Implementations of ASN.1 APIs reworked.
Added tests.
Added wolfssl_bn_set_neg for ASN.1 code.
Added wolfssl_sk_new_type() and wolfssl_sk_pop_type() to generically
handle creating and popping a stack of elements of a type.
No longer freeing pathlen field of ASN1 OBJECT in
wolfSSL_X509_EXTENSION_free(). This is happening in
wolfSSL_ASN1_OBJECT_free().
Stop wolfSSL_i2d_X509_NAME_canon from double freeing ASN.1 STRING's data
field.
Fixed up GetFormattedTime() to be better code.
Added ASN_CLASS_MASK to mask off the class part of an ASN.1 tag.
NO_ASN_TIME means no implementation to get the current time. Disable
features that won'r work without time.
* Add introspection for math build.
* Raise build error if more than one multi-precision math library used.
* Fix ESP32 to support using any multi-precision math option.
* Refactor math headers to use `wolfmath.h`
* Refactor of the opaque math variable type `MATH_INT_T` used by crypto hardware (QuickAssist, SE050, ESP32 and STM32).
* Cleanups for building with `WOLFCRYPT_ONLY` and `NO_BIG_INT`.
* Stop forcing use of fast math by default for platforms in settings.h. Note: For users that still want to use fast math (tfm.c) they will need to add USE_FAST_MATH to their build settings.
Applies To:
```
WOLFSSL_ESPWROOM32
WOLFSSL_ESPWROOM32SE
MICROCHIP_PIC32
WOLFSSL_PICOTCP_DEMO
WOLFSSL_UTASKER
WOLFSSL_NRF5x
FREERTOS_TCP
WOLFSSL_TIRTOS
EBSNET
FREESCALE_COMMON
FREESCALE_KSDK_BM
WOLFSSL_DEOS
MICRIUM
WOLFSSL_SGX
```
src/internal.c: fix for -Wdeclaration-after-statement and clang-diagnostic-unreachable-code-break;
tests/api.c: fix for -Wunused-variable and clang-analyzer-deadcode.DeadStores;
olfcrypt/src/pkcs12.c: fixes for cppcheck uselessAssignmentPtrArg and arrayIndexThenCheck, and clang-tidy clang-analyzer-deadcode.DeadStores and clang-analyzer-core.NonNullParamChecker;
wolfssl/src/tls.c: fix for clang-analyzer-deadcode.DeadStores;
wolfcrypt/src/tfm.c: fix for clang-diagnostic-newline-eof;
src/tls13.c: fix for clang-analyzer-core.NonNullParamChecker.
HAVE_ECH only used by TLS 1.3 add protection around all code.
ssl->options.onlyPskDheKe only available when HAVE_SUPPORTED_CURVES.
CleanupClientTickets() defined when HAVE_SUPPORTED_CURVES.
TLSX_KeyShare_DeriveSecret only defined when HAVE_SUPPORTED_CURVES.
DecodeResponseData - initialize variable single.
New OpenSSL compatibility BN code requires mp_read_radix - turn on in
integer.c, sp_int.c when OPENSSL_EXTRA defined.
rsa.c:_CheckProbablePrime - make sure tmp1 and tmp2 are initialized
before error handling jumps to freeing them.
test_remove_hs_message uses 1024-bit DH key which is not supported when
using SP math with SP.
Long server IDs were being truncated. Hash long IDs instead.
TLS 1.3 session ticket on client side no longer added session to client
cache. Explicit call added.
BN APIs from ssl.c have been moved out to ssl_bn.c that is included in
ssl.c.
Added defines for BN_rand() and BN_pseudo_rand() to indicate which bits
are to be set.
'internal' field now always maps to the ;mpi' field that is a MP
integer.
SetIndividualInternal/External renamed to wolfssl_bn_get/set_value.
Fixed BN APIs to work as closely to OpenSSL as possible.
Added tests.
Moved wolfssl_make_rng out to ssl.c as BN APIs are using it now.
SP int and TFM now check trials are in a valid range for
mp_prime_is_prime_ex().
configure.ac: remove implied --disable-examples when --disable-inline;
tests/api.c: always include wolfcrypt/src/misc.c;
wolfcrypt/src/misc.c: use WC_MISC_STATIC to designate staticness of the misc functions;
wolfssl/wolfcrypt/misc.h: define WC_MISC_STATIC to empty or static depending on #ifdef NO_INLINE or not (respectively);
wolfssl/wolfcrypt/types.h: #ifdef NO_INLINE, #define WC_INLINE WC_MAYBE_UNUSED, and move the definitions of WARN_UNUSED_RESULT and WC_MAYBE_UNUSED above the definition of WC_INLINE for clarity;
wolfssl/wolfcrypt/mem_track.h: always declare TrackMalloc() et al static (no functional change, because the state and callback bindings were and remain unconditionally static, but previously #ifdef NO_INLINE the functions were globals, leading to multiple definitions.
* stunnel 5.67 support
- ssl->session->ticketAdd may be set when parsing a ticket before PSK
- PSK binders should be calculated based on ciphersuite associated with PSK
- Add option to prioritise PSK order instead of ciphersuite order
- Update ctx->method->version when using Set_CTX_max_proto_version API
- Simplify wolfSSL_parse_cipher_list
- Keep copy of old list and then add in the previous ciphersuites depending on whether we are doing only TLS 1.3 ciphersuites or not
- Specify CRL revocation in alert
- Match reason string to match OpenSSL
- Add support for external data callbacks for WOLFSSL_SESSION
* Upref the session for stunnel instead of duplicating it
* Add small stack option for wolfSSL_parse_cipher_list
wolfSSL_EC_POINT_set_affine_coordinates_GFp wasn't checking the point is
valid for the curve. Added call to check point when setting.
Made check available for opensslextra.
Fixed test to have valid ordinates to set.
wolfSSL_EC_POINT_set_affine_coordinates_GFp wasn't checking the point is
valid for the curve. Added call to check point when setting.
Made check available for opensslextra.
Fixed test to have valid ordinates to set.
used and size have no reason to be negative - change type.
Change code to match unsigned change. Mostly change variables to be
unsigned where possible.
integer.c: Only have mp_rand_prime available when needed and
mp_prime_is_prime_ex is available.
Fixes from regression testing.
fix file descriptor leaks in AF_ALG code, and fix return codes (WC_AFALG_SOCK_E, not -1) in afalg_aes.c;
fixes for sanitizer-detected forbidden null pointer args in AfalgHashUpdate() and AfalgHashCopy();
fixes for resource leaks in api.c test_wolfSSL_AES_cbc_encrypt() (missing wc_AesFree()s);
fixes for resource leaks in test.c openssl_test() (missing wolfSSL_EVP_CIPHER_CTX_cleanup());
also some local fixes for bugprone-signed-char-misuse, readability-redundant-preprocessor, and clang-diagnostic-strict-prototypes, in src/pk.c and src/ssl.c.
Reworked the implementations of the EC APIs including:
wolfSSL_EC_curve, wolfSSL_EC_METHOD, wolfSSL_EC_GROUP,
wolfSSL_EC_POINT, wolfSSL_EC_KEY, wolfSSL_ECDSA_SIG, wolfSSL_ECDSA and
wolfSSL_ECDH.
Added support for EC parameters to PEM decoding.
EccEnumToNID() moved out of wolfCrypt - it maps NIDs defined in
wolfssl/openssl/ec.h to those in wolfssl/wolfcrypt/ecc.h.
Moved wc_OBJ_sn2nid() out of wolfCrypt - implementation uses
EccEnumToNID().
Changed reference counding to use wolfSSL_Ref.
Added tests to api.c that increase coverage of EC APIs.
- Allocate ssl->suites when necessary for the WOLFSSL object to have its own instance. Use AllocateSuites() to allocate the object.
- Move cipher negotiation options from Suites into Options
ZD15346
* Fixed `PEM_BUFSIZE` macro redefined when building with coexist.
* Updated the `user_settings_all.h` and `user_settings_wolfboot_keytools.h` to include latest options.
* Improved API unit test error case checking where `TEST_RES_CHECK` is not used.
* Changed `TEST_SKIPPED` to unique value.
* Added CI tests for enable-all, small stack, and user setting templates.
This method requires some explanation. Its sibling is
int SetSSL_CTX(WOLFSSL* ssl, WOLFSSL_CTX* ctx, int writeDup)
which re-inits the WOLFSSL* with all settings in the new CTX.
That one is the right one to use *before* a handshake is started.
This method was added by OpenSSL to be used *during* the handshake, e.g.
when a server inspects the SNI in a ClientHello callback and
decides which set of certificates to use.
Since, at the time the SNI callback is run, some decisions on
Extensions or the ServerHello might already have been taken, this
method is very restricted in what it does:
- changing the server certificate(s)
- changing the server id for session handling
and everything else in WOLFSSL* needs to remain untouched.
Configuration
- thread-local storaoge is selected when available
- '--enable-error-queue-per-thread' and '--disable-error-queue-per-thread' can
be used as before to explicitly en-/disable the feature.
Implementation:
- with thread-local-storage, error queue is realized in one struct without
allocations. Queue size is restricted to 16 entries (per thread), which
is the same limit in OpenSSL 1.1.x.
- without thread-local-storage, all error queue operations are mutex locked
- wc_PeekErrorNodeLineData() and wc_GetErrorNodeErr() added for use by SSL
functions to allow locked queue iterations/manipulations.
Change API test cases to return a result.
Test success is now TEST_SUCCESS (1).
Test result can be returned with use of macro TEST_RES_CHECK().
Always print the id, name of the test and the result (skipped or
otherwise) before and after running the test case.
Changed test case output to go to stderr.
Fixed some formatting.
Add option to take index and/or name of test case to run.
Added option to list all API tests.
Added option to only run API tests.
Added options to show usage.
This fixes hash based dir lookup of ASN1/DER CRLs in OpenSSL
compatible API. The function wolfSSL_X509_load_crl_file is
called with entry->dir_type, rather than hardcoded filetype.
A new test crl was added, and existing crl 0fdb2da4.r0 was
reorganized to a new dir.
Also, completes the stub wolfSSL_X509_LOOKUP_add_dir. A new
test function test_X509_LOOKUP_add_dir was added to tests/api.c
clang-analyzer-deadcode.DeadStores in examples/server/server.c;
-Werror=use-after-free and LeakSanitizer Direct leak in tests/api.c;
nullPointerRedundantCheck in src/pk.c which identified a semantically consequential flub.
Ed25519 and Ed448 need to enable certs.
If no system CA certs can't be parsed,
wolfSSL_CTX_load_system_CA_certs() will fail. Don't try test if RSA and
ECC are not enabled.
Fix benchmark.c so that e is defined when WOLFSSL_BENCHMARK_ALL defined.
MAX_LENGTH_SZ is 4 and supports lengths up to 2^24 - one byte for length
and 3 bytes of length.
(new gcc compiler fix)
Reorganized the DH APIs into groups.
Reworked all DH APIs.
Improved testing of DH API.
Implemented wolfSSL_PEM_read_RSAPublicKey() and
wolfSSL_PEM_write_RSA_PUBKEY().
Fix public key PEM write implementations to use the correct
header/footer names.
Added support for "RSA PUBLIC KEY" in header and footer in DerToPemEx().
Reworked PEM read/write APIs to be independent. No longer create an EVP
to use common function - common functionality refectored out.
Similarly file APIs don't create a BIO and call the partner APIs.
Improved testing of PEM read/write APIs.
Generic read BIO from memory BIO now returns the buffer instead of
allocating memory and reading.
No longer reading chunks when a file BIO.
Added wolfssl_make_rng() to create or get get global random. All RSA and
DH APIs now use this. DH_generate_parameters() creates a random object
and use global on error rather than just using global random.
Changed implementations to use BIO_new_fp() instead of create a new BIO
and setting file pointer.
* Fix api.c build error with `NO_FILESYSTEM` and `WOLFSSL_CERT_EXT`.
* Fix for building tests/suites.c with static memory (missing `LARGEST_MEM_BUCKET`).
* Always expose `wc_ecc_set_rng` for compatibility.
This new function, wolfSSL_CTX_load_system_CA_certs, currently only supports
Linux-based OS's. It searches through conventional CA directories and once it
finds one, attempts to load CA certs from it. After the first directory is
found, we don't check the others.
This commit also adds a function wolfSSL_get_system_CA_dirs, which returns a
pointer to an array of directories where wolfSSL_CTX_load_system_CA_certs will
look for CA certs. This is used in a unit test, where we only want to expect
success if one of these directories actually exists on the test system.
Finally, this commit adds support for SSL_CTX_set_default_verify_paths to the
compatibility layer. It doesn't model the exact behavior of its OpenSSL
counterpart; it's mostly a wrapper around wolfSSL_CTX_load_system_CA_certs,
manipulating the return value of that function to conform to OpenSSL's
conventions.
- Mark old epochs as invalid so we don't attempt to decrypt with them
- Return a non-zero value if possible in unit tests
- Move Dtls13CheckAEADFailLimit to dtls13.c
- Reset state in processreply
wolfSSL_clear: check return from InitSSL_Suites() call.
TLS13: check ClientHello cipher suite length is even.
Silently remove duplicate cipher suites from user input.
Add tests of duplicate cipher suite removal.
Handle multiple instances of the same cipher suite being in the server's
list.
Fix client order negotiation of cipher suite when doing pre-shared keys.
- Use WOLF_SK_COMPARE_CB() to generate the correct types instead of using void* for the callback parameters.
- Remove WOLFSSL_STACK.comp entirely since it is not used anywhere. Ignore input parameters that used to set this member.
This function would only accept the string "SHA" for SHA-1-based HMAC, but it
should also accept "SHA1." This is similar to how wolfSSL_EVP_get_digestbyname
allows both "SHA" and "SHA1." We didn't have a test for this in api.c. I added
one, and it failed before my fix here.
Don't reserve the full message length of memory. Instead we only allocate memory for the fragments that we have already received. We also dynamically combine memory fragments when we receive overlap.
- wire the wc_ChaCha_* implementation into the EVP API
as `wolfSSL_EVP_chacha20`
- follow IV conversions of OpenSSL
- add test case
- have QUIC support use this for header protection when
CHACHA20_POLY1305 has been negotiated in the handshake.
Ed25519 and Ed448 make public key APIs now check whether the private key
was set.
The private key and public key flags setting and clearing also updated.
Testing of unset private key and calling make public key added for
Ed25519 and Ed448.
Detect when certificate's RSA public key size is too big and fail on
loading of certificate.
Fix unit test to only attempt to use 3072 bit RSA-PSS keys when
RSA_MAX_SIZE supports it.
* Add test case for OpenSSLs capability to init a evp context partially in several calls.
* EVP handling of CHACHA20_POLY1305 improvment
- save key at ctx for Init()s without IV
- reuse stored key for Init()s with new IV, reusing ctx
- free and zero key on ctx clenaup
* Adding type cast to XMALLOC to force compiler compatibility.
* EVP: using same DYNAMIC_TYPE in alloc and free of chacha20_poly1305 key.
Co-authored-by: Stefan Eissing <stefan.eissing@greenbytes.de>
in api.c test_wolfSSL_dtls12_fragments_spammer(), inhibit clang-analyzer-deadcode.DeadStores;
in asn.c ParseCRL_Extensions() old (!WOLFSSL_ASN_TEMPLATE) version, fix bounds check to prevent overshift;
in misc.c ctMaskCopy(), use `*(x + i)`, not x[i], to tiptoe around cppcheck-2.9 objectIndex bug.
This function was just freeing the stack object itself of GENERAL_NAMES with
wolfSSL_sk_free, but this doesn't free the data in the items of the stack. The
fix is to replace wolfSSL_sk_free with wolfSSL_sk_GENERAL_NAME_free.
1. In api.c, change the name of the devId global to testDevId.
2. In api.c, for the SNI test, some of the pointers were multiply
typecast eventually to `void **`. An older compiler didn't like
that. Just use void pointer.
OpenSSL compat expects ASN_SELF_SIGNED_E when a self signed cert can't be verified. This is useful when translating the error with GetX509Error into a X509_V_ERR_* error.
Support wolfSSL_CTX_set1_curves_list being available when X25519 and/or
X448 only defined.
Don't recognize X25519 or X448 if corresponding define is not set.
Enable test in api.c.
Add support for parsing and verifying certificates with RSA-PSS
signatures. Including check PSS parameters in key with those in
signature algorithm.
Add support for parsing private RSA PSS key.
Add support for parsing public RSA PSS key.
- Modify all existing tests to return an int. This moves us in the direction of
being able to return error/success from a test rather than just calling abort
when something fails. Also, all tests now have the same signature, so they can
easily be members of an array of test cases.
- Wrap each test in a TEST_CASE struct, which just stores a pointer to the test
function and the name of the test, for now. In the future, other metadata can
be added (e.g. should this test be run/skipped) to TEST_CASE, if desired.
- Modify all tests to return 0 on success. Right now, this doesn't do us much
good because the failure mechanism isn't returning some value != 0, it's
abort.
- Add TestSetup and TestCleanup functions that run before and after each test,
respectively. The former does nothing right now, and the latter clears the
error queue, if the error queue is compiled in.
wolfSSL_ERR_get_error and wolfSSL_ERR_peek_error_line_data should return the
earliest error in the queue (i.e. the error at the front), but prior to this
commit, they returned the latest/most recent one instead.
In DoAlert, we were adding an error to the queue for all alerts. However, a
close_notify isn't really an error. This commit makes it so DoAlert only adds
errors to the queue for non-close_notify alerts. In ReceiveData, similarly, we
were adding an error to the queue when the peer sent a close_notify, as
determined by ssl->error == ZERO_RETURN. Now, we don't add an error in this
case.
add wolfCrypt error codes IO_FAILED_E "Input/output failure" and SYSLIB_FAILED_E "System/library call failed";
tests/api.c and tests/unit.c: flush stdout for error message in Fail() macro, add fflush(stdout) after printf()s, print success message at end of unit_test(), and send several error messages to stderr instead of stdout;
wolfcrypt/test/test.c: add fallthrough macro definition of printf() that pairs it with fflush(stdout);
unit.h: in definition of macro AssertPtr(), add PRAGMA_GCC("GCC diagnostic ignored \"-Wpedantic\"");
sp_int.c: refactor several lingering instances of "if (0) { ... }" code pattern to #if 0 ... #endif.
1. Rename _DtlsUpdateWindow() as wolfSSL_Dtls_UpdateWindow() and make
it public so it may be tested.
2. Rename the internal functions DtlsWindowUpdate(), DtlsWindowCheck(),
and DtlsUpdateWindowGTSeq() as _DtlsWindowUpdate() and
_DtlsWindowCheck(), and _DtlsUpdateWindowGTSeq().
3. When updating the DTLS sequence window, and the next sequence
number (lo) wraps to zero, increment the next sequence number (hi)
by 1.
4. Fix an off-by-one error that wrapped around when saving the
packet sequence number in the bit-field window.
5. Adding a test for wolfSSL_DtlsUpdateWindow() function. With many test
cases. It is set up in a table format with running check values.
6. Change location of incrementing the difference when calculating the
location for setting the bit.
7. Updated the check of the sequence difference in the GT scenario.
8. In the DTLS window update functions remove newDiff and just use diff.
9. Handle the cases where the DTLS window crosses the high order word
sequence number change.
10. Add a debug option to print out the state of the DTLS sequence number
window.
In defense against attack, assume the imported public key is not trusted
and check it matches the private key if set.
Added APIs that allow application to explicitly trust public key.
Original APIs default to not trusting public key.
Add a define WOLFSSL_CHECK_MEM_ZERO to turn on code that checks that
memory that must be zeroized before going out of use is zero.
Everytime sensitive data is put into a allocated buffer or stack buffer;
the address, its length and a name is stored to be checked later.
Where the stack buffer is about to go out of use, a call is added to
check that the required parts are zero.
wc_MemZero_Add() adds an address with length and name to a table of
addressed to be checked later.
wc_MemZero_Check() checks that the memory associated with the address is
zeroized where required.
mp_memzero_add() adds mp_int's data pointer with length and name to
table.
mp_memzero_check() checks that the data pointer is zeroized where
required.
Freeing memory will check the address. The length was prepended on
allocation.
Realloction was changed for WOLFSSL_CHECK_MEM_ZERO to perform an
allocate, check, copy, free.
With PR 5170, I added logic that requires a EVP_CTRL_GCM_SET_IV_FIXED command be
issued before a EVP_CTRL_GCM_IV_GEN command. This matches OpenSSL's behavior.
However, OpenSSL also clears the flag enabling EVP_CTRL_GCM_IV_GEN after
EVP_CTRL_GCM_SET_IV_FIXED if EVP_CipherInit is called with a NULL key.
Otherwise, the flag retains its value. We didn't mirror this logic, and that
caused problems in OpenSSH unit testing. This commit aligns our logic with
OpenSSL's and adds a regression test to test_evp_cipher_aes_gcm for this case.
* PKCS7 should use allocated buffer for RSA.
* PKCS7 small stack typo for `keyAlgArray` size in `wc_PKCS7_AddRecipient_KTRI`.
* Fix for use of `free`, which should be `XFREE` in api.c.
* Cleanup old RSA benchmarking MDK5/WINCE code no longer needed with `WC_DECLARE_ARRAY_DYNAMIC_DEC` and `WC_DECLARE_ARRAY_DYNAMIC_EXE`.
Discovered the AES-GCM flow using this command didn't work in our OpenSSH port.
This commit makes the behavior match OpenSSL and adds testing using known
OpenSSL-generated test vectors to prevent regressions. This was one of those
problems where two ends of a connection would work fine if they were both using
wolfSSL but not if one was using OpenSSL (i.e. OpenSSH interop with AES-GCM
was broken).
Prior to this commit, if you wanted access to the Finished messages from a
handshake, you needed to turn on the compatibility layer, via one of
OPENSSL_ALL, WOLFSSL_HAPROXY, or WOLFSSL_WPAS. With this commit, defining any
of these causes WOLFSSL_HAVE_TLS_UNIQUE to be defined (a reference to the
tls-unique channel binding which these messages are used for) in settings.h.
This allows a user to define WOLFSSL_HAVE_TLS_UNIQUE to access the Finished
messages without bringing in the whole compat layer.
Hideki Miyazaki
Sean Parkinson
David Garske
Lealem Amedie
Juliusz Sosinowicz
JacobBarthelmeh
Daniel Pouzzner
Kareem
Marco Oliverio
tim-weller-wolfssl
Chris Conlon
Andras Fekete
Anthony Hu
Juliusz Sosinowicz
tmael
jordan
John Safranek
Eric Blankenhorn
kaleb-himes
Stefan Eissing
Hayden Roche
Uriah Pollock
Lealem Amedie
Satoshi Yamaguchi
Stefan Eissing
Stanislav Klima
Stefan Eissing
TakayukiMatsuo
Disyer
Takashi Kojo