If MBEDTLS_KEY_EXCHANGE_SOME_PSK_ENABLED is defined, then the return value will be overridden by the extra code running after the removed return instruction.
Signed-off-by: gabor-mezei-arm <gabor.mezei@arm.com>
The extra code running after the removed return instruction should not generate any output. Only the read config value must be printed.
Signed-off-by: gabor-mezei-arm <gabor.mezei@arm.com>
This reverts commit 9c46a60e6c.
When the library is dynamically linked against Glibc (as is usually
the case with Glibc), it now requires a recent Glibc at runtime if it
was compiled with a recent Glibc. This is a loss of functionality for
no demonstrated benefit.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
“Prior to Mbed TLS 2.24” suggests that 2.24 itself didn't use the old
policy anymore, but it did. Change to “Until”, and also give the exact
version number “2.24.0”.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
... as opposed to PSA_ERROR_BAD_STATE.
The spec on psa_cipher_finish() states that PSA_ERROR_INVALID_ARGUMENT
should be returned when:
"The total input size passed to this operation is not valid for this
particular algorithm. For example, the algorithm is a based on block
cipher and requires a whole number of blocks, but the total input size
is not a multiple of the block size."
Currently, there is a distinction between encryption and decryption
on whether INVALID_ARGUMENT or BAD_STATE is returned, but this is not
a part of the spec.
This fix ensures that PSA_ERROR_INVALID_ARGUMENT is returned
consistently on invalid cipher input sizes.
Signed-off-by: Fredrik Strupe <fredrik.strupe@silabs.com>
From now on, external contributions are no longer acknowledged in the
changelog file. They of course remain acknowledged in the Git history.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
Starting with commit 49e94e3, the do/while loop in
`rsa_prepare_blinding()` was changed to a `do...while(0)`, which
prevents retry from being effective and leaves dead code.
Restore the while condition to retry, and lift the calls to finish the
computation out of the while loop by by observing that they are
performed only when `mbedtls_mpi_inv_mod()` returns zero.
Signed-off-by: Peter Kolbus <peter.kolbus@garmin.com>
Python should not be required for the build when the no_test target is
used. This commit adds the generated file to the source tree and the
check-generated-files script, and removes the generation from (c)make.
Fixes#3524
Signed-off-by: Cameron Nemo <cnemo@tutanota.com>
The toplevel directory is actually just ../..: the makefile commands
are executed in the subdirectory. $(PWD) earlier was wrong because it
comes from the shell, not from make. Looking up $(MAKEFILE_LIST) is
wrong because it indicates where the makefile is (make -f), not which
directory to work in (make -C).
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
According to https://www.bearssl.org/ctmul.html even single-precision
multiplication is not constant-time on some older platforms.
An added benefit of the new code is that it removes the somewhat mysterious
constant 0x1ff - which was selected because at that point the maximum value of
padlen was 256. The new code is perhaps a bit more readable for that reason.
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
The previous code used comparison operators >= and == that are quite likely to
be compiled to branches by some compilers on some architectures (with some
optimisation levels).
For example, take the following function:
void old_update( size_t data_len, size_t *padlen )
{
*padlen *= ( data_len >= *padlen + 1 );
}
With Clang 3.8, let's compile it for the Arm v6-M architecture:
% clang --target=arm-none-eabi -march=armv6-m -Os foo.c -S -o - |
sed -n '/^old_update:$/,/\.size/p'
old_update:
.fnstart
@ BB#0:
.save {r4, lr}
push {r4, lr}
ldr r2, [r1]
adds r4, r2, #1
movs r3, #0
cmp r4, r0
bls .LBB0_2
@ BB#1:
mov r2, r3
.LBB0_2:
str r2, [r1]
pop {r4, pc}
.Lfunc_end0:
.size old_update, .Lfunc_end0-old_update
We can see an unbalanced secret-dependant branch, resulting in a total
execution time depends on the value of the secret (here padlen) in a
straightforward way.
The new version, based on bit operations, doesn't have this issue:
new_update:
.fnstart
@ BB#0:
ldr r2, [r1]
subs r0, r0, #1
subs r0, r0, r2
asrs r0, r0, #31
bics r2, r0
str r2, [r1]
bx lr
.Lfunc_end1:
.size new_update, .Lfunc_end1-new_update
(As a bonus, it's smaller and uses less stack.)
While there's no formal guarantee that the version based on bit operations in
C won't be translated using branches by the compiler, experiments tend to show
that's the case [1], and it is commonly accepted knowledge in the practical
crypto community that if we want to sick to C, bit operations are the safest
bet [2].
[1] https://github.com/mpg/ct/blob/master/results
[2] https://github.com/veorq/cryptocoding
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
Move key identifier related macros and functions from
crypto_types.h to crypto_values.h as the latter is
the intended file to put them in.
Signed-off-by: Ronald Cron <ronald.cron@arm.com>
No obvious reason to not enable owner identifier encoding
in baremetal as multi-client support is expected to be needed
for some embedded platforms. Thus enable it.
Signed-off-by: Ronald Cron <ronald.cron@arm.com>
