After opening a file containing sensitive data, call mbedtls_setbuf() to
disable buffering. This way, we don't expose sensitive data to a memory
disclosure vulnerability in a buffer outside our control.
This commit adds a call to mbedtls_setbuf() after each call to fopen(),
but only in sample programs that were calling mbedtls_platform_zeroize().
Don't bother protecting stdio buffers in programs where application buffers
weren't protected.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
GCC 12 emits a warning because it thinks `buffer1` is used after having been
freed. The code is correct C because we're only using the value of
`(uintptr_t)buffer1`, not `buffer1`. However, we aren't using the value for
anything useful: it doesn't really matter if an alloc-free-alloc sequence
returns the same address twice. So don't print that bit of information, and
this way we don't need to save the old address.
Fixes#5974.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
Add one RSA PSS signature algorithm to the
test list of signature algorithms. This allows
certificate chains exposing an RSA key with
signatures using SHA-1 to be used in tests
where an TLS 1.3 handshake is performed.
Signed-off-by: Ronald Cron <ronald.cron@arm.com>
Remaining hits seem to be hex data, certificates,
and other miscellaneous exceptions.
List generated by running codespell -w -L
keypair,Keypair,KeyPair,keyPair,ciph,nd
Signed-off-by: Andrzej Kurek <andrzej.kurek@arm.com>
- key_opaque_set_alg_usage(): set alg/usage in loop
- key_opaque_set_alg_usage(): add key paramteter to set default alg/usage if it is not specified by command line parameters
- unify default alg/usage for client and server
- optimize opaque code on client and server side
Signed-off-by: Przemek Stekiel <przemyslaw.stekiel@mobica.com>
The QueryPerformanceCounter implementation previously was within
defined(_MSC_VER), but it works just as well on other Windows
toolchains, like mingw.
For most common mingw x86 build configurations, one of the earlier
inline assembly implementations would end up used, but for non-x86
(arm, aarch64), it would end up falling back on the gettimeofday
implementation.
This implementation did build successfully (as mingw toolchains do
provide gettimeofday, contrary to MSVC), but the header providing
gettimeofday, <sys/time.h>, wasn't ever included when building
targeting Windows - thus the function was called without a proper
declaration.
Clang 15 changes such implicit function declarations into a hard
error by default, when building in C99 mode (or newer) [1].
(While Clang 15 still is under development, this may still change
before it's released, but it's a valid issue in any case.)
[1] 7d644e1215
Signed-off-by: Martin Storsjö <martin@martin.st>
remove use of MBEDTLS_SSL_MINOR_VERSION_*
remove use of MBEDTLS_SSL_MAJOR_VERSION_*
(only remaining use is in tests/suites/test_suite_ssl.data)
Signed-off-by: Glenn Strauss <gstrauss@gluelogic.com>
Store the TLS version in tls_version instead of major, minor version num
Note: existing application use which accesses the struct member
(using MBEDTLS_PRIVATE) is not compatible, as the struct is now smaller.
Reduce size of mbedtls_ssl_ciphersuite_t
members are defined using integral types instead of enums in
order to pack structure and reduce memory usage by internal
ciphersuite_definitions[]
Signed-off-by: Glenn Strauss <gstrauss@gluelogic.com>
This commit modifies programs/test/selftest to include a check that
none of the standard integer types (unsigned) [short, int, long, long]
uses padding bits, which we currently don't support.
Signed-off-by: Dave Rodgman <dave.rodgman@arm.com>
time() is only needed to seed the PRNG non-deterministically. If it isn't
available, do seed it, but pick a static seed.
Signed-off-by: Andrzej Kurek <andrzej.kurek@arm.com>
The added null byte was accounted for twice, once by taking
opt.buffer_size+1 when allocating the buffer and once by taking opt.buffer-1
when filling the buffer. Make opt.buffer_size the size that is actually
read, it's less confusing that way.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
Make sure that buf always has enough room for what it will contain. Before,
this was not the case if the buffer was smaller than the default response,
leading to memory corruption in ssl_server2.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
MBEDTLS_HAVE_TIME_ALT implies MBEDTLS_HAVE_TIME, so an extra
check for MBEDTLS_HAVE_TIME is not needed.
Signed-off-by: David Horstmann <david.horstmann@arm.com>
Remove direct inclusion of mbedtls_config.h and replace with
build_info.h, as is the convention in Mbed TLS 3.0.
Signed-off-by: David Horstmann <david.horstmann@arm.com>
Allow programs/test/udp_proxy.c to build when MBEDTLS_HAVE_TIME is
not defined. In this case, do not attempt to seed the pseudo-random
number generator used to sometimes produce corrupt packets and other
erroneous data.
Signed-off-by: David Horstmann <david.horstmann@arm.com>
MBEDTLS_HAVE_TIME is documented as: "System has time.h and time()."
If that is not defined, do not attempt to include time.h.
A particular problem is platform-time.h, which should only be included if
MBEDTLS_HAVE_TIME is defined, which makes everything messier. Maybe it
should be refactored to have the check inside the header.
Signed-off-by: Daniel Axtens <dja@axtens.net>
In fuzz_privkey, we switched over to using dummy_rand(), which uses
ctr_drbg internally, and thus requires an initialised ctr_drbg_context
to be passed in via p_rng when calling mbedtls_pk_parse_key().
Signed-off-by: Paul Elliott <paul.elliott@arm.com>
ctr_drbg is a local variable and thus needs initialisation every time
LLVMFuzzerTestOneInput() is called, the rest of the variables inside the
if(initialised) block are all static.
Add extra validation to attempt to catch this issue in future.
Signed-off-by: Paul Elliott <paul.elliott@arm.com>
Used to print "cipher:" when it was the cipher part of a program that
had both cipher and PSA. Now it doesn't really make sense. Align the
output to match the PSA version of this program.
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
Visual Studio and CMake didn't like having targets with the same name,
albeit in different directories.
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
- avoid hardcoded sizes when there's a macro for that
- avoid mutable global variables
- zeroize potentially-sensitive local buffer on exit
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
MSVC 2013, still supported and used in our CI, did not support that.
aead_psa.c(78): error C2099: initializer is not a constant
aead_psa.c(168): error C2057: expected constant expression
aead_psa.c(168): error C2466: cannot allocate an array of constant size 0
aead_psa.c(168): error C2133: 'out' : unknown size
aead_psa.c(169): warning C4034: sizeof returns 0
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
Having two programs might make comparison easier, and will make it
easier to people to use just the PSA one as an example.
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
If no key is loaded in a slot, say "none", not "invalid PK".
When listing two key types, use punctuation that's visibly a sequence
separator (",").
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
Existing example programs in this directory are already incompatible
with that option, so this is probably acceptable here too.
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
This is meant to highlight similarities and differences in the
multi-part HMAC APIs.
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
At the end of the benchmark program, heap stats are printed, and these
stats will be wrong if we reset counters in the middle.
Also remove the function to reset counters, in order to encourage other
programs to behave correctly as well.
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
This no longer makes sense since pre-computed multiples of the base
point are now static. The function was not doing anything since `grp.T`
was set to `NULL` when exiting `ecp_mul_comb()` anyway.
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
The "proper" fix would be to define the function only when it's needed,
but the condition for that would be tedious to write (enumeration of all
symmetric crypto modules) and since this is a utility program, not the
core library, I think it's OK to keep unused functions.
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
Fix library references, tests and programs.
Testing is performed in the already present all.sh test.
Signed-off-by: Andrzej Kurek <andrzej.kurek@arm.com>
As we have now a minimal viable implementation of TLS 1.3,
let's remove EXPERIMENTAL from the config option enabling
it.
Signed-off-by: Ronald Cron <ronald.cron@arm.com>
If -f was used as an argument twice to the program, then it would leak
the file resource, due to overwriting it on the second pass
Signed-off-by: Paul Elliott <paul.elliott@arm.com>
Remove support signature PKCS1 v1.5 in CertificateVerify.
Remove useless server states in test script
Signed-off-by: XiaokangQian <xiaokang.qian@arm.com>
Remove useless component in all.sh
Remove use server logs in ssh-opt.sh
Remove useless guards in ssl_client2.c
Signed-off-by: XiaokangQian <xiaokang.qian@arm.com>
Change client test code to support rsa pss signatures
Add test cases for rsa pss signature in ssl-opt.sh
Signed-off-by: XiaokangQian <xiaokang.qian@arm.com>
Non-regression for the fix in https://github.com/ARMmbed/mbedtls/pull/5126:
libmbedtls and libmbedx509 did not declare their dependencies on libmbedx509
and libmbedcrypto when built with make.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
I had originally thought to support directories with
mbedtls_x509_crt_parse_path but it would have complicated the code more than
I cared for. Remove a remnant of the original project in the documentation.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
Change one occurrence of ${PYTHON} to ${MBEDTLS_PYTHON_EXECUTABLE}
and add implied ${MBEDTLS_PYTHON_EXECUTABLE} to the start of a
different command.
Signed-off-by: David Horstmann <david.horstmann@arm.com>
When the option is On, CMake will have rules to generate the generated
files using scripts etc. When the option is Off, CMake will assume the
files are available from the source tree; in that mode, it won't require
any extra tools (Perl for example) compared to when we committed the
files to git.
The intention is that users will never need to adjust this option:
- in the development branch (and features branches etc.) the option is
always On (development mode);
- in released tarballs, which include the generated files, we'll switch
the option to Off (release mode) in the same commit that re-adds the
generated files.
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
This one's a bit funny too as the generated file is not a source to the
executable (ie, it's not passed as an argument to the compiler), so
CMake's dependency resolution didn't work even though the file is in the
same directory.
For some reason, the following didn't work either:
add_dependencies(psa_constant_names
${CMAKE_CURRENT_BINARY_DIR}/psa_constant_names_generated.c)
So, apply the same strategy as for cross-directory use of a generated
file by creating a target and using it as a dependency.
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
query_config was added twice, and while at it let's declare all the
sources in one place
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
This option only gated an ability to set a callback,
but was deemed unnecessary as it was yet another define to
remember when writing tests, or test configurations. Fixes#4653.
Signed-off-by: Andrzej Kurek <andrzej.kurek@arm.com>
Declare all AES and DES functions that return int as needing to have
their result checked, and do check the result in our code.
A DES or AES block operation can fail in alternative implementations of
mbedtls_internal_aes_encrypt() (under MBEDTLS_AES_ENCRYPT_ALT),
mbedtls_internal_aes_decrypt() (under MBEDTLS_AES_DECRYPT_ALT),
mbedtls_des_crypt_ecb() (under MBEDTLS_DES_CRYPT_ECB_ALT),
mbedtls_des3_crypt_ecb() (under MBEDTLS_DES3_CRYPT_ECB_ALT).
A failure can happen if the accelerator peripheral is in a bad state.
Several block modes were not catching the error.
This commit does the following code changes, grouped together to avoid
having an intermediate commit where the build fails:
* Add MBEDTLS_CHECK_RETURN to all functions returning int in aes.h and des.h.
* Fix all places where this causes a GCC warning, indicating that our code
was not properly checking the result of an AES operation:
* In library code: on failure, goto exit and return ret.
* In pkey programs: goto exit.
* In the benchmark program: exit (not ideal since there's no error
message, but it's what the code currently does for failures).
* In test code: TEST_ASSERT.
* Changelog entry.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
mbedtls_cipher_setkey takes an int argument. Cast explicitly, otherwise MSVC
complains.
Where possible, just stick to size_t.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
The structures mbedtls_x509_time, mbedtls_x509_crl_entry, mbedtls_x509_crl,
mbedtls_x509_crt, mbedtls_x509_san_other_name,
mbedtls_x509_subject_alternative_name, mbedtls_x509_csr are designed to
expose the result of parsing X.509 data. Document many of their fields as
being publicly readable.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
On platforms with BSD-like sockets, it is useful for applications to have
access to the underlying file descriptor so that they can use functions like
select() and poll().
Do not promise that the field will exist on other platforms such as
Windows (where the type and name of the field are technically wrong because
Windows socket handles are actually not file descriptors).
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
The whole point of this structure is to provide information, both for the
library's own sake and to applications.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
If socket setup fail, ssl structure setup won't be called.
And the order of them do not affect final result, but it
will break ssl setup negative tests.
Change the order can fix that.
issue: #4844
Change-Id: I2488ed5f74773421eb1eac0cfd7f1ce4fbb0b32d
Signed-off-by: Jerry Yu <jerry.h.yu@arm.com>
To support tls1.3 relative tests, add `tls1_3`
parameter for `{min,max}_version` and `force_version`
issues: #4844
Change-Id: I1b22a076582374b8aabc733086562e9d03a94a2a
Signed-off-by: Jerry Yu <jerry.h.yu@arm.com>
The makefiles look for python3 on Unix-like systems where python is often
Python 2. This uses sh code so it doesn't work on Windows. On Windows, the
makefiles just assume that python is Python 3.
The code was incorrectly deciding not to try python3 based on WINDOWS_BUILD,
which indicates that the build is *for* Windows. Switch to checking WINDOWS,
which indicates that the build is *on* Windows.
Fix#4774
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
Generate programs/test/cpp_dummy_build.cpp dynamically instead of
maintaining it manually. This removes the need to update it when the list of
headers changes.
Include all the headers unconditionally except for the ones that cannot be
included directly.
Support this dynamic generation both with make and with cmake.
Adapt all.sh accordingly. Remove the redundant C build from
component_build_default_make_gcc_and_cxx (it was also done in
component_test_default_out_of_box), leaving a component_test_make_cxx. Also
run the C++ program, because why not. Do this in the full configuration
which may catch a bit more problems in headers.
Fixes#2570 for good.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
We previously introduced a safety check ensuring that if a datagram had
already been dropped twice, it would no longer be dropped or delayed
after that.
This missed an edge case: if a datagram is dropped once, it can be
delayed any number of times. Since "delay" is not defined in terms of
time (x seconds) but in terms of ordering with respect to other messages
(will be forwarded after the next message is forwarded), depending on
the RNG results this could result in an endless loop where all messages
are delayed until the next, which is itself delayed, etc. and no message
is ever forwarded.
The probability of this happening n times in a row is (1/d)^n, where d
is the value passed as delay=d, so for delay=5 and n=5 it's around 0.03%
which seems small but we still happened on such an occurrence in real
life:
tests/ssl-opt.sh --seed 1625061502 -f 'DTLS proxy: 3d, min handshake, resumption$'
results (according to debug statements added for the investigation) in
the ClientHello of the second handshake being dropped once then delayed
5 times, after which the client stops re-trying and the test fails for
no interesting reason.
Make sure this doesn't happen again by putting a cap on the number of
times we fail to forward a given datagram immediately.
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
