Kyber/ML-KEM: make both available #8143
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
SparkiDev
force-pushed
the
kyber_plus_mlkem
branch
4 times, most recently
from
4a579a3 to
c7c586a
Compare
Make Kyber and ML-KEM individually available as well as at the same time. Modified TLS layer to support both Kyber and ML-KEM. Added new identifiers in TLS layer for ML-KEM.
SparkiDev
force-pushed
the
kyber_plus_mlkem
branch
from
c7c586a to
7d42dda
Compare
douzzer
approved these changes
Nov 9, 2024
kp-thomas-yau
added a commit
to expressvpn/wolfssl-rs
that referenced
this pull request
Nov 15, 2024
We would use a patch to use WolfSSL's implementation of both Kyber and ML-KEM so that we can remove liboqs while maintaing support for Kyber at the moment. This patch uses commits and code changes from the following PR in WolfSSL: - wolfSSL/wolfssl#8143 - wolfSSL/wolfssl#8172 - wolfSSL/wolfssl#8183 - wolfSSL/wolfssl#8185
kp-thomas-yau
added a commit
to expressvpn/wolfssl-rs
that referenced
this pull request
Nov 15, 2024
We would use a patch to use WolfSSL's implementation of both Kyber and ML-KEM so that we can remove liboqs while maintaing support for Kyber at the moment. This patch uses commits and code changes from the following PR in WolfSSL: - wolfSSL/wolfssl#8143 - wolfSSL/wolfssl#8172 - wolfSSL/wolfssl#8183 - wolfSSL/wolfssl#8185
kp-thomas-yau
added a commit
to expressvpn/wolfssl-rs
that referenced
this pull request
Nov 15, 2024
We would use a patch to use WolfSSL's implementation of both Kyber and ML-KEM so that we can remove liboqs while maintaing support for Kyber at the moment. This patch uses commits and code changes from the following PR in WolfSSL: - wolfSSL/wolfssl#8143 - wolfSSL/wolfssl#8172 - wolfSSL/wolfssl#8183 - wolfSSL/wolfssl#8185
kp-thomas-yau
added a commit
to expressvpn/wolfssl-rs
that referenced
this pull request
Nov 15, 2024
We would use a patch to use WolfSSL's implementation of both Kyber and ML-KEM so that we can remove liboqs while maintaing support for Kyber at the moment. This patch uses commits and code changes from the following PR in WolfSSL: - wolfSSL/wolfssl#8143 - wolfSSL/wolfssl#8172 - wolfSSL/wolfssl#8183 - wolfSSL/wolfssl#8185
kp-thomas-yau
added a commit
to expressvpn/wolfssl-rs
that referenced
this pull request
Nov 18, 2024
We would use a patch to use WolfSSL's implementation of both Kyber and ML-KEM so that we can remove liboqs while maintaining support for Kyber at the moment. This patch uses commits and code changes from the following PR in WolfSSL: - wolfSSL/wolfssl#8143 - wolfSSL/wolfssl#8172 - wolfSSL/wolfssl#8183 - wolfSSL/wolfssl#8185
kp-thomas-yau
added a commit
to expressvpn/wolfssl-rs
that referenced
this pull request
Nov 18, 2024
We would use a patch to use WolfSSL's implementation of both Kyber and ML-KEM so that we can remove liboqs while maintaining support for Kyber at the moment. This patch uses commits and code changes from the following PR in WolfSSL: - wolfSSL/wolfssl#8143 - wolfSSL/wolfssl#8172 - wolfSSL/wolfssl#8183 - wolfSSL/wolfssl#8185
kp-thomas-yau
added a commit
to expressvpn/lightway-core
that referenced
this pull request
Nov 25, 2024
Remove references to liboqs and edit build flag/configs for each of the platform. Since WolfSSL would officially release their Kyber/ML-KEM implementations a few months later, we would use the git patch to essentially patch their PRs on top of the 5.7.4 release. The patch consists of the commits and code changes from the following PR from WolfSSL: - wolfSSL/wolfssl#8143 - wolfSSL/wolfssl#8172 - wolfSSL/wolfssl#8183 - wolfSSL/wolfssl#8185 Configuration for enabling ML-KEM/Kyber: 1. For only ML-KEM: ./configure --enable-kyber ./configure --enable-kyber=all,ml-kem 2. For just Kyber: ./configure --enable-kyber=all,original 3. For ML-KEM and Kyber ./configure --enable-kyber=all,original,ml-kem ./configure --enable-kyber=all,ml-kem,original
8 tasks
kp-thomas-yau
added a commit
to expressvpn/lightway-core
that referenced
this pull request
Nov 26, 2024
Remove references to liboqs and edit build flag/configs for each of the platform. Since WolfSSL would officially release their Kyber/ML-KEM implementations a few months later, we would use the git patch to essentially patch their PRs on top of the 5.7.4 release. The patch consists of the commits and code changes from the following PR from WolfSSL: - wolfSSL/wolfssl#8143 - wolfSSL/wolfssl#8172 - wolfSSL/wolfssl#8183 - wolfSSL/wolfssl#8185 Configuration for enabling ML-KEM/Kyber: 1. For only ML-KEM: ./configure --enable-kyber ./configure --enable-kyber=all,ml-kem 2. For just Kyber: ./configure --enable-kyber=all,original 3. For ML-KEM and Kyber ./configure --enable-kyber=all,original,ml-kem ./configure --enable-kyber=all,ml-kem,original
dgarske
added a commit
to LinuxJedi/wolfssl
that referenced
this pull request
Nov 27, 2024
kp-thomas-yau
added a commit
to expressvpn/lightway-core
that referenced
this pull request
Dec 2, 2024
Remove references to liboqs and edit build flag/configs for each of the platform. Since WolfSSL would officially release their Kyber/ML-KEM implementations a few months later, we would use the git patch to essentially patch their PRs on top of the 5.7.4 release. The patch consists of the commits and code changes from the following PR from WolfSSL: - wolfSSL/wolfssl#8143 - wolfSSL/wolfssl#8172 - wolfSSL/wolfssl#8183 - wolfSSL/wolfssl#8185 Configuration for enabling ML-KEM/Kyber: 1. For only ML-KEM: ./configure --enable-kyber ./configure --enable-kyber=all,ml-kem 2. For just Kyber: ./configure --enable-kyber=all,original 3. For ML-KEM and Kyber ./configure --enable-kyber=all,original,ml-kem ./configure --enable-kyber=all,ml-kem,original
kp-thomas-yau
added a commit
to expressvpn/lightway-core
that referenced
this pull request
Dec 3, 2024
Remove references to liboqs and edit build flag/configs for each of the platform. Since WolfSSL would officially release their Kyber/ML-KEM implementations a few months later, we would use the git patch to essentially patch their PRs on top of the 5.7.4 release. The patch consists of the commits and code changes from the following PR from WolfSSL: - wolfSSL/wolfssl#8143 - wolfSSL/wolfssl#8172 - wolfSSL/wolfssl#8183 - wolfSSL/wolfssl#8185 Configuration for enabling ML-KEM/Kyber: 1. For only ML-KEM: ./configure --enable-kyber ./configure --enable-kyber=all,ml-kem 2. For just Kyber: ./configure --enable-kyber=all,original 3. For ML-KEM and Kyber ./configure --enable-kyber=all,original,ml-kem ./configure --enable-kyber=all,ml-kem,original
gasbytes
pushed a commit
to gasbytes/wolfssl
that referenced
this pull request
Dec 21, 2024
gasbytes
pushed a commit
to gasbytes/wolfssl
that referenced
this pull request
Dec 21, 2024
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. linuxkm: work around aarch64 dependency on alt_cb_patch_nops for enable-linuxkm-pie (FIPS support). wolfssl/wolfcrypt/aes.h: #define WC_NO_COMPAT_AES_BLOCK_SIZE in OPENSSL_COEXIST builds. see comment in source code with usage instructions. wolfhsm-mldsa-fixes SP ARM: big-endian support Handle reading and writing from big-endian byte array when compiling for big endian. Rework little endian to be more effiecient too. Fix wolfSSL_X509_STORE_get0_objects to handle case where no CA has been loaded Fix conversion on various files. Work from Reda. Additional conversion warnings. Addressing CI/CD before continuing with the fixing More Wconversion fixing (Renesas specific) Moved variable to the top of the scope fixes for OPENSSL_COEXIST with FIPS and with/without TEST_OPENSSL_COEXIST. adjustments to x509.h macro list Add STM32MP13 to Cube IDE Add STM32MP13 HAL support for more SHA types This adds STM32 HAL support for: * SHA384 * SHA512 (with -224 and -256) * SHA3 (all variants apart from SHAKE) The partial FIFO block calculations have been adjusted based in the STM32 code to support the larger hash sizes. This should work with other chips such as the STM32U5xx, but is not enabled for that yet. Fix STM32 example broken in wolfSSL#8143. Fixes for building with SP RSA small and RSA Public only. ZD 18996 src/ssl_sess.c: in wolfSSL_CTX_flush_sessions(), add missing check of s->sessionIDSz, similar to the fix to TlsSessionCacheGetAndLock() in wolfSSL#8182 (ef67b1c). also, add missing macro to .wolfssl_known_macro_extras. Fix test environment Add size checks to sessionID Fix for Compressed Keys with FIPS build dsa in visual studio wc_port: change zephyr struct k_thread tid member to pointer. Fix issue with wc_lms_impl.c or wc_lms not including settings.h. Caused issue enabling LMS from user_settings.h. Fixes for ML-DSA and LMS cast warnings and spelling errors. Expose compatibility get_verify functions with openssl_extra. Fix broken verify on Ada wrapper The Ada wrapper had an `&` operator for the verification mode. This effectively caused the verification mode to equal `0`. The operator has been switched to `or` now, in addition, a getter has been added to the API. This allows for the test I've added to the server code to verify that it is being set correctly. `OPENSSL_ALL` flag added to Ada so that the verify mode getter function is compiled in. Fixes wolfSSL#7461 Thanks to @dalybrown for reporting it. Add libspdm action Depends on wolfSSL/osp#217 configure.ac: add --enable-fips=cert4718 alias for v5, and make --enable-fips=v5 set FIPS to 5.2.1; set DEF_FAST_MATH and DEF_SP_MATH to "no" when "yes" would conflict with user-supplied arguments. configure.ac: fix SC1105 ("Shells disambiguate (( differently or not at all."). wolfssl/wolfcrypt/types.h and wolfssl/wolfcrypt/hash.h: define WOLF_AGG_DUMMY_MEMBER, pivoting on HAVE_EMPTY_AGGREGATES, and use WOLF_AGG_DUMMY_MEMBER in wc_Hashes. src/ssl_crypto.c: revert FIPS gate threshold in wolfSSL_AES_decrypt() changed in d85c108 -- original value was correct, misdiagnosed by faulty test. update fips-check.sh for cert wolfSSL#4718: remap linuxv5 as an alias for linuxv5.2.1, and add linuxv5-RC12. fips-check.sh: add support for WOLFSSL_REPO and noautogen option; tweak git fetching to keep wolfssl and fips tags distinct, and fetch all needed tags by name to assure availability for checkout. also, hide stdout noise from pushd/popd. peer review: refactor HAVE_ANONYMOUS_INLINE_AGGREGATES and HAVE_EMPTY_AGGREGATES to conform to wolfssl convention -- defined() for true, !defined() for false -- while retaining ability for user override-off by passing in explicit 0 definition. src/internal.c: in HashSkeData(), remove unneeded logically faulty nullness check around XFREE(ssl->buffers.digest.buffer, ...). Add nss interop Add sanity check for configuration method Disable hitch OSP test Fix from review move !defined(EXTERNAL_OPTS_OPENVPN) assert from src/internal.c to wolfssl/wolfcrypt/types.h with refinements; refine logic+message of assert in wolfssl/wolfcrypt/settings.h re "wolfssl/options.h included in compiled wolfssl library object..". wolfssl/wolfcrypt/settings.h: use #warning, not #error, for "No configuration for wolfSSL detected, check header order", to avoid unnecessary breakage of old projects with nonstandard custom settings. Revert to ubuntu-22.04 add support for WOLFSSL_NO_OPTIONS_H: * activate WOLFSSL_NO_OPTIONS_H in linuxkm/Kbuild for in-module test.o and benchmark.o. * refine explanatory comments in settings.h re WOLFSSL_USE_OPTIONS_H, WOLFSSL_NO_OPTIONS_H, and WOLFSSL_CUSTOM_CONFIG. * add safety catch to options.h/options.h.in to inhibit inclusion if defined(WOLFSSL_NO_OPTIONS_H). * for good measure, add explicit check for WOLFSSL_NO_OPTIONS_H to wolfcrypt/benchmark/benchmark.c and wolfcrypt/test/test.c. Improve Espressif SHA HW/SW mutex messages .wolfssl_known_macro_extras: regenerate Use proper ref count handling when adding to x509 store Always keep original x509 pointer with proper refcounts even for self signed trusted CA Dont use specific free function Free x509 on fail to push CMAKE: look for pthreads when importing wolfSSL if required All required dependencies of a package must also be found in the package configuration file. Consumers of wolfSSL can't know if it was built with or without threads support. This change adds find_package(Threads) lookup in the file used for find_package(wolfssl) if wolfSSL was built with threads support. Initial implementation for using PKCS11 to retrieve certificate for SSL CTX Updates per review comments Add support for cert format in get cert crypto callback Use char instead of sword8, sanity length check on CKA_VALUE No redundant NULL check on free Remove redundant NULL check Aarch64 Poly1305: fix corner case Don't mask top 26 bits as it may have next bit set as reduction step was only approximate. Fix memory leak make new sanity check be a warning defining custom config avoids warning of library builds pulling in options.h Add support for the RFC822 Mailbox attribute. Aarch64: make code compile when no hardware crypto avail Detects availability of instructions for Aarch64. WOLFSSL_ALWAYS_KEEP_SNI enabled by default with --enable-jni wolfSSL_CTX_set_tlsext_use_srtp() should return 1 on failure and 0 upon success. Same with wolfSSL_set_tlsext_use_srtp(). See https://docs.openssl.org/1.1.1/man3/SSL_CTX_set_tlsext_use_srtp/ Add a test. Various cleanups and fixes: * 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 a10260c. * 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 Fixes for macro names. Cleanup the gating for `WOLFSSL_NO_AES_CFB_1_8`. Revert "Aarch64: make code compile when no hardware crypto avail" fix for sig fault harden build linuxkm/Kbuild and linuxkm/module_exports.c.template: on kernel >=6.13, add quotes around the namespace arg to EXPORT_SYMBOL_NS_GPL() (upstream change actually made in 6.13-rc2). Aarch64: make code compile when no hardware crypto avail Detects availability of instructions for Aarch64. MacOS: allow SHA-3 instructions to be explicitly not used Some iPads and iPhones don't support SHA-3 instructions. Allow SHA-3 instructions to explicitly not be used for these devices. Fix compile issue with NO_WOLFSSL_DIR `test_wolfSSL_CTX_load_system_CA_certs()` would try to use DIR functions when `NO_WOLFSSL_DIR` was used. EdDSA Ed448: sc_muladd now does full reduction sc_muladd was reducing to word boundary and not to order. Now reduces to order as last step. CID also supported in DTLS 1.2 Add CID interop with mbedtls add shebang use unique key fix redirect order Use source hostap repo Initialize vars & change types to appease Windows/VS fips-check.sh fixes + enhancements: * change default WOLFSSL_REPO to the canonical upstream. * refactor tag calculation without bash associative arrays, for backward compat. * add support for fetching FIPS tags/branches into a persistent fips repo if one is found at ../fips. * use --shared in git clones where applicable. * always check out the master FIPS branch, for its tooling, and always make sure it's up to date with $FIPS_REPO. * after each fetch for a previously unknown tag, explicitly associate the tag with the FETCH_HEAD. Enable support for using certificate manager only. Fixes for building without TLS enabled (NO_TLS). ZD 19054. Tested using `./configure --disable-tlsv12 --disable-tls13 CFLAGS="-DNO_TLS" && make check` Fix issues in `test_tls13_apis` with no filesystem or no RSA/ECC. Fix nested `NO_TLS`. Further fixes with NO_TLS to support use with compatibility layer. Add `--disable-tls` option that can be used with `--enable-all` to disable TLS features and set `NO_TLS`. Useful for allowing certificate manager and crypto compatibility API's only. configure.ac: fix faulty logic in FIPS v6 feature calculation re ENABLED_ARMASM_CRYPTO, originally added in 6e0a901. wolfcrypt/src/aes.c: add missing WOLFSSL_ARMASM gate clause around wolfCrypt_FIPS_aes_ro_sanity, necessitated by 514a92d/wolfSSL#8293. wolfCrypt -Wconversion expansion: fix numerous warnings, all benign, from -Warith-conversion -Wenum-conversion -Wfloat-conversion -Wsign-conversion. Espressif Managed Component wolfSSL 5.7.4 post-release update fedora crypto-policies: initial support. Fix RA6M jankins failure Printing the rfc822Mailbox x509 attribute fix: cast int operands to size_t in bio buffer size calc to prevent loss of precision Fix C4333 warning by adjusting right shift operation on byte cast
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Description
Make Kyber and ML-KEM individually available as well as at the same time.
Modified TLS layer to support both Kyber and ML-KEM. Added new identifiers in TLS layer for ML-KEM.
Testing
Regression tested Kyber/ML-KEM.
./configure -enable-kyber=all,original now has both implementations.
Checklist