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>
* Reworked the cipher context once again to be more robustly defined
* Removed redundant memset
* Unified behaviour on failure between driver and software in cipher_finish
* Cipher test driver setup function now also returns early when its status
is overridden, like the other test driver functions
* Removed redundant test cases
* Added bad-order checking to verify the driver doesn't get called where
the spec says it won't.
Signed-off-by: Steven Cooreman <steven.cooreman@silabs.com>
As pointed out by Ronald. The key slot is populated using
get_key_from_slot, and after calling the driver the slot is
validated to not contain an external key, so calling
get_transparent_key is superfluous.
Signed-off-by: Steven Cooreman <steven.cooreman@silabs.com>
Added zeroization of the wrapper context on failure/abort, and reliance on
the crypto core to not call an uninitialised wrapper.
Signed-off-by: Steven Cooreman <steven.cooreman@silabs.com>
Once an operation has been 'accepted' by a driver, the remainder is bound
to the same driver, since driver-specific context structs cannot be shared.
This provides a pretty good gate mechanism for the fallback logic, too.
Signed-off-by: Steven Cooreman <steven.cooreman@silabs.com>
ssl_tls1_3_keys.c exports a structure containing all labels used
in the TLS 1.3 key schedule, but the TLS 1.3 key scheduling unit
tests so far replicated those labels in the test file. In particular,
wrong label values in ssl_tls1_3_keys.c wouldn't have been caught
by the unit tests.
This commit modifies the TLS 1.3 key schedule unit tests to use
the TLS 1.3 labels as exported by ssl_tls1_3_keys.c. This not only
makes sure that those labels are correct, but also avoids hardcoding
their hex-encoding in the test file.
Signed-off-by: Hanno Becker <hanno.becker@arm.com>
Needed to make additional fixes so that when MBEDTLS_USE_PSA_CRYPTO
is defined, the depends-hashes test will succeed. There are two
versions of the ecdsa_verify_wrap() function, one with
MBEDTLS_USE_PSA_CRYPTO and when when it is not enabled. The non PSA
version is not using the md_alg parameter since it is not required.
The PSA version was using that parameter to derive a different value
it needed for PSA_ALG_ECDSA. The arguement of PSA_ALG_ECDSA is
ignored for psa_sign_hash and psa_verify_hash. It is present because
it is used and must be a valid hash, not zero, for psa_sign_hash
(but not psa_verify_hash) with PSA_ALG_DETERMINISTIC_ECDSA, and it is
needed for psa_sign_message and psa_verify_message which are not
implemented yet. The local parameter now uses PSA_ALG_ECDSA_ANY for
the verify function to avoid using the md_alg parameter and avoids
returning incorrect error codes.
Fixes#3587
Signed-off-by: John Durkop <john.durkop@fermatsoftware.com>
This is in line with how the entries of the TLS 1.3 label
structure `mbedtls_ssl_tls1_3_labels` are initialized.
Signed-off-by: Hanno Becker <hanno.becker@arm.com>
`common.h` takes care of the logic of chosing the correct
configuration file, so we don't need to replicate it in
each source file.
Signed-off-by: Hanno Becker <hanno.becker@arm.com>
This commit introduces a new file library/ssl_tls13_key.c
which will subsequently be populated with functionality relating
to the TLS 1.3 key schedule.
Those functions are expected to be internal and are documented
in the internal header library/ssl_tls13_keys.h.
The first function to be implemented is the key expansion
function `HKDF-Expand-Label`. See the documentation in
library/ssl_tls13_keys.h for more information.
Signed-off-by: Hanno Becker <hanno.becker@arm.com>
Trying to compile in the PSA accelerator test driver under MBEDTLS_TEST_HOOKS
turned out to be awkward regarding existing builds. We'll put it under a
custom (not in config.h) define instead, since it's something that only
should happen in test.
Signed-off-by: Steven Cooreman <steven.cooreman@silabs.com>
This is a more natural place for handling the drivers belonging to the
'previous' SE driver spec. It makes for a cleaner psa_crypto.c, and
potentially an easier overview of how to migrate from the old SE driver
interface to the 'opaque accelerator' interface.
Signed-off-by: Steven Cooreman <steven.cooreman@silabs.com>
Use macros instead of accessing directly the key identifier
fields for coding consistency and ease maintenance.
Signed-off-by: Ronald Cron <ronald.cron@arm.com>
Fix PSA code and unit tests for the unit tests
to pass with key identifiers encoding owner
identifiers.
The changes in PSA code just make the enablement
of key identifiers encoding owner identifiers
platform independent. Previous to this commit,
such key identifiers were used only in the case
of PSA SPM platforms.
Signed-off-by: Ronald Cron <ronald.cron@arm.com>
With PSA crypto v1.0.0, a volatile key identifier may
contain a owner identifier but no file is associated
to it. Thus rename the type psa_key_file_id_t to
mbedtls_svc_key_id_t to avoid a direct link with a
file when a key identifier involves an owner
identifier.
The new type name is prefixed by mbedtls to highlight
that the type is specific to Mbed TLS implementation
and not defined in the PSA Cryptography API
specification.
The svc in the type name stands for service as this
is the key identifier type from the point of view of
the service providing the Cryptography services.
The service can be completely provided by the present
library or partially in case of a multi-client service.
As a consequence rename as well:
. MBEDTLS_PSA_CRYPTO_KEY_FILE_ID_ENCODES_OWNER to
MBEDTLS_PSA_CRYPTO_KEY_ID_ENCODES_OWNER
. PSA_KEY_ID_INIT to MBEDTLS_SVC_KEY_ID_INIT
. PSA_KEY_FILE_GET_KEY_ID to MBEDTLS_SVC_KEY_ID_GET_KEY_ID
. psa_key_file_id_make to mbedtls_svc_key_id_make
Signed-off-by: Ronald Cron <ronald.cron@arm.com>
Define always psa_key_id_t as defined in the PSA
Cryptography API specification independently of
whether the MBEDTLS_PSA_CRYPTO_KEY_FILE_ID_ENCODES_OWNER
configuration file is set or not.
As a consequence, get rid of `psa_app_key_id_t` that is
not needed anymore.
Signed-off-by: Ronald Cron <ronald.cron@arm.com>
The purpose of this commit and the following is for
psa_key_id_t to always be as defined by the PSA
Cryptography API specification.
Currently psa_key_id_t departs from its specification
definition when MBEDTLS_PSA_CRYPTO_KEY_FILE_ID_ENCODES_OWNER
configuration flag is set. In that configuration, it is set
to be equal to psa_key_file_id_t which in that configuration
encodes an owner identifier along the key identifier.
Type psa_key_file_id_t was meant to be the key identifier type
used throughout the library code. If
MBEDTLS_PSA_CRYPTO_KEY_FILE_ID_ENCODES_OWNER is set it
includes both a key and owner identifier, otherwise it is
equal to psa_key_id_t.
It has not been the key identifier type throughout the
library so far because when the PSA Cryptography
specification was developped the library Doxygen
documentation was used to generate the PSA Cryptography API
specification thus the need to use psa_key_id_t and not
psa_key_file_id_t.
As this constraint does not hold anymore, move
to psa_key_file_id_t as the key identifier type throughout
the library code.
By the way, this commit updates the key identifier
initialization in the tests to be compatible with a
composit key identifier. A psa_key_id_make()
inline function is introduced to initialize key
identifiers (composit ot not) at runtime.
Signed-off-by: Ronald Cron <ronald.cron@arm.com>
Minor updates to changelog for more concise wording and fixed styling
in other files as needed.
Signed-off-by: John Durkop <john.durkop@fermatsoftware.com>
Add guards in pk_wrap.c to ensure if ECDSA is not defined, errors
are returned.
Remove warnings in pk.c for unused variables.
Add new test (test_depends_pkalgs_psa) to all.sh to confirm
when USE_PSA_CRYPTO is defined that features are working properly.
Fix#3294
Signed-off-by: John Durkop <john.durkop@fermatsoftware.com>
Probably the `W[2 << MBEDTLS_MPI_WINDOW_SIZE]` notation is based on a transcription of 2**MBEDTLS_MPI_WINDOW_SIZE.
Signed-off-by: Daniel Otte <d.otte@wut.de>
Wrap multi-line string literals in parentheses
to prevent a Clang 12 -Wstring-concatenation warning
(activated by -Wall), which caused the build to fail.
Fixes https://github.com/ARMmbed/mbedtls/issues/3586
Signed-off-by: Guido Vranken <guidovranken@gmail.com>
Update copyright notices to newly added files since merge of original
PR #3546 "Update copyright notices to use Linux Foundation guidance".
Generated using the same script.
Signed-off-by: Dan Handley <dan.handley@arm.com>
* development:
Update copyright notices to use Linux Foundation guidance
Undef ASSERT before defining it to ensure that no previous definition has sneaked in through included files.
Add ChangeLog entry for X.509 CN-type vulnerability
Improve documentation of cn in x509_crt_verify()
Fix comparison between different name types
Add test: DNS names should not match IP addresses
Remove obsolete buildbot reference in compat.sh
Fix misuse of printf in shell script
Fix added proxy command when IPv6 is used
Simplify test syntax
Fix logic error in setting client port
ssl-opt.sh: include test name in log files
ssl-opt.sh: remove old buildbot-specific condition
ssl-opt.sh: add proxy to all DTLS tests
Signed-off-by: Dan Handley <dan.handley@arm.com>
Currently the new component in all.sh fails because
mbedtls_ssl_cf_memcpy_offset() is not actually constant flow - this is on
purpose to be able to verify that the new test works.
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
mbedtls_md_setup() allocates a hash-specific context and then, if
requested, an extra HMAC context. If the second allocation failed, the
hash context was not freed.
Fix this by ensuring that the mbedtls_md_context_t object is always in
a consistent state, in particular, that the md_info field is always
set. For robustness, ensure that the object is in a consistent state
even on errors (other than BAD_INPUT_DATA if the object was not in a
consistent state on entry).
Fix#3486
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
The tests are supposed to be failing now (in all.sh component
test_memsan_constant_flow), but they don't as apparently MemSan doesn't
complain when the src argument of memcpy() is uninitialized, see
https://github.com/google/sanitizers/issues/1296
The next commit will add an option to test constant flow with valgrind, which
will hopefully correctly flag the current non-constant-flow implementation.
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
This paves the way for a constant-flow implementation of HMAC checking, by
making sure that the comparison happens at a constant address. The missing
step is obviously to copy the HMAC from the secret offset to this temporary
buffer with constant flow, which will be done in the next few commits.
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
As a result, the copyright of contributors other than Arm is now
acknowledged, and the years of publishing are no longer tracked in the
source files.
Also remove the now-redundant lines declaring that the files are part of
MbedTLS.
This commit was generated using the following script:
# ========================
#!/bin/sh
# Find files
find '(' -path './.git' -o -path './3rdparty' ')' -prune -o -type f -print | xargs sed -bi '
# Replace copyright attribution line
s/Copyright.*Arm.*/Copyright The Mbed TLS Contributors/I
# Remove redundant declaration and the preceding line
$!N
/This file is part of Mbed TLS/Id
P
D
'
# ========================
Signed-off-by: Bence Szépkúti <bence.szepkuti@arm.com>