AES-GCM: don't generate M0 when using assembly unless falling back to C
and then use new assembly code.
HMAC: add option to copy hashes (--enable-hash-copy
-DWOLFSSL_HMAC_COPY_HASH) to improve performance when using the same key
for multiple operations.
AES GCM streaming - fix GHASH_ONE_BLOCK to use CPU feature information.
AES-GCM uses EOR3 (SHA-3 instruction) - split assembly code.
Kyber uses SQRDMLSH - split assembly code.
Changed define from WOLFSSL_AARCH64_NO_SQRMLSH to
WOLFSSL_AARCH64_NO_SQRDMLSH to match instruction.
Improved array data format for inline assembly code.
* Fix to properly set configure.ac LMS/XMSS enables and build of those code files.
* Remove duplicate aes.c `wc_AesSetKeyLocal` call to `wc_AesSetIV`. Moved earlier in function in commit a10260ca5f.
* Benchmark missing time.h with NO_ASN_TIME.
* Added option to support disabling AES CFB 1/8 `WOLFSSL_NO_AES_CFB_1_8`.
* Fixes for building with combinations of `WOLFSSL_RSA_VERIFY_ONLY` and `WOLFSSL_RSA_PUBLIC_ONLY`.
* Fix for building `--enable-stacksize=verbose` with single threaded.
* Various tab and formatting cleanups.
ZD 18996
This adds support for the STM32MP13 HAL, tested on the STM32MP135F MPU.
Using the HAL this modifies our previous RNG, AES-CBC, AES-GCM, HASH,
ECDSA and DES3 ST HAL acceleration to work with the MPU. It also works
around bugs found in the AES-GCM code of the HAL.
The HAL does not appear to have support for MD5 HASH at the moment, so
this has been given a flag to disable it on this MPU.
wrapper/CSharp/wolfSSL_CSharp/wolfCrypt.cs: update for new calling conventions around wc_AesNew, wc_curve25519_new, wc_ed25519_new, wc_HashNew, and wc_NewRsaKey, and the corresponding delete functions.
wolfssl/wolfcrypt/aes.h: add Aes.streamData_sz;
src/tls13.c: fix devId passed to wc_HmacInit() in CreateCookieExt() and TlsCheckCookie();
src/keys.c: in SetKeys(), call wc_HmacInit() on hmacs only if newly allocated;
wolfcrypt/src/aes.c:
* in wc_Gmac(), wc_GmacVerify(), and AesSivCipher(), use wc_AesNew() and wc_AesDelete();
* in wc_AesInit(), zero the object on entry, and remove superseded piecemeal initializations to zero;
* in wc_AesFree(), zero aes->streamData, and zero the entire object as final cleanup;
wolfcrypt/src/curve25519.c: in wc_curve25519_free(), zero the entire object rather than zeroing piecemeal;
wolfcrypt/test/test.c:
* add fallback implementations (for old FIPS) of wc_HashNew(), wc_HashDelete(), wc_curve25519_new(), wc_curve25519_delete(), wc_ed25519_new(), and wc_ed25519_delete();
* update constructor calls throughout for new semantics;
* refactor ed25519_test() for proper cleanup and error encoding.
add delete APIs, matching recently added wc_AesNew, wc_curve25519_new, wc_ed25519_new, wc_HashNew, and wc_NewRsaKey:
* wc_AesDelete()
* wc_HashDelete()
* wc_DeleteRsaKey()
* wc_curve25519_delete()
* wc_ed25519_delete()
* remove handling in corresponding preexisting free APIs for recently added .isAllocated member -- this restores preexisting semantics;
* add WC_NO_CONSTRUCTORS gate, and auto-activate it when NO_WOLFSSL_MEMORY && WOLFSSL_NO_MALLOC (unless preempted by XMALLOC_USER or XMALLOC_OVERRIDE);
* exclude recently added .isAllocated members from wolfcrypt structs when defined(WC_NO_CONSTRUCTORS);
* adjust wolfcrypt/test/test.c for consistency with the above, and fix cleanup codes/dynamics in several tests.
* Adds RNG, ECC(ECIES and ECDHE), RSA, ED25519/Curve25519, AES-GCM, and Hashing to the CSharp wrapper.
* Adds GitHub action for building the CSharp wrapper solution and running wolfCrypt test and a TLS server/client example.
* Adds "new" API's for wolfCrypt for platforms that cannot tolerate the structs directly.
* Fixes for several scan-build warnings.
wolfcrypt/src/aes.c:
* when SAVE_VECTOR_REGISTERS2_DOES_NOTHING, define do-nothing VECTOR_REGISTERS_PUSH and VECTOR_REGISTERS_POP, to mollify Coverity CONSTANT_EXPRESSION_RESULT;
* in AesGcmDecryptUpdate_aesni(), omit " && (c != NULL)" clause from computation of endA argument to AesGcmAadUpdate_aesni(), to mollify Coverity FORWARD_NULL (impermissible nullness is already checked and BAD_FUNC_ARGed by the sole caller, wc_AesGcmDecryptUpdate());
wolfcrypt/src/misc.c: add readUnalignedWord64(), writeUnalignedWord64(), readUnalignedWords64(), and writeUnalignedWords64(), for safe word64 access to possibly-unaligned data;
wolfcrypt/src/wc_kyber_poly.c: use readUnalignedWords64() and readUnalignedWord64() to mitigate sanitizer-reported "load of misaligned address".
* Proposed new interface for AesSivEncrypt with number of ADs != 1.
* Implement AES SIV S2V computation with a number of ADs not equal to 1.
* Add Example A.1 from RFC5297 to AES SIV test vectors.
* Add tests for new AES SIV interface, and add test vectors for examples given in RFC5297.
* Include the nonce in count of maximum number of ADs.
* Addressing review comments.
* Addressing review comments: Use uppercase 'U' suffix on unsigned constant.
* Rename local variables named 'ad0' to 'ad', since the zero makes no sense, especially since in the RFC 5297 document they're actually counting the ADs from 1.
tests/api.c: add error_test() adapted from wolfcrypt/test/test.c, checking all error strings for expected presence/absence and length, called from existing test_wolfSSL_ERR_strings().
wolfssl/ssl.h, wolfssl/error-ssl.h, and wolfssl/wolfcrypt/error-crypt.h:
* move several negative error return codes from ssl.h to error-ssl.h,
* renumber them to conform to existing sequence, and
* include error-ssl.h from ssl.h;
* add special-case WOLFSSL_DEBUG_TRACE_ERROR_CODES macros for WOLFSSL_FAILURE;
* add missing WOLFSSL_API attribute to wc_backtrace_render().
add numerous WC_NO_ERR_TRACE()s to operand error code uses, cleaning up error traces in general, and particularly when WOLFSSL_DEBUG_TRACE_ERROR_CODES_ALWAYS.
* crypto lib (36),
* crypto test&benchmark (20),
* TLS lib (179),
* examples (122),
* linuxkm (3),
* tests/api.c (2272).
