This commit introduces a new SSL error code
`MBEDTLS_ERR_SSL_VERSION_MISMATCH`
which can be used to indicate operation failure due to a
mismatch of version or configuration.
It is put to use in the implementation of `mbedtls_ssl_session_load()`
to signal the attempt to de-serialize a session which has been serialized
in a build of Mbed TLS using a different version or configuration.
This commit improves the test exercising the behaviour of
session deserialization when facing an unexpected version
or config, by testing ver/cfg corruption at any bit in the
ver/cfg header of the serialized data; previously, it had
only tested the first bit of each byte.
The size of the ticket used in this test dropped from 192 to 143 bytes, so
move all sizes used in this test down 50 bytes. Also, we now need to adapt the
server response size as the default size would otherwise collide with the new
mtu value.
We have explicit recommendations to use US spelling for technical writing, so
let's apply this to code as well for uniformity. (My fingers tend to prefer UK
spelling, so this needs to be fixed in many places.)
sed -i 's/\([Ss]eriali\)s/\1z/g' **/*.[ch] **/*.function **/*.data ChangeLog
This test works regardless of the serialisation format and embedded pointers
in it, contrary to the load-save test, though it requires more maintenance of
the test code (sync the member list with the struct definition).
This uncovered a bug that led to a double-free (in practice, in general could
be free() on any invalid value): initially the session structure is loaded
with `memcpy()` which copies the previous values of pointers peer_cert and
ticket to heap-allocated buffers (or any other value if the input is
attacker-controlled). Now if we exit before we got a chance to replace those
invalid values with valid ones (for example because the input buffer is too
small, or because the second malloc() failed), then the next call to
session_free() is going to call free() on invalid pointers.
This bug is fixed in this commit by always setting the pointers to NULL right
after they've been read from the serialised state, so that the invalid values
can never be used.
(An alternative would be to NULL-ify them when writing, which was rejected
mostly because we need to do it when reading anyway (as the consequences of
free(invalid) are too severe to take any risk), so doing it when writing as
well is redundant and a waste of code size.)
Also, while thinking about what happens in case of errors, it became apparent
to me that it was bad practice to leave the session structure in an
half-initialised state and rely on the caller to call session_free(), so this
commit also ensures we always clear the structure when loading failed.
This test appeared to be passing for the wrong reason, it's not actually not
appropriate for the current implementation. The serialised data contains
values of pointers to heap-allocated buffers. There is no reason these should
be identical after a load-save pair. They just happened to be identical when I
first ran the test due to the place of session_free() in the test code and the
fact that the libc's malloc() reused the same buffers. The test no longer
passes if other malloc() implementations are used (for example, when compiling
with asan which avoids re-using the buffer, probably for better error
detection).
So, disable this test for now (we can re-enable it when we changed how
sessions are serialised, which will be done in a future PR, hence the name of
the dummy macro in depends_on). In the next commit we're going to add a test
that save-load is the identity instead - which will be more work in testing as
it will require checking each field manually, but at least is reliable.
This initial test ensures that a load-save function is the identity. It is so
far incomplete in that it only tests sessions without tickets or certificate.
This will be improved in the next commits.
When running make with parallelization, running both "clean" and "lib"
with a single make invocation can lead to each target building in
parallel. It's bad if lib is partially done building something, and then
clean deletes what was built. This can lead to errors later on in the
lib target.
$ make -j9 clean lib
CC aes.c
CC aesni.c
CC arc4.c
CC aria.c
CC asn1parse.c
CC ./library/error.c
CC ./library/version.c
CC ./library/version_features.c
AR libmbedcrypto.a
ar: aes.o: No such file or directory
Makefile:120: recipe for target 'libmbedcrypto.a' failed
make[2]: *** [libmbedcrypto.a] Error 1
Makefile:152: recipe for target 'libmbedcrypto.a' failed
make[1]: *** [libmbedcrypto.a] Error 2
Makefile:19: recipe for target 'lib' failed
make: *** [lib] Error 2
make: *** Waiting for unfinished jobs....
To avoid this sort of trouble, always invoke clean by itself without
other targets throughout the library. Don't run clean in parallel with
other rules. The only place where clean was run in parallel with other
targets was in list-symbols.sh.
- Replace 'RSA with MD2' OID '2a864886f70d010102' by
'RSA with SHA-256' OID '2a864886f70d01010b':
Only the last byte determines the hash, and
`MBEDTLS_OID_PKCS1_MD2 == MBEDTLS_OID_PKCS1 "\x02"`
`MBEDTLS_OID_PKCS1_SHA256 == MBEDTLS_OID_PKCS1 "\x0b"`
See oid.h.
- Replace MD2 dependency by SHA256 dependency.
- Adapt expected CRT info output.
While 'session hash' is currently unique, so suitable to prove that the
intended code path has been taken, it's a generic enough phrase that in the
future we might add other debug messages containing it in completely unrelated
code paths. In order to future-proof the accuracy of the test, let's use a
more specific string.
This commit temporarily comments the copying of the negotiated CIDs
into the established ::mbedtls_ssl_transform in mbedtls_ssl_derive_keys()
until the CID feature has been fully implemented.
While mbedtls_ssl_decrypt_buf() and mbedtls_ssl_encrypt_buf() do
support CID-based record protection by now and can be unit tested,
the following two changes in the rest of the stack are still missing
before CID-based record protection can be integrated:
- Parsing of CIDs in incoming records.
- Allowing the new CID record content type for incoming records.
- Dealing with a change of record content type during record
decryption.
Further, since mbedtls_ssl_get_peer_cid() judges the use of CIDs by
the CID fields in the currently transforms, this change also requires
temporarily disabling some grepping for ssl_client2 / ssl_server2
debug output in ssl-opt.sh.
Part of the record encryption/decryption tests is to gradually
increase the space available at the front and/or at the back of
a record and observe when encryption starts to succeed. If exactly
one of the two parameters is varied at a time, the expectation is
that encryption will continue to succeed once it has started
succeeding (that's not true if both pre- and post-space are varied
at the same time).
Moreover, previously the test would take turns when choosing which
transform should be used for encryption, and which for decryption.
With the introduction of the CID feaature, this switching of transforms
doesn't align with the expectation of eventual success of the encryption,
since the overhead of encryption might be different for the parties,
because both parties may use different CIDs for their outgoing records.
This commit modifies the tests to not take turns between transforms,
but to always use the same transforms for encryption and decryption
during a single round of the test.
We've observed that sometimes check-names.sh exits unexpectedly with
status 2 and no error message. The failure is not reproducible. This
commits makes the script print a trace if it exits unexpectedly.
We called in tinycrypt in the file names, but uecc in config.h, all.sh and
other places, which could be confusing. Just use tinycrypt everywhere because
that's the name of the project and repo where we took the files.
The changes were made using the following commands (with GNU sed and zsh):
sed -i 's/uecc/tinycrypt/g' **/*.[ch] tests/scripts/all.sh
sed -i 's/MBEDTLS_USE_UECC/MBEDTLS_USE_TINYCRYPT/g' **/*.[ch] tests/scripts/all.sh scripts/config.pl
This commit adds tests to check the behavior of the record encryption
routine `ssl_encrypt_buf` when the buffer surrounding the plaintext is
too small to hold the expansion in the beginning and end (due to IV's,
padding, and MAC).
Each test starts successively increases the space available at the
beginning, end, or both, of the record buffer, and checks that the
record encryption either fails with a BUFFER_TOO_SMALL error, or
that it succeeds. Moreover, if it succeeds, it is checked that
decryption succeeds, too, and results in the original record.
This commit adds tests exercising mutually inverse pairs of
record encryption and decryption transformations for the various
transformation types allowed in TLS: Stream, CBC, and AEAD.
requires_config_enabled doesn't support multiple config options.
Tests having multiple configuration dependencies must be prefixed
with multiple invocations of requires_config_enabled instead.
When doing ABI/API checking, its useful to have a list of all the
identifiers that are defined in the internal header files, as we
do not promise compatibility for them. This option allows for a
simple method of getting them for use with the ABI checking script.
Run ssl-opt.sh on x86_32 with ASan. This may detect bugs that only
show up on 32-bit platforms, for example due to size_t overflow.
For this component, turn off some memory management features that are
not useful, potentially slow, and may reduce ASan's effectiveness at
catching buffer overflows.
* origin/pr/2470:
Silence pylint
check-files.py: readability improvement in permission check
check-files.py: use class fields for class-wide constants
check-files.py: clean up class structure
abi_check.py: Document more methods
check-files.py: document some classes and methods
Fix pylint errors going uncaught
Call pylint3, not pylint
New, documented pylint configuration
* origin/pr/2364:
Increase okm_hex buffer to contain null character
Minor modifications to hkdf test
Add explanation for okm_string size
Update ChangeLog
Reduce buffer size of okm
Reduce Stack usage of hkdf test function
* restricted/pr/553:
Fix mbedtls_ecdh_get_params with new ECDH context
Add changelog entry for mbedtls_ecdh_get_params robustness
Fix ecdh_get_params with mismatching group
Add test case for ecdh_get_params with mismatching group
Add test case for ecdh_calc_secret
Fix typo in documentation
It was failing to set the key in the ENCRYPT direction before encrypting.
This just happened to work for GCM and CCM.
After re-encrypting, compare the length to the expected ciphertext
length not the plaintext length. Again this just happens to work for
GCM and CCM since they do not perform any kind of padding.
* origin/pr/2436:
Use certificates from data_files and refer them
Specify server certificate to use in SHA-1 test
refactor CA and SRV certificates into separate blocks
refactor SHA-1 certificate defintions and assignment
refactor server SHA-1 certificate definition into a new block
define TEST_SRV_CRT_RSA_SOME in similar logic to TEST_CA_CRT_RSA_SOME
server SHA-256 certificate now follows the same logic as CA SHA-256 certificate
add entry to ChangeLog
* restricted/pr/550:
Update query_config.c
Fix failure in SSLv3 per-version suites test
Adjust DES exclude lists in test scripts
Clarify 3DES changes in ChangeLog
Fix documentation for 3DES removal
Exclude 3DES tests in test scripts
Fix wording of ChangeLog and 3DES_REMOVE docs
Reduce priority of 3DES ciphersuites
* public/pr/2429:
Add ChangeLog entry for unused bits in bitstrings
Improve docs for ASN.1 bitstrings and their usage
Add tests for (named) bitstring to suite_asn1write
Fix ASN1 bitstring writing
The test used 3DES as the suite for SSLv3, which now makes the handshake fails
with "no ciphersuite in common", failing the test as well. Use Camellia
instead (as there are not enough AES ciphersuites before TLS 1.2 to
distinguish between the 3 versions).
Document some dependencies, but not all. Just trying to avoid introducing new
issues by using a new cipher here, not trying to make it perfect, which is a
much larger task out of scope of this commit.
Line issue trackers are conceptually a subclass of file issue
trackers: they're file issue trackers where issues arise from checking
each line independently. So make it an actual subclass.
Pylint pointed out the design smell: there was an abstract method that
wasn't always overridden in concrete child classes.
Make check-python-files.sh run pylint on all *.py files (in
directories where they are known to be present), rather than list
files explicitly.
Fix a bug whereby the return status of check-python-files.sh was only
based on the last file passing, i.e. errors in other files were
effectively ignored.
Make check-python-files.sh run pylint unconditionally. Since pylint3
is not critical, make all.sh to skip running check-python-files.sh if
pylint3 is not available.
The pylint configuration in .pylint was a modified version of the
output of `pylint --generate-rcfile` from an unknown version of
pylint. Replace it with a file that only contains settings that are
modified from the default, with an explanation of why each setting is
modified.
The new .pylintrc was written from scratch, based on the output of
pylint on the current version of the files and on a judgement of what
to silence generically, what to silence on a case-by-case basis and
what to fix.
Add a test case for doing an ECDH calculation by calling
mbedtls_ecdh_get_params on both keys, with keys belonging to
different groups. This should fail, but currently passes.
When all.sh invokes check_headers_in_cpp, a backup config.h exists. This
causes a stray difference vs cpp_dummy_build.cpp. Fix by only collecting
the *.h files in include/mbedtls.
Change-Id: Ifd415027e856858579a6699538f06fc49c793570
`test_hkdf` in the hkdf test suites consumed stack of ~6KB with
6 buffers of ~1KB each. This causes stack overflow on some platforms
with smaller stack. The buffer sizes were reduced. By testing, the sizes
can be reduced even further, as the largest seen size is 82 bytes(for okm).
Wildcard patterns now work with command line COMPONENT arguments
without --except as well as with. You can now run e.g.
`all.sh "check_*` to run all the sanity checks.
After backing up and restoring config.h, `git diff-files` may report
it as potentially-changed because it isn't sure whether the index is
up to date. Use `git diff` instead: it actually reads the file.
Only look for armcc if component_build_armcc is to be executed,
instead of requiring the option --no-armcc.
You can still pass --no-armcc, but it's no longer required when
listing components to run. With no list of components or an exclude
list on the command line, --no-armcc is equivalent to having
build_armcc in the exclude list.
Build the list of components to run in $RUN_COMPONENTS as part of
command line parsing. After parsing the command line, it no longer
matters how this list was built.
Extract the list of available components by looking for definitions of
functions called component_xxx. The previous code explicitly listed
all components in run_all_components, which opened the risk of
forgetting to list a component there.
Add a conditional execution facility: if a function support_xxx exists
and returns false then component_xxx is not executed (except when the
command line lists an explicit set of components to execute).
MAKEFLAGS was set to -j if it was already set, instead of being set if
not previously set as intended. So now all.sh will do parallel builds
if invoked without MAKEFLAGS in the environment.
Don't bail out of all.sh if the OS isn't Linux. We only expect
everything to pass on a recent Linux x86_64, but it's useful to call
all.sh to run some components on any platform.
In all.sh, always run both MemorySanitizer and Valgrind. Valgrind is
slower than ASan and MSan but finds some things that they don't.
Run MSan unconditionally, not just on Linux/x86_64. MSan is supported
on some other OSes and CPUs these days.
Use `all.sh --except test_memsan` if you want to omit MSan because it
isn't supported on your platform. Use `all.sh --except test_memcheck`
if you want to omit Valgrind because it's too slow.
Make the test scripts more portable (tested on FreeBSD): don't insist
on GNU sed, and recognize amd64 as well as x86_64 for `uname -m`. The
`make` utility must still be GNU make.
Call `set disable-randomization off` only if it seems to be supported.
The goal is to neither get an error about disable-randomization not
being supported (e.g. on FreeBSD), nor get an error if it is supported
but fails (e.g. on Ubuntu).
Only fiddle with disable-randomization from all.sh, which cares
because it reports the failure of ASLR disabling as an error. If a
developer invokes the Gdb script manually, a warning about ASLR
doesn't matter.
Use `cmake -D CMAKE_BUILD_TYPE=Asan` rather than manually setting
`-fsanitize=address`. This lets cmake determine the necessary compiler
and linker flags.
With UNSAFE_BUILD on, force -Wno-error. This is necessary to build
with MBEDTLS_TEST_NULL_ENTROPY.
Merge the work on all.sh that was done on mbedtls-2.14.0 with the
changes from mbedtls-2.14.0 to mbedtls-2.16.0.
There is a merge conflict in test/scripts/all.sh, which is the only
file that was modified in the all.sh work branch. I resolved it by
taking the copy from the all.sh branch and applying the changes
between mbedtls-2.14.0 and mbedtls-2.16.0. These changes consisted of
two commits:
* "Add tests to all.sh for CHECK_PARAMS edge cases": adds two
test components which are reproduced here as
test_check_params_without_platform and component_test_check_params_silent.
* "tests: Backup config.h before modifying it": moot because the
component framework introduced in the all.sh branch backs up config.h
systematically.
In all.sh, always save config.h before running a component, instead of
doing it manually in each component that requires it (except when we
forget, which has happened). This would break a script that requires
config.h.bak not to exist, but we don't have any of those.
Call cleanup from run_component instead of calling it from each
individual component function.
Clean up after each component rather than before. With the new
structure it makes more sense for each component to leave the place
clean. Run cleanup once at the beginning to start from a clean slate.
Move almost all the code of this script into functions. There is no
intended behavior change. The goal of this commit is to make
subsequent improvements easier to follow.
A very large number of lines have been reintended. To see what's going
on, ignore whitespace differences (e.g. diff -w).
I followed the following rules:
* Minimize the amount of code that gets moved.
* Don't change anything to what gets executed or displayed.
* Almost all the code must end up in a function.
* One function does one thing. For most of the code, that's from one
"cleanup" to the next.
* The test sequence functions (run_XXX) are independent.
The change mostly amounts to putting chunks of code into a function
and calling the functions in order. A few test runs are conditional;
in those cases the conditional is around the function call.
tests/Makefile had some unused warnings disabled unnecessarily, which
test-ref-configs.pl was turning back on. We don't need to disable these warnings
so I'm turning them back on.
Dependent on configured options, not all of the helper functions were being
used, which was leading to warning of unused functions with Clang.
To avoid any complex compile time options, or adding more logic to
generate_test_code.py to screen out unused functions, those functions which were
provoking the warning were changed to remove static, remove them from file
scope, and expose them to the linker.
