_ALG_RSA_PSS:PSA_WANT_ALG_SHA_{hbits}:PSA_WANT_KEY_TYPE_RSA_PUBLIC_KEY:MBEDTLS_PK_PARSE_C:MBEDTLS_MD_C
verify_hash:PSA_KEY_TYPE_RSA_PUBLIC_KEY:"{key_hex}":PSA_ALG_RSA_PSS(PSA_ALG_SHA_{hbits}):"{input_hex}":"{output_hex}"

PSA verify hash: RSA-{kbits} PSS-any-salt SHA-{hbits}, {remark}
depends_on:PSA_WANT_ALG_RSA_PSS:PSA_WANT_ALG_SHA_{hbits}:PSA_WANT_KEY_TYPE_RSA_PUBLIC_KEY:MBEDTLS_PK_PARSE_C:MBEDTLS_MD_C
verify_hash:PSA_KEY_TYPE_RSA_PUBLIC_KEY:"{key_hex}":PSA_ALG_RSA_PSS_ANY_SALT(PSA_ALG_SHA_{hbits}):"{input_hex}":"{output_hex}"
""")

def rand(n):
    return bytes(x & 0xff for x in range(n))

def test_case(kbits, hbits, slen):
    priv = key[kbits]
    pub_spki = priv.publickey().export_key('DER')
    pub_raw = PublicKey._expand_subject_public_key_info(pub_spki)[1]
    hash_op = hash_module[hbits].new(b'abc')
    digest = hash_op.copy().digest()
    output = pss.new(priv, salt_bytes=slen, rand_func=rand).sign(hash_op)
    print_test_case(f"slen={slen}", pub_raw, kbits, hbits, digest, output)

test_case(1024, 256, 0)
test_case(1024, 256, 31)
test_case(1024, 256, 32)
test_case(1024, 256, 94)
test_case(1024, 512, 61)
test_case(1024, 512, 62)
test_case(528, 512, 0)
```

Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
2021-10-28 10:07:24 +02:00
Gilles Peskine 19ff9a6173 Test that a PSS policy doesn't allow PSS_ANY_SALT and vice versa
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
2021-10-28 10:07:21 +02:00