Integrators in a client-server architecture need to provide this function on
the client side.
Fixesmbedtls/issues#10341.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
This is meant to hold threading-related definitions that are not public, but
are used in the test framework.
To be populated later.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
Add more detail around how generation of configuration-independent files
chooses a C compiler. Mention that setting HOSTCC or CC is recommended
where there are multiple toolchains.
Mention that the fallback location is the cc executable, which may help
users troubleshooting when the file generation picks up the wrong
toolchain (as in Mbed-TLS/mbedtls#10360).
Signed-off-by: David Horstmann <david.horstmann@arm.com>
Test that `make lib GEN_FILES=` does build the generated files if they're
missing. Also, test that this only requires the expected commands: `$(CC)`,
`$(AR)`, `$(PERL)` and `$(PYTHON)`. For Python (and Perl), we don't test for
reliance on unexpected third-party packages: that would be desirable, but
out of scope of this commit.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
Don't test `make clean` at the toplevel, that would be too much work (we'd
need to support `$(RM)` in all makefiles, and arrange for `find` as well for
`clean_more_on_top`).
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
Fix `make lib GEN_FILES=` attempting to rebuild
`psa_crypto_driver_wrappers*` if the files' timestamps are older than their
dependencies. Turning off `GEN_FILES` is supposed to avoid that.
Fixes#10335.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
Test that `make lib GEN_FILES=` works in a minimal environment (just `${CC}`
and `${AR}`). We promise that in `README.md`.
Non-regression test for #10335.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
Add a custom target that depends on crypto
generated files, and make both the static and
shared crypto libraries depend on it.
This ensures that when both libraries are built,
the files are not generated concurrently
by the static and shared library targets.
Do the same for the TLS libraries.
Signed-off-by: Ronald Cron <ronald.cron@arm.com>
The dependencies declared in ci.requirements.txt are only used in
scripts that we run on the Linux CI.
Signed-off-by: Bence Szépkúti <bence.szepkuti@arm.com>
Recent versions of cryptography require a Rust toolchain to install on
FreeBSD, which we do not have set up yet.
Signed-off-by: Bence Szépkúti <bence.szepkuti@arm.com>
The version was unspecified because of our use of Python 3.5 on the CI,
whichi has since been eliminated.
Signed-off-by: Bence Szépkúti <bence.szepkuti@arm.com>
In some cases, we were calling `mbedtls_test_ssl_endpoint_free()` on an
uninitialized `mbedtls_test_ssl_endpoint` object if the test case failed
early, e.g. due to `psa_crypto_init()` failing. This was largely harmless,
but could have caused weird test results in case of failure, and was flagged
by Coverity.
Use a more systematic style for initializing the stack object as soon as
it's declared.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
In some cases, we were calling `mbedtls_test_ssl_endpoint_free()` on an
uninitialized `mbedtls_test_ssl_endpoint` object if the test case failed
early, e.g. due to `psa_crypto_init()` failing. This was largely harmless,
but could have caused weird test results in case of failure, and was flagged
by Coverity.
Use a more systematic style for initializing the stack object as soon as
it's declared.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
