Since Mbed TLS 3.6.0, all officially supported versions of Visual Studio
a printf function family that is sufficiently compliant to C99 for our
purposes, in particular supporting `%zu` for `size_t`. The only platform
without `%zu` that we semi-officially support is older versions of MinGW,
still used in our CI. MinGW provides either a Windows legacy printf or a
standards-compliant printf depending on the value of
`__USE_MINGW_ANSI_STDIO` when compiling each C file. Force the use of the
compliant version. Don't rely on `MBEDTLS_PRINTF_SIZET`, which is defined in
`<mbedtls/debug.h>` and no longer considers the Windows legacy version in
Mbed TLS >= 4.1.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
Always activate `__USE_MINGW_ANSI_STDIO` unless overridden on the command
line. This is necessary with older versions of MingW and/or Windows,
where snprintf does not always zero-terminate the buffer, and does
not support formats such as `"%zu"` for size_t and `"%lld"` for long long.
Simplify debug.h accordingly. The macros `MBEDTLS_PRINTF_SIZET`,
`MBEDTLS_PRINTF_SIZET_HAX` and `MBEDTLS_PRINTF_LONGLONG` are no longer
needed, but they are still used in our code base and must stay in debug.h
for backward compatibility.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
In `test_suite_debug`, test `mbedtls_debug_snprintf()`, which uses
`mbedtls_vsnprintf()` like `mbedtls_debug_print_msg()`. Do this instead of
testing `mbedtls_snprintf()`, which might be subtly different (older
Windows runtimes had slightly different behavior for vsnprintf() vs
snprintf(); TF-PSA-Crypto might pick up a different function if the
platform configuration is different in TF-PSA-Crypto and Mbed TLS).
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
On some platforms, the system headers expose different interfaces depending
on what macros are defined, for example to provide different standards
compliance level. Create a common place where we can declare such macros,
so that our code can behave in the same way when it's in different files.
Individual .c files can still override these requirements by defining
macros before including the common header, if it's really necessary.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
We already have `x509_internal.h` which is common to all parts of the X.509
library, and `ssl_misc.h` which is common to all parts of the TLS library.
Also create `mbedtls_common.h` which is for the Mbed TLS project as a whole.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
Don't ship two slightly different wheels.
This reduces our platform adherence by using only `clock_gettime()` in the
library and not `gettimeofday()` as well.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
Nowadays, the timing module just builds on a function that provides a timer
with millisecond resolution. In terms of platform requirements, this is
almost exactly equivalent to `mbedtls_ms_time()`
provides (`mbedtls_ms_time()` is arguably a little stronger because it is
supposed to last longer than a single timer object, but an application could
start a timer when it starts, so there's no real difference.) So it's a bit
silly that `timing.c` essentially reimplements this. Rely on
`mbedtls_ms_time()` instead.
This is an API break because in Mbed TLS 4.0, it was possible to enable
`MBEDTLS_TIMING_C` without `MBEDTLS_HAVE_TIME`. However, `timing.c` only
provided an implementation for Windows and Unix-like platforms, and on those
platforms, it is very likely that the default implementation of
`MBEDTLS_HAVE_TIME` would also work. (The main exception would be a platform
that has the traditional Unix function `gettimeofday()`, but not the 1990s
novelty `clock_gettime()`.) So make this an official requirement, as a
belated change that really should have gone into 4.0 if we'd taken the time
to dig into it.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
Improve DTLS proxy 3d tests with OpenSSL and
GnuTLS servers. Have a better control of which
message is fragmented and verify it is the
case.
Signed-off-by: Ronald Cron <ronald.cron@arm.com>
Improve DTLS reassembly tests with OpenSSL
and GnuTLS server. Check that some messages
have been reassembled.
Signed-off-by: Ronald Cron <ronald.cron@arm.com>
We are about to have full support for TLS 1.2
CH reassembly on server side. The equivalent
positive test would be a duplicate of one of
the tests generated by generate_tls_handshake_tests.py.
Thus just removing the negative test.
Signed-off-by: Ronald Cron <ronald.cron@arm.com>
In the TLS-Exporter for TLS 1.3 we mistakenly call PSA_HASH_LENGTH() on
an mbedtls_md_type_t when it should be called on a psa_algorithm_t.
Fortunately, these two types have almost the same values, since we have
previously aligned them to make conversion more efficient. As a result,
PSA_HASH_LENGTH() produces exactly the same value when called on an
mbedtls_md_type_t as with the equivalent psa_algorithm_t.
Thanks to this happy coincidence, fix a largely cosmetic issue (rather
than a major functional bug).
Signed-off-by: David Horstmann <david.horstmann@arm.com>
TLS-PRF uses either SHA-256 and SHA-384, so the removed paragraph was not
correct. The correct version is already available few lines below in the
same header file.
Signed-off-by: Valerio Setti <valerio.setti@nordicsemi.no>
Specify that the cryptographic operations of Mbed TLS are governed by
its threat model and point specifically to block ciphers as an important
case of this.
Signed-off-by: David Horstmann <david.horstmann@arm.com>
To avoid confusion about the threat model of cryptographic code, add a
link to the SECURITY.md of TF-PSA-Crypto. This should help users who are
unaware that the cryptography has been split into a separate repository.
Signed-off-by: David Horstmann <david.horstmann@arm.com>
