========================================================================
README for Intel(R) Multi-Buffer Crypto for IPsec Library

March 2018
========================================================================


Contents
========

- Overview
- Processor Extensions
- Recommendations
- Package Content
- Compilation
- Installation
- Legal Disclaimer


Overview
========
Intel Multi-Buffer Crypto for IPsec Library is highly-optimized
software implementations of the core cryptographic processing for IPsec,
which provides industry-leading performance on a range of Intel(R) Processors.

For information on how to build and use this library, see the
Intel White Paper:
"Fast Multi-buffer IPsec Implementations on Intel Architecture Processors".
Jim Guilford, Sean Gulley, et. al.

The easiest way to find it is to search the Internet for the title and
Intel White Paper.

Table 1. List of supported cipher algorithms and their implementations.
+---------------------------------------------------------------------+
|               |                   Implementation                    |
| Encryption    +-----------------------------------------------------|
|               | x86_64 | SSE    | AVX    | AVX2   | AVX512 | VAES(6)|
|---------------+--------+--------+--------+--------+--------+--------|
| AES128-GCM    | N      | Y  by8 | Y  by8 | Y  by8 | Y  by8 | Y x4by8|
| AES192-GCM    | N      | Y  by8 | Y  by8 | Y  by8 | Y  by8 | Y x4by8|
| AES256-GCM    | N      | Y  by8 | Y  by8 | Y  by8 | Y  by8 | Y x4by8|
| AES128-CCM    | Y(1)   | Y  by4 | Y  by8 | N      | N      | N      |
| AES128-CBC    | N      | Y(2)   | Y(4)   | N      | N      | N      |
| AES192-CBC    | N      | Y(2)   | Y(4)   | N      | N      | N      |
| AES256-CBC    | N      | Y(2)   | Y(4)   | N      | N      | N      |
| AES128-CTR    | N      | Y  by4 | Y  by8 | N      | N      | N      |
| AES192-CTR    | N      | Y  by4 | Y  by8 | N      | N      | N      |
| AES256-CTR    | N      | Y  by4 | Y  by8 | N      | N      | N      |
| NULL          | Y      | N      | N      | N      | N      | N      |
| AES128-DOCSIS | N      | Y(3)   | Y(5)   | N      | N      | N      |
| DES-DOCSIS    | Y      | N      | N      | N      | Y  x16 | N      |
| 3DES          | Y      | N      | N      | N      | Y  x16 | N      |
| DES           | Y      | N      | N      | N      | Y  x16 | N      |
+---------------------------------------------------------------------+

Notes:
(1)   - AES128-CCM scheduler code is implemented in C at the moment.
        Underlaying AES128-CTR algorithm utlizes SSE and AVX.
(2,3) - decryption is by4 and encryption is x4
(4,5) - decryption is by8 and encryption is x8
(6)   - AVX512 plus VAES and VPCLMULQDQ extensions

Legend:
  byY - single buffer Y blocks at a time
   xY - Y buffers at a time

As an example of how to read table 1 and 2, if one uses AVX512 interface
to perform AES128-CBC encryption then there is no native AVX512
implementation for this cipher. In such case, the library uses best
available implementation which is AVX for AES128-CBC.


Table 2. List of supported integrity algorithms and their implementations.
+-------------------------------------------------------------------------+
|                   |                   Implementation                    |
| Integrity         +-----------------------------------------------------|
|                   | x86_64 | SSE    | AVX    | AVX2   | AVX512 | VAES(4)|
|-------------------+--------+--------+--------+--------+--------+--------|
| AES-XCBC-96       | N      | Y   x4 | Y   x8 | N      | N      | N      |
| HMAC-MD5-96       | Y(1)   | Y x4x2 | Y x4x2 | Y x8x2 | N      | N      |
| HMAC-SHA1-96      | N      | Y(3)x4 | Y   x4 | Y   x8 | Y  x16 | N      |
| HMAC-SHA2-224_112 | N      | Y(3)x4 | Y   x4 | Y   x8 | Y  x16 | N      |
| HMAC-SHA2-256_128 | N      | Y(3)x4 | Y   x4 | Y   x8 | Y  x16 | N      |
| HMAC-SHA2-384_192 | N      | Y   x2 | Y   x2 | Y   x4 | Y   x8 | N      |
| HMAC-SHA2-512_256 | N      | Y   x2 | Y   x2 | Y   x4 | Y   x8 | N      |
| AES128-GMAC       | N      | Y  by8 | Y  by8 | Y  by8 | Y  by8 | Y x4by8|
| AES192-GMAC       | N      | Y  by8 | Y  by8 | Y  by8 | Y  by8 | Y x4by8|
| AES256-GMAC       | N      | Y  by8 | Y  by8 | Y  by8 | Y  by8 | Y x4by8|
| NULL              | N      | N      | N      | N      | N      | N      |
| AES128-CCM        | Y(2)   | Y   x4 | Y   x8 | N      | N      | N      |
| AES128-CMAC-96    | Y      | Y   x4 | Y   x8 | N      | N      | N      |
+-------------------------------------------------------------------------+

Notes:
(1)   - MD5 over one block implemented in C
(2)   - AES128-CCM scheduler code is implemented in C.
        Underlaying AES128-CBC algorithm utlizes SSE and AVX.
(3)   - Implementation using SHANI extentions is x2
(4)   - AVX512 plus VAES and VPCLMULQDQ extensions

Legend:
  byY - single buffer Y blocks at a time
   xY - Y buffers at a time

Table 3. Encryption and integrity algorithm combinations
+---------------------------------------------------------------------+
| Encryption    | Allowed Integrity Algorithms                        |
|---------------+-----------------------------------------------------|
| AES128-GCM    | AES128-GMAC                                         |
|---------------+-----------------------------------------------------|
| AES192-GCM    | AES192-GMAC                                         |
|---------------+-----------------------------------------------------|
| AES256-GCM    | AES256-GMAC                                         |
|---------------+-----------------------------------------------------|
| AES128-CCM    | AES128-CCM                                          |
|---------------+-----------------------------------------------------|
| AES128-CBC,   | AES-XCBC-96,                                        |
| AES192-CBC,   | HMAC-SHA1-96, HMAC-SHA2-224_112, HMAC-SHA2-256_128, |
| AES256-CBC,   | HMAC-SHA2-384_192, HMAC-SHA2-512_256,               |
| AES128-CTR,   | AES128-CMAC-96,                                     |
| AES192-CTR,   | NULL                                                |
| AES256-CTR,   |                                                     |
| NULL,         |                                                     |
| AES128-DOCSIS,|                                                     |
| DES-DOCSIS,   |                                                     |
| 3DES,         |                                                     |
| DES,          |                                                     |
+---------------+-----------------------------------------------------+


Processor Extensions
====================

Table 4. Processor extensions used in the library
+-------------------------------------------------------------------------+
| Algorithm         | Interface | Extensions                              |
|-------------------+-----------+-----------------------------------------|
| HMAC-SHA1-96,     | AVX512    | AVX512F, AVX512BW, AVX512VL             |
| HMAC-SHA2-224_112,|           |                                         |
| HMAC-SHA2-256_128,|           |                                         |
| HMAC-SHA2-384_192,|           |                                         |
| HMAC-SHA2-512_256 |           |                                         |
|-------------------+-----------+-----------------------------------------|
| DES, 3DES,        | AVX512    | AVX512F, AVX512BW                       |
| DOCSIS-DES        |           |                                         |
|-------------------+-----------+-----------------------------------------|
| HMAC-SHA1-96,     | SSE       | SHANI                                   |
| HMAC-SHA2-224_112,|           | - presence is autodetected and library  |
| HMAC-SHA2-256_128,|           |   falls back to SSE implementation      |
| HMAC-SHA2-384_192,|           |   if not present                        |
| HMAC-SHA2-512_256 |           |                                         |
|-------------------+-----------+-----------------------------------------|


Recommendations
===============

Legacy or to be avoided algorithms listed in the table below are implemented
in the library in order to support legacy applications. Please use corresponding
alternative algorithms instead.

+----------------------------------------------------------+
| # | Algorithm          | Recommendation | Alternative    |
|---+--------------------+----------------+----------------|
| 1 | DES encryption     | Avoid          | AES encryption |
|---+--------------------+----------------+----------------|
| 2 | 3DES encryption    | Avoid          | AES encryption |
|---+--------------------+----------------+----------------|
| 3 | HMAC-MD5 integrity | Legacy         | HMAC-SHA1      |
+----------------------------------------------------------+


Package Content
===============

LibTestApp - Library test application
LibPerfApp - Library performance application
sse - Intel(R) SSE optimized routines
avx - Intel(R) AVX optimized routines
avx2 - Intel(R) AVX2 optimized routines
avx512 - Intel(R) AVX512 optimized routines
no-aesni - Non-AESNI accelerated routines

Compilation
===========

Linux (64-bit only)
-------------------

Required tools:
- GNU make
- NASM version 2.13.03 (or newer)
- gcc (GCC) 4.8.3 (or newer)

Shared library:
> make

Static library:
> make SHARED=n

Clean the build:
> make clean
or
> make clean SHARED=n

Build with debugging information:
> make DEBUG=y

Note: Building with debugging information is not advised for production use.

Windows (x64 only)
------------------

Required tools:
- Microsoft (R) Visual Studio 2015:
  - NMAKE: Microsoft (R) Program Maintenance Utility Version 14.00.24210.0
  - CL: Microsoft (R) C/C++ Optimizing Compiler Version 19.00.24215.1 for x64
  - LIB: Microsoft (R) Library Manager Version 14.00.24215.1
  - LINK: Microsoft (R) Incremental Linker Version 14.00.24215.1
  - Note: Building on later versions should work but is not verified
- NASM version 2.13.03 (or newer)

Shared library (DLL):
> nmake /f win_x64.mak

Static library:
> nmake /f win_x64.mak SHARED=n

Clean the build:
> nmake /f win_x64.mak clean
or
> nmake /f win_x64.mak clean SHARED=n

Build with debugging information:
> nmake /f win_x64.mak DEBUG=y

Note: Building with debugging information is not advised for production use.

Installation
============

Linux (64-bit only)
-------------------

First compile the library and then install:
> make
> sudo make install

To uninstall the library run:
> sudo make uninstall

If you want to change install location then define PREFIX
> sudo make install PREFIX=<path>

If there is no need to run ldconfig at install stage please use NOLDCONFIG=y option.
> sudo make install NOLDCONFIG=y

If library was compiled as an archive (not a default option) then install it
using SHARED=n option:
> sudo make install SHARED=n

Windows (x64 only)
------------------

First compile the library and then install from a command prompt in
administrator mode:
> nmake /f win_x64.mak
> nmake /f win_x64.mak install

To uninstall the library run:
> nmake /f win_x64.mak uninstall

If you want to change install location then define PREFIX (default C:\Program Files)
> nmake /f win_x64.mak install PREFIX=<path>

If library was compiled as a static library (not a default option) then install it
using SHARED=n option:
> nmake /f win_x64.mak install SHARED=n

Legal Disclaimer
================

THIS SOFTWARE IS PROVIDED BY INTEL"AS IS". NO LICENSE, EXPRESS OR
IMPLIED, BY ESTOPPEL OR OTHERWISE, TO ANY INTELLECTUAL PROPERTY RIGHTS
ARE GRANTED THROUGH USE. EXCEPT AS PROVIDED IN INTEL'S TERMS AND
CONDITIONS OF SALE, INTEL ASSUMES NO LIABILITY WHATSOEVER AND INTEL
DISCLAIMS ANY EXPRESS OR IMPLIED WARRANTY, RELATING TO SALE AND/OR
USE OF INTEL PRODUCTS INCLUDING LIABILITY OR WARRANTIES RELATING TO
FITNESS FOR A PARTICULAR PURPOSE, MERCHANTABILITY, OR INFRINGEMENT
OF ANY PATENT, COPYRIGHT OR OTHER INTELLECTUAL PROPERTY RIGHT.