We had a message in the data file, and were computing its hash in the
test function. It is more efficient (and simpler when it comes to
dependencies) to directly have the message hash in the data file.
It was probably this way because some test vectors provide the message
for the sake of all-in-one implementation that hash-and-sign at once.
But our API gets a hash as the input and signs it. In unit tests, this
should be reflected in the signature of the test function, which should
take a hash as input.
The changes to the .data file were done using the following python
script:
import hashlib
suite = 'pkcs1_v21'
functions = {
'pkcs1_rsassa_pss_sign': (6, 8),
'pkcs1_rsassa_pss_verify': (4, 6),
'pkcs1_rsassa_pss_verify_ext': (4, 8),
}
def hash_ctx(s):
if s == 'MBEDTLS_MD_MD5':
return hashlib.md5()
if s == 'MBEDTLS_MD_SHA1':
return hashlib.sha1()
if s == 'MBEDTLS_MD_SHA224':
return hashlib.sha224()
if s == 'MBEDTLS_MD_SHA256':
return hashlib.sha256()
if s == 'MBEDTLS_MD_SHA384':
return hashlib.sha384()
if s == 'MBEDTLS_MD_SHA512':
return hashlib.sha512()
def fix(l):
parts = l.rstrip().split(":")
fun = parts[0]
if fun not in functions:
return l
(digest_idx, msg_idx) = functions[fun]
alg_str = parts[digest_idx]
if alg_str == "MBEDTLS_MD_NONE":
return l
h = hash_ctx(alg_str)
msg_str = parts[msg_idx]
msg_hex = msg_str[1:-1]
msg = bytes.fromhex(msg_hex)
h.update(msg)
msg_hash = h.hexdigest()
msg_hash_str = '"' + msg_hash + '"'
parts[msg_idx] = msg_hash_str
return ":".join(parts) + '\n'
filename = 'tests/suites/test_suite_' + suite + '.data'
with open(filename) as f:
lines = f.readlines()
lines = [fix(l) for l in lines]
with open(filename, 'w') as f:
f.writelines(lines)
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
We had a message in the data file, and were computing its hash in the
test function. It is more efficient (and simpler when it comes to
dependencies) to directly have the message hash in the data file.
It was probably this way because some test vectors provide the message
for the sake of all-in-one implementation that hash-and-sign at once.
But our API gets a hash as the input and signs it. In unit tests, this
should be reflected in the signature of the test function, which should
take a hash as input.
The changes to the .data file were done using the following python
script:
import hashlib
suite = 'ecdsa'
functions = {
'ecdsa_det_test_vectors': (3, 4),
'ecdsa_write_restart': (3, 4),
}
def hash_ctx(s):
if s == 'MBEDTLS_MD_MD5':
return hashlib.md5()
if s == 'MBEDTLS_MD_SHA1':
return hashlib.sha1()
if s == 'MBEDTLS_MD_SHA224':
return hashlib.sha224()
if s == 'MBEDTLS_MD_SHA256':
return hashlib.sha256()
if s == 'MBEDTLS_MD_SHA384':
return hashlib.sha384()
if s == 'MBEDTLS_MD_SHA512':
return hashlib.sha512()
def fix(l):
parts = l.rstrip().split(":")
fun = parts[0]
if fun not in functions:
return l
(alg_idx, msg_idx) = functions[fun]
alg_str = parts[alg_idx]
if alg_str == "MBEDTLS_MD_NONE":
return l
h = hash_ctx(alg_str)
msg_str = parts[msg_idx][1:-1]
h.update(msg_str.encode('ascii'))
msg_hash = h.hexdigest()
msg_hash_str = '"' + msg_hash.upper() + '"'
parts[msg_idx] = msg_hash_str
return ":".join(parts) + '\n'
filename = 'tests/suites/test_suite_' + suite + '.data'
with open(filename) as f:
lines = f.readlines()
lines = [fix(l) for l in lines]
with open(filename, 'w') as f:
f.writelines(lines)
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
If generation of files is turned off, and a file is missing, when
building in tree with cmake, you can end up with the generated file
being turned into a symlink to itself. This will also break any future
attempt at building with make. Fix this by testing if the file exists
prior to attempting to link it.
Signed-off-by: Paul Elliott <paul.elliott@arm.com>
Again, the guards are slightly different, let's see in the CI if this
causes any trouble.
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
There is a small difference: the global function only works for hashes
that are included in the build, while the old one worked for all hashes
regardless of whether they were enabled or not.
We'll see in the CI is this causes any issues.
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
This is mostly:
sed -i 's/mbedtls_psa_translate_md/mbedtls_hash_info_psa_from_md/' \
library/*.c tests/suites/*.function
This should be good for code size as the old inline function was used
from 10 translation units inside the library, so we have 10 copies at
least.
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
Using static inline functions is bad for code size; the function from
md_internal.h was already used from 3 different C files, so already was
copied at least 3 times in the library, and this would only get worse
over time.
Use actual functions, and also share the actual data between them.
Provide a consistent set of operations. Conversion to/from
human-readable string was omitted for now but could be added later if
needed.
In the future, this can be used to replace other similar (inline)
functions that are currently scattered, including (but perhaps not
limited to):
- mbedtls_psa_translate_md() from psa_util.h
- mbedtls_md_info_from_psa() (indirectly) from psa_crypto_hash.h
- get_md_alg_from_psa() from psa_crypto_rsa.c
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
Fixes#1910
With ebx added to the MULADDC_STOP clobber list to fix#1550, the inline
assembly fails to build with GCC < 5 in PIC mode with the following error:
include/mbedtls/bn_mul.h:46:13: error: PIC register clobbered by ‘ebx’ in ‘asm’
This is because older GCC versions treated the x86 ebx register (which is
used for the GOT) as a fixed reserved register when building as PIC.
This is fixed by an improved register allocator in GCC 5+. From the release
notes:
Register allocation improvements: Reuse of the PIC hard register, instead of
using a fixed register, was implemented on x86/x86-64 targets. This
improves generated PIC code performance as more hard registers can be used.
https://www.gnu.org/software/gcc/gcc-5/changes.html
As a workaround, detect this situation and disable the inline assembly,
similar to the MULADDC_CANNOT_USE_R7 logic.
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Trying to compile MD_C without any of the hash modules would result in a
bunch of unused parameter warning (hence errors in -Werror builds).
We could silence those warnings by casting the parameters to void, but
still, compiling the module in such a configuration would mean all of
its functions are useless (always returning an error).
Seems better to just document the dependency.
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
It was only used in test_suite_pk which was fixed to no longer compute
hashes in a temporary buffer.
Also, it's not entirely clear is this macro was really a good idea:
perhaps it's better for each user to have an explicit #if
defined(MBEDTSL_USE_PSA_CRYPTO) and use either MBEDTLS_MD_MAX_SIZE or
PSA_HASH_MAX_SIZE in each branch of that #if.
So, removing it for the time being.
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
Same rationale as previous "Rm useless use of MD" commits.
Here the first two test functions were already not depending on MD_C,
but the new version is much simpler, as it avoids having two versions of
the code depending on the value of USE_PSA.
Changes to the data file generated with the following Python script:
import hashlib
suite = 'pk'
functions = {
'pk_rsa_verify_test_vec': (2, 1, True),
'pk_rsa_verify_ext_test_vec': (2, 1, True),
'pk_sign_verify_restart': (6, 7, False),
}
def hash_ctx(s):
if s == 'MBEDTLS_MD_MD5':
return hashlib.md5()
if s == 'MBEDTLS_MD_SHA1':
return hashlib.sha1()
if s == 'MBEDTLS_MD_SHA224':
return hashlib.sha224()
if s == 'MBEDTLS_MD_SHA256':
return hashlib.sha256()
if s == 'MBEDTLS_MD_SHA384':
return hashlib.sha384()
if s == 'MBEDTLS_MD_SHA512':
return hashlib.sha512()
if s == 'MBEDTLS_MD_RIPEMD160':
return hashlib.new("ripemd160")
def fix(l):
parts = l.rstrip().split(":")
fun = parts[0]
if fun not in functions:
return l
(alg_idx, msg_idx, is_hex) = functions[fun]
alg_str = parts[alg_idx]
if alg_str == "MBEDTLS_MD_NONE" or alg_str == "255":
return l
h = hash_ctx(alg_str)
msg_str = parts[msg_idx][1:-1]
msg = bytes.fromhex(msg_str) if is_hex else bytes(msg_str, 'ascii')
h.update(msg)
msg_hash = h.hexdigest()
msg_hash_str = '"' + msg_hash + '"'
parts[msg_idx] = msg_hash_str
return ":".join(parts) + '\n'
filename = 'tests/suites/test_suite_' + suite + '.data'
with open(filename) as f:
lines = f.readlines()
lines = [fix(l) for l in lines]
with open(filename, 'w') as f:
f.writelines(lines)
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
Tests are not here to demonstrate best practice, but to test a specific
part of the code. Using an RNG provided by the test framework also makes
the test code more focused on what we actually mean to test.
This brings the number of tests skipped in test_suite_rsa in
test_psa_crypto_config_accel_hash_use_psa down to 0 (from 50).
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
Probably the result of copy-pasting: the test functions actually don't
use those modules at all.
This brings the number of tests skipped in test_suite_rsa in
test_psa_crypto_config_accel_hash_use_psa down to 50 (from 61).
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
This brings the number of tests skipped in test_suite_rsa in
test_psa_crypto_config_accel_hash_use_psa down to 61 (from 117).
The changes to the test data file were generated with:
sed -i -f or_psa_hash.sed tests/suites/test_suite_rsa.data
with or_psa_hash.sed containing:
s/MBEDTLS_MD5_C/MBEDTLS_OR_PSA_WANT_ALG_MD5/g
s/MBEDTLS_RIPEMD160_C/MBEDTLS_OR_PSA_WANT_ALG_RIPEMD160/g
s/MBEDTLS_SHA1_C/MBEDTLS_OR_PSA_WANT_ALG_SHA_1/g
s/MBEDTLS_SHA224_C/MBEDTLS_OR_PSA_WANT_ALG_SHA_224/g
s/MBEDTLS_SHA256_C/MBEDTLS_OR_PSA_WANT_ALG_SHA_256/g
s/MBEDTLS_SHA384_C/MBEDTLS_OR_PSA_WANT_ALG_SHA_384/g
s/MBEDTLS_SHA512_C/MBEDTLS_OR_PSA_WANT_ALG_SHA_512/g
Here the MBEDTLS_OR_PSA_xxx macros are the right choice as we just need
data about the hashes to be available.
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
Same rationale as two commits ago.
Changes to the data file generated with the following script:
import hashlib
suite = 'rsa'
functions = {
'mbedtls_rsa_pkcs1_sign': (3, 1),
'mbedtls_rsa_pkcs1_verify': (3, 1),
}
def hash_ctx(s):
if s == 'MBEDTLS_MD_MD5':
return hashlib.md5()
if s == 'MBEDTLS_MD_SHA1':
return hashlib.sha1()
if s == 'MBEDTLS_MD_SHA224':
return hashlib.sha224()
if s == 'MBEDTLS_MD_SHA256':
return hashlib.sha256()
if s == 'MBEDTLS_MD_SHA384':
return hashlib.sha384()
if s == 'MBEDTLS_MD_SHA512':
return hashlib.sha512()
if s == 'MBEDTLS_MD_RIPEMD160':
return hashlib.new("ripemd160")
def fix(l):
parts = l.rstrip().split(":")
fun = parts[0]
if fun not in functions:
return l
(alg_idx, msg_idx) = functions[fun]
alg_str = parts[alg_idx]
if alg_str == "MBEDTLS_MD_NONE" or alg_str == "255":
return l
h = hash_ctx(alg_str)
msg_str = parts[msg_idx]
msg_hex = msg_str[1:-1]
msg = bytes.fromhex(msg_hex)
h.update(msg)
msg_hash = h.hexdigest()
msg_hash_str = '"' + msg_hash + '"'
parts[msg_idx] = msg_hash_str
return ":".join(parts) + '\n'
filename = 'tests/suites/test_suite_' + suite + '.data'
with open(filename) as f:
lines = f.readlines()
lines = [fix(l) for l in lines]
with open(filename, 'w') as f:
f.writelines(lines)
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
This brings the number of tests skipped in test_suite_pkcs1_v15 in
test_psa_crypto_config_accel_hash_use_psa to 0.
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
We had a message in the data file, and were computing its hash in the
test function. It is more efficient (and simpler when it comes to
dependencies) to directly have the message hash in the data file.
It was probably this way because some test vectors provide the message
for the sake of all-in-one implementation that hash-and-sign at once.
But our API gets a hash as the input and signs it. In unit tests, this
should be reflected in the signature of the test function, which should
take a hash as input.
The changes to the .data file were done using the following python
script:
import hashlib
suite = 'pkcs1_v15'
functions = {
'pkcs1_rsassa_v15_sign': (10, 12),
'pkcs1_rsassa_v15_verify': (6, 8),
}
def hash_ctx(s):
if s == 'MBEDTLS_MD_MD5':
return hashlib.md5()
if s == 'MBEDTLS_MD_SHA1':
return hashlib.sha1()
if s == 'MBEDTLS_MD_SHA224':
return hashlib.sha224()
if s == 'MBEDTLS_MD_SHA256':
return hashlib.sha256()
if s == 'MBEDTLS_MD_SHA384':
return hashlib.sha384()
if s == 'MBEDTLS_MD_SHA512':
return hashlib.sha512()
def fix(l):
parts = l.rstrip().split(":")
fun = parts[0]
if fun not in functions:
return l
(alg_idx, msg_idx) = functions[fun]
alg_str = parts[alg_idx]
if alg_str == "MBEDTLS_MD_NONE":
return l
h = hash_ctx(alg_str)
msg_str = parts[msg_idx]
msg_hex = msg_str[1:-1]
msg = bytes.fromhex(msg_hex)
h.update(msg)
msg_hash = h.hexdigest()
msg_hash_str = '"' + msg_hash + '"'
parts[msg_idx] = msg_hash_str
return ":".join(parts) + '\n'
filename = 'tests/suites/test_suite_' + suite + '.data'
with open(filename) as f:
lines = f.readlines()
lines = [fix(l) for l in lines]
with open(filename, 'w') as f:
f.writelines(lines)
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
The test script all.sh contains the component
component_build_armcc
testing that Mbed TLS builds using Arm Compiler 5 and 6,
on a variety of platforms.
However, the component does not exercise inline assembly
for Arm, since
- MBEDTLS_HAVE_ASM is unset, and
- Some Arm inline assembly is only used if the level of
optimization is not 0.
This commit changes the test component to ensure that
inline assembly is built by setting MBEDTLS_HAVE_ASM
as well as enabling optimization level 1 (-O1).
Signed-off-by: Hanno Becker <hanno.becker@arm.com>
Within the M-profile of the Arm architecture, some instructions
admit both a 16-bit and a 32-bit encoding. For those instructions,
some assemblers support the use of the .n (narrow) and .w (wide)
suffixes to force a choice of instruction encoding width.
Forcing the size of encodings may be useful to ensure alignment
of code, which can have a significant performance impact on some
microarchitectures.
It is for this reason that a previous commit introduced explicit
.w suffixes into what was believed to be M-profile only assembly
in library/bn_mul.h.
This change, however, introduced two issues:
- First, the assembly block in question is used also for Armv7-A
systems, on which the .n/.w distinction is not meaningful
(all instructions are 32-bit).
- Second, compiler support for .n/.w suffixes appears patchy,
leading to compilation failures even when building for M-profile
targets.
This commit removes the .w annotations in order to restore working
code, deferring controlled re-introduction for the sake of performance.
Fixes#6089.
Signed-off-by: Hanno Becker <hanno.becker@arm.com>
The encrypt/decrypt cases don't depend on actually computing a hash,
just on information about it being available, and this information is
guarded by MBEDTLS_OR_PSA_WANT_xxx.
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
Currently just replacing existing uses, but the real point of having
these conditions as a single macro is that we'll be able to use them in
tests case dependencies, see next commit.
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
Having the whole .function file depend on SHA-1 was wrong: dependencies
in .function files are for compile-time dependencies all functions in
this file build just fine without SHA-1. It's just some tests cases that
do need SHA-1 at runtime, use dependencies on those specific tests in
the .data file then.
This reduces the number of cases skipped in this test suite in
test_psa_crypto_config_accel_hash_use_psa from 28 (all of them) down to
12 (those that actually use SHA-1 as opposed to no hash).
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
