Context: During a handshake, the SSL/TLS handshake logic constructs
an instance of ::mbedtls_ssl_session representing the SSL session
being established. This structure contains information such as the
session's master secret, the peer certificate, or the session ticket
issues by the server (if applicable).
During a renegotiation, the new session is constructed aside the existing
one and destroys and replaces the latter only when the renegotiation is
complete. While conceptually clear, this means that during the renegotiation,
large pieces of information such as the peer's CRT or the session ticket
exist twice in memory, even though the original versions are removed
eventually.
This commit removes the simultaneous presence of two peer CRT chains
in memory during renegotiation, in the following way:
- Unlike in the case of SessionTickets handled in the previous commit,
we cannot simply free the peer's CRT chain from the previous handshake
before parsing the new one, as we need to verify that the peer's end-CRT
hasn't changed to mitigate the 'Triple Handshake Attack'.
- Instead, we perform a binary comparison of the original peer end-CRT
with the one presented during renegotiation, and if it succeeds, we
avoid re-parsing CRT by moving the corresponding CRT pointer from the
old to the new session structure.
- The remaining CRTs in the peer's chain are not affected by the triple
handshake attack protection, and for them we may employ the canonical
approach of freeing them before parsing the remainder of the new chain.
Note that this commit intends to not change any observable behavior
of the stack. In particular:
- The peer's CRT chain is still verified during renegotiation.
- The tail of the peer's CRT chain may change during renegotiation.
Context: During a handshake, the SSL/TLS handshake logic constructs
an instance of ::mbedtls_ssl_session representing the SSL session
being established. This structure contains information such as the
session's master secret, the peer certificate, or the session ticket
issues by the server (if applicable).
During a renegotiation, the new session is constructed aside the existing
one and destroys and replaces the latter only when the renegotiation is
complete. While conceptually clear, this means that during the renegotiation,
large pieces of information such as the peer's CRT or the session ticket
exist twice in memory, even though the original versions are removed
eventually.
This commit starts removing this memory inefficiency by freeing the old
session's SessionTicket before the one for the new session is allocated.
Context:
The existing API `mbedtls_x509_parse_crt_der()` for parsing DER
encoded X.509 CRTs unconditionally makes creates a copy of the
input buffer in RAM. While this comes at the benefit of easy use,
-- specifically: allowing the user to free or re-use the input
buffer right after the call -- it creates a significant memory
overhead, as the CRT is duplicated in memory (at least temporarily).
This might not be tolerable a resource constrained device.
As a remedy, this commit adds a new X.509 API call
`mbedtls_x509_parse_crt_der_nocopy()`
which has the same signature as `mbedtls_x509_parse_crt_der()`
and almost the same semantics, with one difference: The input
buffer must persist and be unmodified for the lifetime of the
established instance of `mbedtls_x509_crt`, that is, until
`mbedtls_x509_crt_free()` is called.
Resolve incompatibilties in the RSA module where changes made for
parameter validation prevent Mbed Crypto from working. Mbed Crypto
depends on being able to pass zero-length buffers that are NULL to RSA
encryption functions.
This reverts commit 2f660d047d.
Context: There are two public key writing functions in Mbed TLS. First,
mbedtls_pk_write_pubkey(), which exports a public key in the form of a
SubjectPublicKey structure containing the raw keying material
(for example, EC point coordinates for an EC public key, without
reference to the underlying curve). Secondly, mbedtls_pk_write_pubkey_der(),
which exports a public key in the form of a SubjectPublicKeyInfo structure,
wrapping the SubjectPublicKey structure by additional information
identifying the type of public key (and for ECC, e.g., it'd also contain
the ECC group identifier). The implementation of mbedtls_pk_write_pubkey_der()
calls mbedtls_pk_write_pubkey() first and then adds the corresponding
algorithm identifier wrapper.
Both of these functions need to be provided for PSA-based opaque PK contexts,
based on PSA's public key export function.
Previously, PSA used the SubjectPublicKeyInfo structure as its export format,
so mbedtls_pk_write_pubkey_der() could be easily implemented, while
mbedtls_pk_write_pubkey() would need to trim the output of the PSA export.
The previous implementation of mbedtls_pk_write_pubkey() is not quite right
because it calls PSA export doesn't do any trimming, hence exporting the large
SubjectPublicKeyInfo structure instead of the small SubjectPublicKey.
mbedtls_pk_write_pubkey_der(), in turn, immediately returns after calling
mbedtls_pk_write_pubkey(), hence also returning the SubjectPublicKeyInfo
structure, which is correct.
By now, the PSA public key export format has changed to the smaller
SubjectPublicKey structure. This means that, now, mbedtls_pk_write_pubkey()
can be implemented by just calling the PSA export, and that
mbedtls_pk_write_pubkey_der() needs to add the algorithm information around
it, just as in the other types of PK contexts. While not correct for the
old format, the existing code for mbedtls_pk_write_pubkey() is therefore
correct for the new PSA public key format, and needs no change apart from
the missing pointer shift in the last commit.
The implementation of mbedtls_pk_write_pubkey_der() needs a special code
path for PSA-based opaque PK contexts, as the PK context only contains
the PSA key handle, and the PSA API needs to be used to extract the
underlying EC curve to be able to write the AlgorithmParameter structure
that's part of the SubjectPublicKeyInfo structure.
That's what this commit does, (hopefully) making both
mbedtls_pk_write_pubkey() and mbedtls_pk_write_pubkey_der() export
the correctly formatted public key based on the new PSA public key format.
Additional changes to temporarily enable running tests:
ssl_srv.c and test_suite_ecdh use mbedtls_ecp_group_load instead of
mbedtls_ecdh_setup
test_suite_ctr_drbg uses mbedtls_ctr_drbg_update instead of
mbedtls_ctr_drbg_update_ret
Previously, PSA used SubjectPublicKeyInfo structures to serialize EC public keys.
This has recently been changed to using ECPoint structures instead, but the wrapper
making PSA ECDSA verification available through Mbed TLS' PK API hasn't yet been
adapted accordingly - which is what this commit does.
Luckily, Mbed TLS' PK API offers two functions mbedtls_pk_write_pubkey()
and mbedtls_pk_write_pubkey_der(), the latter exporting a SubjectPublicKeyInfo
structure and the former exporting an ECPoint structure in case of EC public
keys. For the adaptation of the ECDSA wrapper ecdsa_verify_wrap() it is therefore
sufficient to use mbedtls_pk_write_pubkey() instead of mbedtls_pk_write_pubkey_der().
Return the error code if failed, instead of returning value `1`.
If not failed, return the call of the underlying function,
in `mbedtls_ecdsa_genkey()`.
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.
mbedtls_mpi_read_binary() calls memcpy() with the source pointer being
the source pointer passed to mbedtls_mpi_read_binary(), the latter may
be NULL if the buffer length is 0 (and this happens e.g. in the ECJPAKE
test suite). The behavior of memcpy(), in contrast, is undefined when
called with NULL source buffer, even if the length of the copy operation
is 0.
This commit fixes this by explicitly checking that the source pointer is
not NULL before calling memcpy(), and skipping the call otherwise.
Context: The function `mbedtls_mpi_fill_random()` uses a temporary stack
buffer to hold the random data before reading it into the target MPI.
Problem: This is inefficient both computationally and memory-wise.
Memory-wise, it may lead to a stack overflow on constrained devices with
limited stack.
Fix: This commit introduces the following changes to get rid of the
temporary stack buffer entirely:
1. It modifies the call to the PRNG to output the random data directly
into the target MPI's data buffer.
This alone, however, constitutes a change of observable behaviour:
The previous implementation guaranteed to interpret the bytes emitted by
the PRNG in a big-endian fashion, while rerouting the PRNG output into the
target MPI's limb array leads to an interpretation that depends on the
endianness of the host machine.
As a remedy, the following change is applied, too:
2. Reorder the bytes emitted from the PRNG within the target MPI's
data buffer to ensure big-endian semantics.
Luckily, the byte reordering was already implemented as part of
`mbedtls_mpi_read_binary()`, so:
3. Extract bigendian-to-host byte reordering from
`mbedtls_mpi_read_binary()` to a separate internal function
`mpi_bigendian_to_host()` to be used by `mbedtls_mpi_read_binary()`
and `mbedtls_mpi_fill_random()`.
The calls to cipher_finish didn't actually do anything:
- the cipher mode is always ECB
- in that case cipher_finish() only sets *olen to zero, and returns either 0
or an error depending on whether there was pending data
- olen is a local variable in the caller, so setting it to zero right before
returning is not essential
- the return value of cipher_finis() was not checked by the caller so that's
not useful either
- the cipher layer does not have ALT implementations so the behaviour
described above is unconditional on ALT implementations (in particular,
cipher_finish() can't be useful to hardware as (with ECB) it doesn't call any
functions from lower-level modules that could release resources for example)
Since the calls are causing issues with parameter validation, and were no
serving any functional purpose, it's simpler to just remove them.
Somehow, mbedtls_sha256_ret() is defined even if MBEDTLS_SHA256_ALT
is set, and it is using SHA256_VALIDATE_RET. The documentation should
be enhanced to indicate that MBEDTLS_SHA256_ALT does _not_ replace
the entire module, but only the core SHA-256 functions.
Somehow, mbedtls_sha512_ret() is defined even if MBEDTLS_SHA512_ALT
is set, and it is using SHA512_VALIDATE_RET. The documentation should
be enhanced to indicate that MBEDTLS_SHA512_ALT does _not_ replace
the entire module, but only the core SHA-512 functions.
Somehow, mbedtls_sha1_ret() is defined even if MBEDTLS_SHA1_ALT
is set, and it is using SHA1_VALIDATE_RET. The documentation should
be enhanced to indicate that MBEDTLS_SHA1_ALT does _not_ replace
the entire module, but only the core SHA-1 functions.
Document when a context must be initialized or not, when it must be
set up or not, and whether it needs a private key or a public key will
do.
The implementation is sometimes more liberal than the documentation,
accepting a non-set-up context as a context that can't perform the
requested information. This preserves backward compatibility.
The MPI_VALIDATE_RET() macro cannot be used for parameter
validation of mbedtls_mpi_lsb() because this function returns
a size_t.
Use the underlying MBEDTLS_INTERNAL_VALIDATE_RET() insteaed,
returning 0 on failure.
Also, add a test for this behaviour.
