Change the name of these confs to be version agnostic,
we will later use these configs to enforce the mbedtls
minimum version
Change-Id: I1f665c2471877ecc833270c511749ff845046f10
Signed-off-by: Ryan Everett <ryan.everett@arm.com>
Make this function return values from crypto_ret_value.
The previous method of returning the mbedtls error code
on failure meant that the authentication module couldn't
correctly parse failures from this function.
Change-Id: I9fe6eba1fc79e8f81004f8cd202781aea907e963
Signed-off-by: Ryan Everett <ryan.everett@arm.com>
The buffer size allocated for the raw ECDSA signatures
was too small for P-384 signatures. This led to mbedtls
aborting the generation of the raw sig.
Fix this by increasing the buffer size to the required
value.
Change-Id: I06a9cfe1f4cb2603c5fbe945714e90460c24edb8
Signed-off-by: Ryan Everett <ryan.everett@arm.com>
This new version uses the multipart PSA AEAD API;
the authentication tag is verified via
a call to psa_aead_verify.
Change-Id: If4b7e6258223ae6fead1794d3e8d0004f0f387b3
Signed-off-by: Ryan Everett <ryan.everett@arm.com>
MbedTLS 3.6.1 fixed the issue which previously
produced this warning, so this hack is no longer
necessary.
Change-Id: I934adefbf2fed16e16b9d98bc8674125b70b08fc
Signed-off-by: Ryan Everett <ryan.everett@arm.com>
This new version uses fewer internal functions
in favour of calling equivalent mbedtls APIs.
Change-Id: I0c2c20a74687211f2d554501f57898da07b01739
Signed-off-by: Ryan Everett <ryan.everett@arm.com>
When the verify_signature function is called with the RSASSA_PSS
signature algorithm and a somewhat well-formed public key, invalid
signatures can be incorrectly verified due to this change [1].
This is primarily because of the introduction of the following code,
where a return check is missing before the goto:
if (pk_alg == MBEDTLS_PK_RSASSA_PSS) {
rc = pk_bytes_from_subpubkey((unsigned char **) &pk_ptr, &pk_len);
goto end2;
}
This code executes before the call to psa_verify_message. The
unconditional goto end2; branch leads to the immediate return of rc.
If the call to pk_bytes_from_subpubkey succeeds (i.e., the key is
formatted correctly), the signature is verified regardless of its
actual content.
This change [1] was included in the v2.11 release. Therefore, anyone
using this release with the PSA Crypto implementation must apply
this patch to ensure proper signature verification.
[1]: https://review.trustedfirmware.org/plugins/gitiles/TF-A/
trusted-firmware-a/+/55aed7d798f3d48d6aa08d58eb46c4cda318bcfb/drivers/
auth/mbedtls/mbedtls_psa_crypto.c#447
Signed-off-by: Manish V Badarkhe <Manish.Badarkhe@arm.com>
Reported-by: Ryan Everett <ryan.everett@arm.com>
Change-Id: Ib484d97a04b7a82dd72592c8b5b153d577d01fc9
Further, remove reliance of mbedtls_md_psa_alg_from_type on
the actual values of the PSA_ALG_... defines.
And work around a prior bug that would try to import a
SubjectPublicKeyInfo into a PSA key. Instead, we import the
SubjectPublicKey itself.
Change-Id: Ib345b0bd4f2994f366629ed162d18814fd05aa2b
Signed-off-by: Jimmy Brisson <jimmy.brisson@arm.com>
Deprecation notice was sent to the community and no objection was
raised, so removing mbedtls 2.x support.
Change-Id: Id3eb98b55692df98aabe6a7c5a5ec910222c8abd
Signed-off-by: Lauren Wehrmeister <lauren.wehrmeister@arm.com>
The ECDSA algorithm signature verification in the PSA differs
from the RSA algorithm in its handling of data formats. In the
case of RSA, an encoded ASN1.0 buffer is passed to the PSA API,
which then decodes the buffer. However, for ECDSA, the PSA API
expects a raw format.
To accomodate this requirement, introduce several static APIs
that allows to retrieve -
1. ECDSA public key data pointer along with its size, and also,
the ECC family in PSA format from the public key.
2. R and S pair of the ECDSA signature along with its size
Change-Id: Icc7d5659aeb3d5c1ab63c3a12c001e68b11a3a86
Signed-off-by: Manish V Badarkhe <Manish.Badarkhe@arm.com>
Use KEY_SIZE 384 to enable ECDSA P384 key support by
setting MBEDTLS_ECP_DP_SECP384R1_ENABLED.
Selected by setting KEY_ALG=ecdsa and KEY_SIZE=384.
Change-Id: I382f34fc4da98f166a2aada5d16fdf44632b47f5
Signed-off-by: Lauren Wehrmeister <lauren.wehrmeister@arm.com>
A PSA crypto API is used for signature verification, namely
psa_verify_message, which calculates the signature and verify it
for the given signed message.
Note, this functionality is as of now tested for RSA based Key
algorithm, subsequent patches will provide the signature verification
support for all the key algorithms available in TF-A (provided by KEY_ALG
build option).
Signed-off-by: Manish V Badarkhe <Manish.Badarkhe@arm.com>
Change-Id: I16214e3cac6a7af445530a6dbd66cfc6ceedcfc6
A PSA crypto API is used for hash verification, namely
'psa_hash_compute', which internally calculates hash of the given
data payload.
Signed-off-by: Manish V Badarkhe <Manish.Badarkhe@arm.com>
Change-Id: Ic90e43e68b836ee2add4b868663cfca2340c8108
A PSA crypto API is used for hash verification, namely
psa_hash_verify, which internally calculates and verifies the hash
of the given data payload.
Signed-off-by: Manish V Badarkhe <Manish.Badarkhe@arm.com>
Change-Id: Ib48aa6b74b59aea6036333ff00f6ca566b910e60
Initialised Mbedtls PSA cryto during Crypto init using
function call 'psa_crypto_init'.
MbedTLS currently requires a Random Number Generator (RNG) once
PSA Crypto support is enabled. However, TF-A itself doesn't engage
in cryptographic operations that demand randomness. Consequently,
we simulate the presence of an external TRNG (through the configuration
option 'MBEDTLS_PSA_CRYPTO_EXTERNAL_RNG) while, in reality, we offer
a dummy implementation of mbedtls_psa_external_get_random() that always
returns an error.
Change-Id: Ife6d03909c0e6081438d2b2519ef500e5dcdb88f
Signed-off-by: Manish V Badarkhe <Manish.Badarkhe@arm.com>
An ad-hoc MbedTLS PSA crypto driver is registered by compiling
a new driver file, namely mbedtls_psa_crypto.c when PSA_CRYPTO=1.
As of now, this file is the same as mbedtls_crypto.c, but subsequent
patches will update crypto functions in this file to trigger
PSA crypto APIs.
Change-Id: I404c347990661d87dcf5d0501d238e36914ec3ee
Signed-off-by: Manish V Badarkhe <Manish.Badarkhe@arm.com>
This is a preparatory patch to provide MbedTLS PSA Crypto
API support, with below changes -
1. Added a build macro PSA_CRYPTO to enable the MbedTLS PSA
Crypto API support in the subsequent patches.
2. Compile necessary PSA crypto files from MbedTLS source code
when PSA_CRYPTO=1.
Also, marked PSA_CRYPTO as an experimental feature.
Change-Id: I45188f56c5c98b169b2e21e365150b1825c6c450
Signed-off-by: Manish V Badarkhe <Manish.Badarkhe@arm.com>
Trusted Board Boot supports multiple hash algorithms, including SHA-256,
SHA-384, and SHA-512. These algorithms produce hashes of different
lengths, so the resulting DER-encoded hash objects are also of different
lengths. However, the common Trusted Board Boot code only stores the
contents of the object, not its length. Before commit
f47547b354, this was harmless: ASN.1
objects are self-delimiting, and any excess padding was ignored.
f47547b354 changed the code to reject
excess padding. However, this breaks using a shorter hash in a build
that supports longer hashes: the shorter hash will have padding after
it, and verify_hash() will reject it. This was found by an Arm
customer: TF-A v2.9 refused to boot, even though TF-A v2.6 (which did
not have f47547b354) worked just fine.
Storing the length of the hash turns out to be quite difficult.
However, it turns out that hashes verified by verify_hash() always come
from the ROTPK or an X.509 certificate extension. Furthermore, _all_
X.509 certificate extensions used by Trusted Board Boot are ASN.1
DER encoded, so it is possible to reject padding in get_ext(). Padding
after the ROTPK is harmless, and it is better to ignore that padding
than to refuse to boot the system.
Change-Id: I28a19d7783e6036b65e86426d78c8e5b2ed6f542
Fixes: f47547b354 ("fix(auth): reject invalid padding in digests")
Signed-off-by: Demi Marie Obenour <demiobenour@gmail.com>
Have only one definition for REGISTER_CRYPTO_LIB macro, with all the
possible fields. Worst case adds 4 u64 to crypto_lib_desc.
While at it, correct some MISRA violations:
MC3R1.R12.1: (advisory) The precedence of operators within expressions
should be made explicit.
Signed-off-by: Yann Gautier <yann.gautier@st.com>
Change-Id: I1342a20e6eef2354753182c2a81ff959e03e5c81
Following discussions in the reviews of the patch that introduced
plat_convert_pk() function [1], it was decided to deprecate it to
avoid weak function declaration.
A new optional function pointer convert_pk is added to crypto_lib_desc_t.
A new function crypto_mod_convert_pk() will either call
crypto_lib_desc.convert_pk() if it is defined, or do the same
as what was done by the weak function otherwise.
[1] https://review.trustedfirmware.org/c/TF-A/trusted-firmware-a/+/17174
Signed-off-by: Yann Gautier <yann.gautier@foss.st.com>
Change-Id: I9358867f8bfd5e96b5ee238c066877da368e43c6
Previously, extensions were parsed twice: once with error checking for
validation, and a second time without error checking to extract the
extension data. This is error prone and caused TFV-10 (CVE-2022-47630).
A simpler approach is to have get_ext() be responsible for all extension
parsing, and to treat a NULL OID as an indicator that get_ext() is only
being called for validation. cert_parse() checks that get_ext() returns
IMG_PARSER_OK and fails otherwise.
Change-Id: I65a2ff053a188351ba54799827a2b7bd833bb037
Signed-off-by: Demi Marie Obenour <demiobenour@gmail.com>
The previous code was correct but unnecessarily verbose.
Change-Id: Ia19c667811a7c3b6957a0274d36076b0b16e36b7
Signed-off-by: Demi Marie Obenour <demiobenour@gmail.com>
The unique IDs are handled identically, so just use a for loop to get
both of them.
Change-Id: I44baaa4747ca7f314d364a79dfcbce97315f5a92
Signed-off-by: Demi Marie Obenour <demiobenour@gmail.com>
TF-A support for mbedtls3.x has been overdue by number of releases.
As per mbedtls support it was advised to use latest and greatest
mbedtls-3.3. But mbedtls-3.x breaks API compatibility with
mbedtls-2.x
To maintain comptability for mbedtls-2.x and enable mbedtls-3.x
support add a functionality into makefile to determine the major version
of mbedtls and use that to selective include or compile files
that are present.
With mbedtls-3.x numerous other config changes have been done.
Some of the config options deprecated or enabled by default.
Thus we decided to introduce a new 3.x config file part of this
change for building TF-A with mbedtls-3.3.
For futher information on migrating to mbedtls 3.x refer to:
https://github.com/Mbed-TLS/mbedtls/blob/development/docs/3.0-migration-guide.md
Change-Id: Ia8106d6f526809df927d608db27fe149623258ed
Signed-off-by: Govindraj Raja <govindraj.raja@arm.com>
Currently we include MBEDTLS_CONFIG_FILE directly and if a custom
config file is used it will included.
However from mbedtls-3.x onwards it discourages usage of
MBEDTLS_CONFIG_FILE include directly, so to resolve this and keep 2.28
compatibility include version.h which would include the custom config
file if present and also would expose us with mbedtls-major-version
number which could be used for selecting features and functions for
mbedtls 2.28 or 3.3
Change-Id: I029992311be2a38b588ebbb350875b03ea29acdb
Signed-off-by: Govindraj Raja <govindraj.raja@arm.com>
* changes:
fix(libc): remove __putchar alias
fix(console): correct scopes for console symbols
fix(auth): use NULL instead of 0 for pointer check
fix(io): compare function pointers with NULL
fix(fdt-wrappers): use correct prototypes
This was triggered by sparse tool:
drivers/auth/mbedtls/mbedtls_x509_parser.c:481:42: warning:
Using plain integer as NULL pointer
Signed-off-by: Yann Gautier <yann.gautier@st.com>
Change-Id: I392316c2a81ef8da7597e35f136e038f152d19d1
get_ext() does not check the return value of the various mbedtls_*
functions, as cert_parse() is assumed to have guaranteed that they will
always succeed. However, it passes the end of an extension as the end
pointer to these functions, whereas cert_parse() passes the end of the
TBSCertificate. Furthermore, cert_parse() does *not* check that the
contents of the extension have the same length as the extension itself.
Before fd37982a19 ("fix(auth): forbid junk after extensions"),
cert_parse() also does not check that the extension block extends to the
end of the TBSCertificate.
This is a problem, as mbedtls_asn1_get_tag() leaves *p and *len
undefined on failure. In practice, this results in get_ext() continuing
to parse at different offsets than were used (and validated) by
cert_parse(), which means that the in-bounds guarantee provided by
cert_parse() no longer holds.
This patch fixes the remaining flaw by enforcing that the contents of an
extension are the same length as the extension itself.
Change-Id: Id4570f911402e34d5d6c799ae01a01f184c68d7c
Signed-off-by: Demi Marie Obenour <demiobenour@gmail.com>
Signed-off-by: Sandrine Bailleux <sandrine.bailleux@arm.com>
This makes the code a little bit smaller. No functional change
intended.
Change-Id: I794d2927fcd034a79e29c9bba1f8e4410203f547
Signed-off-by: Demi Marie Obenour <demiobenour@gmail.com>
Certificates must not allow trailing junk after them.
Change-Id: Ie33205fb051fc63af5b72c326822da7f62eec1d1
Signed-off-by: Demi Marie Obenour <demiobenour@gmail.com>
This is already checked by the crypto module or by mbedTLS, but checking
it in the X.509 parser is harmless.
Change-Id: Ifdbe3b4c6d04481bb8e93106ee04b49a70f50d5d
Signed-off-by: Demi Marie Obenour <demiobenour@gmail.com>
Since the two instances of the signature algorithm in a certificate must
be bitwise identical, it is not necessary to parse both of them.
Instead, it suffices to parse one of them, and then check that the other
fits in the remaining buffer space and is equal to the first.
Change-Id: Id0a0663165f147879ac83b6a540378fd4873b0dd
Signed-off-by: Demi Marie Obenour <demiobenour@gmail.com>
This reduces the likelihood of future problems later.
Change-Id: Ia748b6ae31a7a48f17ec7f0fc08310a50cd1b135
Signed-off-by: Demi Marie Obenour <demiobenour@gmail.com>
Digests must not have padding after the SEQUENCE or OCTET STRING.
Change-Id: Id25ab23111781f8c8a97c2c3c8edf1cc4a4384c0
Signed-off-by: Demi Marie Obenour <demiobenour@gmail.com>
X.509 and RFC5280 allow omitting the extensions entirely, but require
that if the extensions field is present at all, it must contain at least
one certificate. TF-A already requires the extensions to be present,
but allows them to be empty. However, a certificate with an empty
extensions field will always fail later on, as the extensions contain
the information needed to validate the next stage in the boot chain.
Therefore, it is simpler to require the extension field to be present
and contain at least one extension. Also add a comment explaining why
the extensions field is required, even though it is OPTIONAL in the
ASN.1 syntax.
Change-Id: Ie26eed8a7924bf50937a6b27ccdf7cc9a390588d
Signed-off-by: Demi Marie Obenour <demiobenour@gmail.com>
The extensions must use all remaining bytes in the TBSCertificate.
Change-Id: Idf48f7168e146d050ba62dbc732638946fcd6c92
Signed-off-by: Demi Marie Obenour <demiobenour@gmail.com>
v1 and v2 are forbidden as at least one extension is required. Instead
of actually parsing the version number, just compare it with a
hard-coded string.
Change-Id: Ib8fd34304a0049787db77ec8c2359d0930cd4ba1
Signed-off-by: Demi Marie Obenour <demiobenour@gmail.com>
The platform.mk can add extra MbedTLS source files to LIBMBEDTLS_SRC.
Change-Id: Ida9abfd59d8b02eae23ec0a7f326db060b42bf49
Signed-off-by: Mate Toth-Pal <mate.toth-pal@arm.com>
Correct the warning due to comparison between signed and
unsigned variable.
drivers/auth/mbedtls/mbedtls_x509_parser.c: In function 'get_ext':
drivers/auth/mbedtls/mbedtls_x509_parser.c:120:30:
error: comparison of integer expressions of different
signedness: 'int' and 'size_t' {aka 'unsigned int'}
[-Werror=sign-compare]
120 | if ((oid_len == strlen(oid_str)) && !strcmp(oid, oid_str)) {
| ^~
Change-Id: Ic12527f5f92a34e925bee3047c168eacf5e99d8a
Signed-off-by: Nicolas Toromanoff <nicolas.toromanoff@st.com>
Updated mbedTLS driver to include mbedTLS functions necessary for a
DRTM supported build.
Signed-off-by: Manish V Badarkhe <manish.badarkhe@arm.com>
Change-Id: If0120374a971519cf84f93e0c59e1a320a72cd97
With RSS now introduced, we have 2 Measured Boot backends. Both backends
can be used in the same firmware build with potentially different hash
algorithms, so now there can be more than one hash algorithm in a build.
Therefore the logic for selecting the measured boot hash algorithm needs
to be updated and the coordination of algorithm selection added. This is
done by:
- Adding MBOOT_EL_HASH_ALG for Event Log to define the hash algorithm
to replace TPM_HASH_ALG, removing reference to TPM.
- Adding MBOOT_RSS_HASH_ALG for RSS to define the hash algorithm to
replace TPM_HASH_ALG.
- Coordinating MBOOT_EL_HASH_ALG and MBOOT_RSS_HASH_ALG to define the
Measured Boot configuration macros through defining
TF_MBEDTLS_MBOOT_USE_SHA512 to pull in SHA-512 support if either
backend requires a stronger algorithm than SHA-256.
Signed-off-by: Lauren Wehrmeister <lauren.wehrmeister@arm.com>
Change-Id: I4ddf06ebdc3835beb4d1b6c7bab5a257ffc5c71a
Upgrade to the latest and greatest 2.x release of Mbed TLS library
(i.e. v2.28.0) to take advantage of their bug fixes.
Note that the Mbed TLS project published version 3.x some time
ago. However, as this is a major release with API breakages, upgrading
to 3.x might require some more involved changes in TF-A, which we are
not ready to do. We shall upgrade to mbed TLS 3.x after the v2.7
release of TF-A.
Actually, the upgrade this time simply boils down to including the new
source code module 'constant_time.c' into the firmware.
To quote mbed TLS v2.28.0 release notes [1]:
The mbedcrypto library includes a new source code module
constant_time.c, containing various functions meant to resist timing
side channel attacks. This module does not have a separate
configuration option, and functions from this module will be
included in the build as required.
As a matter of fact, if one is attempting to link TF-A against mbed
TLS v2.28.0 without the present patch, one gets some linker errors
due to missing symbols from this new module.
Apart from this, none of the items listed in mbed TLS release
notes [1] directly affect TF-A. Special note on the following one:
Fix a bug in mbedtls_gcm_starts() when the bit length of the iv
exceeds 2^32.
In TF-A, we do use mbedtls_gcm_starts() when the firmware decryption
feature is enabled with AES-GCM as the authenticated decryption
algorithm (DECRYPTION_SUPPORT=aes_gcm). However, the iv_len variable
which gets passed to mbedtls_gcm_starts() is an unsigned int, i.e. a
32-bit value which by definition is always less than 2**32. Therefore,
we are immune to this bug.
With this upgrade, the size of BL1 and BL2 binaries does not appear to
change on a standard sample test build (with trusted boot and measured
boot enabled).
[1] https://github.com/Mbed-TLS/mbedtls/releases/tag/v2.28.0
Change-Id: Icd5dbf527395e9e22c8fd6b77427188bd7237fd6
Signed-off-by: Sandrine Bailleux <sandrine.bailleux@arm.com>
Common mbedTLS implementation include the fixed configuration
file of mbedTLS and that does not gives flexilibility to the
platform to include their own mbedTLS configuration.
Hence changes are done so that platform can include their own
mbedTLS configuration file.
Signed-off-by: Lucian Paul-Trifu <lucian.paul-trifu@arm.com>
Signed-off-by: Manish V Badarkhe <manish.badarkhe@arm.com>
Change-Id: I04546589f67299e26b0a6a6e151cdf1fdb302607
