The overall function is still not constant-time, but it just got a lot
less leaky.
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
The path using builtin should be OK, as it should be using dedicated CPU
instructions which are constant time.
This fixes the no-builing path.
GCC gained support for __has_builtin in version 10. We're still testing
with older GCC on the CI, so the non-builtin path is tested on the CI.
https://gcc.gnu.org/gcc-10/changes.html
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
A == N (as pointers) will not happen in pratice: in our context, it
would mean we know at compile time that A == N (as values), and we
wouldn't be calling this function if we knew that already.
N == 1 when I != NULL is also not going to happen: we don't care about
operations mod 1.
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
This function has specific code to handle carries and it's not clear how
to exercises that code through the modinv function, so well, that's what
unit tests are for.
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
This is consistent with the general rules documented at the top of the
file:
- when computing GCD(A, N), there is no modular arithmetic, so the
output can alias any of the inputs;
- when computing a modular inverse, N is the modulus, so it can't be
aliased by any of the outputs (we'll use it for modular operations
over the entire course of the function's execution).
But since this function has two modes of operations with different
aliasing rules (G can alias N only if I == NULL), I think it should
really be stated explicitly.
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
This is a direct translation of sict_mi2() from
https://github.com/mpg/cryptohack/blob/main/ct-pres.py
which was presented in the book club's special session.
This commit only includes two test cases which is very little. Most of
the test cases will be generated by Python modules that belong to the
framework. However we can't have the framework generate those before we
have the corresponding test function in the consuming branches. So,
extended tests are coming as a 2nd step, after the test function has
been merged.
(The test cases in .misc should stay, as they can be convenient when
working on the test function.)
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
Fix a race condition in `mbedtls_aes_ni_has_support()` with some compilers.
A compiler could hoist the assignment `done = 1` above the assignment to `c`,
in which case if two threads call `mbedtls_aes_ni_has_support()` at almost
the same time, they could be interleaved as follows:
Initially: done = 0, c = 0
thread A thread B
if (!done)
done = 1; # hoisted
if (!done)
return c & what; # wrong!
c = cpuid();
return c & what
This would lead to thread B using software AES even though AESNI was
available. This is a very minor performance bug. But also, given a very
powerful adversary who can block thread A indefinitely (which may be
possible when attacking an SGX enclave), thread B could use software AES for
a long time, opening the way to a timing side channel attack.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
Correct base64 input (excluding ignored characters such as spaces) consists
of exactly 4*k, 4*k-1 or 4*k-2 digits, followed by 0, 1 or 2 equal signs
respectively.
Previously, any number of trailing equal signs up to 2 was accepted, but if
there fewer than 4*k digits-or-equals, the last partial block was counted in
`*olen` in buffer-too-small mode, but was not output despite returning 0.
Now `mbedtls_base64_decode()` insists on correct padding. This is
backward-compatible since the only plausible useful inputs that used to be
accepted were inputs with 4*k-1 or 4*k-2 digits and no trailing equal signs,
and those led to invalid (truncated) output. Furthermore the function now
always reports the exact output size in buffer-too-small mode.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
