Known problems in Intel® Integrated Performance Primitives Cryptography XTS-AES, GFp, and HMAC functions

The following issues were identified in the Intel® Integrated Performance Primitives (Intel® IPP) Cryptography XTS-AES, GFp, and HMAC functions. The problems affect the Intel® IPP 2017 Update 2 and earlier releases.

These issues will be fixed in the future versions of Intel® IPP. If your code is affected, use the following workaround to fix the problem, and improve the code security:

• ippsAESEncryptXTS_Direct and ippsAESDecryptXTS_Direct
  Problem: The ippsAESEncryptXTS_Direct and ippsAESDecryptXTS_Direct functions do not check the number of blocks in AES-XTS encryption/decryption operations.  The AES-XTS operations are required not to exceed 2^20 AES blocks.
  
  Workaround: To avoid issues with the large AES blocks number, check the blocks number in the application code.

•  ippsGFpxGetSize and ippsGFpECGetsize
  Problem:  The ippsGFpxGetSize and ippsGFpECGetsize functions do not perform check for integer overflow.
  
  Workaround:  Check the GF tower construction height in your application, and limit the extension of basic prime GF to less than 8.  The parameter in ippsGFpxGetSize() function should be  2<= degree <=8.  

• ippsHMACGetTag_rmf and ippsHMACGetTag
  Problem: The ippsHMACGetTag_rmf and ippsHMACGetTag functions leave some sensitive data after exit. This may lead to a leak of these data.
  
  Workaround: Use the following pairs of sequential calls to replace the ippsHMACGetTag_rmf and ippsHMACGetTag function:   
      ippsHMAC_Duplicate()and ippsHMAC_Final()
      ippsHMACDuplicate_rmf()and ippsHMACFinal_rmf()