Instead of generating blinding values in a not-quite-uniform way
(https://github.com/ARMmbed/mbedtls/issues/4245) with copy-pasted
code, use mbedtls_mpi_random().
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
Since mbedtls_mpi_random() is not specific to ECC code, move it from
the ECP module to the bignum module.
This increases the code size in builds without short Weierstrass
curves (including builds without ECC at all) that do not optimize out
unused functions.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
Rename mbedtls_ecp_gen_privkey_sw to mbedtls_mpi_random since it has
no particular connection to elliptic curves beyond the fact that its
operation is defined by the deterministic ECDSA specification. This is
a generic function that generates a random MPI between 1 inclusive and
N exclusive.
Slightly generalize the function to accept a different lower bound,
which adds a negligible amount of complexity.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
Add unit tests for private key generation on short Weierstrass curves.
These tests validate that the result is within the desired range.
Additionally, they validate that after performing many iterations, the
range is covered to an acceptable extent: for tiny ranges, all values
must be reached; for larger ranges, all value bits must reach both 0
and 1.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
mbedtls_ecp_gen_privkey_mx generates a random number with a certain
top bit set. Depending on the size, it would either generate a number
with that top bit being random, then forcibly set the top bit to
1 (when high_bit is not a multiple of 8); or generate a number with
that top bit being 0, then set the top bit to 1 (when high_bit is a
multiple of 8). Change it to always generate the top bit randomly
first.
This doesn't make any difference in practice: the probability
distribution is the same either way, and no supported or plausible
curve has a size of the form 8n+1 anyway. But it slightly simplifies
reasoning about the behavior of this function.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
Don't calculate the bit-size of the initially generated random number.
This is not necessary to reach the desired distribution of private
keys, and creates a (tiny) side channel opportunity.
This changes the way the result is derived from the random number, but
does not affect the resulting distribution.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
The library rejected an RNG input of all-bits-zero, which led to the
key 2^{254} (for Curve25519) having a 31/32 chance of being generated
compared to other keys. This had no practical impact because the
probability of non-compliance was 2^{-256}, but needlessly
complicated the code.
The exception was added in 98e28a74e3 to
avoid the case where b - 1 wraps because b is 0. Instead, change the
comparison code to avoid calculating b - 1.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
Test the exact output from known RNG input. This is overly
constraining, but ensures that the code has good properties.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
If a fallback is not explicitly configured in the
mbedtls_test_rnd_buf_info structure, fail after the buffer is
exhausted.
There is no intended behavior change in this commit: all existing uses
of mbedtls_test_rnd_buffer_rand() have been updated to set
mbedtls_test_rnd_std_rand as the fallback.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
For Montgomery keys, n_bits is actually the position of the highest
bit and not the number of bits, which would be 1 more (fence vs
posts). Rename the variable accordingly to lessen the confusion.
No semantic change.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
Put the Montgomery and short Weierstrass implementations of
mbedtls_ecp_gen_privkey into their own function which can be tested
independently, but will not be part of the public ABI/API.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
When mbedtls_nist_kw_wrap was called with output=NULL and out_size=0, it
performed arithmetic on the null pointer before detecting that the output
buffer is too small and returning an error code. This was unlikely to have
consequences on real-world hardware today, but it is undefined behavior and
UBSan with Clang 10 flagged it. So fix it (fix#4025).
Fix a similar-looking pattern in unwrap, though I haven't verified that it's
reachable there.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
PSA_KEY_LIFETIME_FROM_PERSISTENCE_AND_LOCATION shifts the location
value (location << 8). This can go outside the range of a 32-bit int
if the location value is above 2^23 as is the case here. Asan
rightfully complained about it. Make the value unsigned to avoid the
integer overflow.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
Test keys with various persistence levels, enumerated from the
metadata tests.
For read-only keys, do not attempt to create or destroy the key
through the API, only to read a key that has been injected into
storage directly through filesystem access.
Do not test keys with a non-default location, since they require a
driver and we do not yet have a dependency mechanism to require the
presence of a driver for a specific location value.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
The persistence level PSA_KEY_PERSISTENCE_READ_ONLY can now only be used
as intended, for keys that cannot be modified through normal use of the API.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
Attempting to create a key with an invalid location or with read-only
persistence must be rejected.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
mbedtls_psa_get_stats() was written back before lifetimes were
structured as persistence and location. Fix its classification of
volatile external keys and internal keys with a non-default
persistence.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
The call to `tests/scripts/generate_psa_tests.py` added by the commit
"generate_psa_tests.py: allow generating each file independently"
assumed that the `check` function supports multiple file names, but in
fact it does not do so. When `generate_psa_tests.py` started generating
more than one file, `check-generated-files.sh` did not detect changes to
files other than the first one listed by `generate_psa_tests.py --list`.
Fix this: change `check` to support either a single directory (with
detection of added/removed files) or a list of files (which is assumed
to be static).
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
When generating expressions to construct test case data, there can be
duplicate values, for example if a value of the form C(A) is present
as such in test_suite_psa_crypto_metadata.data and also constructed by
enumerating the argument A for the constructor C. Eliminate such
duplicates in generate_expressions.
This commit removes many test cases that were exact duplicates (and
were near-duplicates differing only in whitespace before the
whitespace normalization).
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
Define the dependency symbols PSA_WANT_ALG_ECDSA_ANY and
PSA_WANT_ALG_RSA_PKCS1V15_SIGN_RAW as de facto synonyms of
PSA_WANT_ALG_ECDSA and PSA_WANT_ALG_RSA_PKCS1V15_SIGN respectively: if
either one is requested, the other is set.
This makes it easier to systematically determine the dependencies of
an algorithm.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
In macro_collector.py, base InputsForTest on PSAMacroEnumerator rather
than PSAMacroCollector. It didn't make much sense to use
PSAMacroCollector anymore since InputsForTest didn't use anything
other than the constructor.
psa_generate_tests now generates arguments for more macros.
In particular, it now collects macro arguments from
test_suite_psa_crypto_metadata. Algorithms with parameters are now
supported.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
First build a list of all keys, then construct all the corresponding
test cases. This allows all required information to be obtained in
one go, which is a significant performance gain as the information
includes numerical values obtained by compiling a C program.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
In generate_psa_tests, use InputsForTest rather than PSAMacroCollector
to gather values. This way, the enumeration of values to test includes
values used in metadata tests in addition to constructors parsed from
header files. This allows greater coverage of values built from
constructors with arguments. This doesn't make a difference yet, but
it will once algorithm constructors with arguments are supported in
generate_psa_tests.
Make the injection of numerical values optional. They are useful for
test_psa_constant_names, so keep them there. Don't use them for
not-supported tests: they might make sense, but the current code
wouldn't work since it doesn't know how to make up fake key material
or what dependencies to generate. Don't use them for storage tests:
they only make sense for supported values.
Don't inject 'PSA_SUCCESS': that's superfluous.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
