xref: /isa-l_crypto/Release_notes.txt (revision 3aa2266bba7e1bce04a3fdb96386b1daf42fb065)

================================================================================
v2.25 Intel Intelligent Storage Acceleration Library Crypto Release Notes
================================================================================

================================================================================
RELEASE NOTE CONTENTS
================================================================================
1. KNOWN ISSUES
2. FIXED ISSUES
3. CHANGE LOG & FEATURES ADDED

================================================================================
1.  KNOWN ISSUES
================================================================================

* Perf tests do not run in Windows environment.

* 32-bit lib is not supported in Windows.

* 32-bit lib is not validated.

================================================================================
2. FIXED ISSUES
================================================================================
v2.25

* Fixed build with gcc 11.1.

* Fixed SHA512 internal reference function parameters.

* Fixed SM3 lane handling for aarch64.

* Replaced non-VEX encoded with VEX-encoded instructions in AES-XTS implementation.

* Fixed clang compilation with older assemblers.

* Fixed SHA512 calculation error on aarch64.

* Fixed MD5/SHA1/SHA256/SHA512 base functions for small inputs (less than block size).

* Fixed AES-XTS OpenSSL calls, limiting the input size, as per restrictions in 3.0 version.

* Fixed Windows build on test applications.

* Fixed SHA1 context structure to force alignment for lengths array.

v2.21

* Put correct vec instruction versions in aes_cbc_enc_{128,192,256}().  May help
  performance on some systems.

v2.20

* Fix issue with new aes_gcm API, aes_gcm_pre_256 was incorrect.

* Multi-buffer hash max length extended.  Previous max length for
  {sha1,sha256,sha512,md5}_mb was 4095MB.  While there is still a 4GB limit for
  each submit, the total hashed length can now be larger then 4GB.

v2.18

* Fix for multi-buffer hash when total length is above 512MB.

v2.14

* Building in unit directories is no longer supported removing the issue of
  leftover object files causing the top-level make build to fail.

v2.9

* Multi-buffer MD5 AVX2 tests fixed to work on FreeBSD 9.1 by explicitly aligning
  structures.

v2.7

* Unit tests and examples are now supported in Windows environment


================================================================================
3. CHANGE LOG & FEATURES ADDED
================================================================================
v2.25

* Added new API including parameter checking (starting with isal_ prefix).

* Added new API returning the library version.

* Added FIPS mode, including self tests in NIST approved algorithms (more information in https://github.com/intel/isa-l_crypto/blob/master/FIPS.md).

* Deprecated previous API in favour of new API (more information in https://github.com/intel/isa-l_crypto/wiki/New-API-introduced-from-v2.25).

* Optimized AES-GCM for AVX512-VAES x86 implementation.

* Optimized SM3 for AVX512 x86 implementation.

* Optimized MD5 and SM3 for aarch64.

* New optimized version of AES-CBC and AES-XTS for aarch64.

* Optimized multi-hash SHA1-Murmur for aarch64.

* Optimized multi-hash SHA1 for aarch64.

* Added ACVP test applications for AES-CBC, AES-GCM, AES-XTS and SHA1/256/512.

* Removed YASM support, so only NASM assembler is supported for x86.

* Bumped minimum NASM version to 2.14.01, which supports all x86 ISA used in this library.

v2.24

* New optimized version of AES-CBC decode

* New AVX2, 8 lane version of multi-buffer SM3

* Added support for big-endian architectures

v2.23

* New optimized versions of block ciphers AES-GCM and AES-XTS.

* New optimized versions of multi-buffer SM3 hashing.  Removed experimental
  status.

v2.22

* New multi-buffer SM3 functions. Experimental base functions only.

* New multi-arch support.

v2.21

* Multi-buffer hash performance improvement for Intel(R) Atom(tm) processors.
  New by-2 shani versions for multi-buffer sha1 & sha256.

* New base functions for multi-buffer hashes.
  md5_mb, sha1_mb, sha256_mb, sha512_mb.

v2.20

* New functions
  - Non-temporal versions of aes_gcm added.

* Multi-buffer hash improvement
  - Increase max length of hash in {sha1,sha256,sha512,md5}_mb to > 4GB.

v2.19

* Multi-buffer hash (sha1_mb, sha256_mb)

  - Choose fast single buffer routine to do flush operation if lanes aren't full.

  - Add SHA-NI support for Goldmont and Cannonlake.

* AES-GCM interface updates.

  - New interface separates the expanded keys and other context into two
  structures.  The old interface is maintained for backward compatibility.

  - User no longer has to append the GCM_IV_END_MARK manually to then end of iv
  as this is now done automatically.  This update should also improve performance
  of small packets.

* Rolling hash is released.

v2.18

* New multi-hash SHA256-based version.

v2.16

* Split lib from non-crypto functions.

v2.15

* Multi-buffer hash updates.  New AVX512 versions for multi-buffer SHA1, SHA256,
  SHA512, MD5 and SHA1+murmur3_x64_128 stitched.

* Removes restrictions on AAD length in AES-GCM.  Previously AAD length was
  limited to a multiple of 4 bytes.  Now any AAD length is allowed.

* Nasm support.  ISA-L ported to build with nasm or yasm assembler.

* Windows DLL support.  Windows builds DLL by default.

* The older, deprecated multi-buffer API has been removed.

v2.14

* New multi-hash sha1 function and multi-hash sha1 + murmur3_x64_128 stitched.
  Multi-hash is designed to give the performance of multi-buffer cryptographic
  hashes with a synchronous single buffer interface.

* New AES-GCM and AES-CBC functions added.

* Autoconf and autotools build allows easier porting to additional systems.
  Previous make system still available to embedded users with Makefile.unx.

* The AES key expand functions that were used for AES-XTS with pre-expanded keys
  now expand the decrypt keys in a different order. The order that decrypt keys
  are stored and used by XTS_AES_128_dec_expanded_key() is reversed from
  previous versions to be compatible with CBC and GCM key expansion. The
  aes_keyexp_*() and XTS_AES_128_dec_expanded_key() functions should work the
  same when paired together.

* Includes update for building on Mac OS X/darwin systems. Add --target=darwin
  to ./configure step.

v2.10

* Added multi-buffer MD5 in the new hash API.  Includes multi-binary capability,
  no restriction on update length and other benefits of the CTX API.

v2.9

* New multi-buffer hash API.  The new API brings the following new features to
  multi-buffer hashes.  The older API is still included but may be deprecated in
  future releases.

  - Multibinary functionality.  Call one function and the appropriate
    architecture-specific version is fixed up at runtime.

  - No restriction on update length.  Submitting an update block no longer has
    to have length a multiple of the fundamental block size.

* New expanded key tests added for AES-XTS 128 and 256

v2.7

* New AVX2 versions for mb_md5 and mb_sha512 hashing code have been added.

v2.6

* Update buffer functionality added to mb_md5, mb_sha256 and mb_sha512 hashing
  code.  Requires API changes to current interface to specify job type and total
  length of hash.

* New AVX2 versions for mb_sha1 and mb_sha256 hashing code have been added.

v2.5

* New feature for multi-buffer SHA-1, update buffer.  mb_sha1 non-finalize jobs
  can now be submitted by setting flags in job structure.  Requires API changes
  to current interface to specify job type and total length of hash.

v2.4

* Added new multi-buffer SHA-512: mb_sha512.  SSE, AVX versions.

v2.3

* Added improved AES XTS versions.

v2.2

* Added new AVX versions of multi-buffer hashes
* Changed type in the interface struct for multi-buffer hashes
  the len field in the following structures :JOB_SHA1,JOB_MD5,
  JOB_SHA256 is now a 32-bit int.

v2.0

* Added AES XTS units aes_xts_128, aes_xts_256

v1.3

* Added new multi-buffer units for SHA-256 and MD5: mb_sha256, mb_md5.