WireGuard to Create Zinc Cryptography Library

It is expected that after the introduction of library into the main Linux kernel, it will accelerate the process of data encryption
03 August 2018   1859

The developers of the VPN protocol WireGuard announced the creation of the Zinc cryptographic library. It is expected that after the introduction of WireGuard into the main Linux kernel, it will accelerate the process of data encryption due to a simplified set of crypto algorithms.

The Zinc library consists of all cryptographic primitives used in WireGuard, assembled in a simplified form. Its task is to increase the performance of data encryption and fast execution of SIMD instructions.

Unlike the CryptoAPI interface used in the current Linux kernel, WireGuard with the Zinc library offers a simplified set of ready-made functions. They can only be used for their intended purpose, which eliminates the appearance of superfluous high-level abstractions. It is believed that the incorrect use of functions is the main source of problems in the development of applications.

Zinc offers the following cryptographic primitives:

  • stream ciphers ChaCha20 and HChaCha20;
  • technology for authentication of Poly1305 messages;
  • function Curve25519 with the Diffie-Hellman protocol to create a private key;
  • hash function BLAKE2s with a performance at MD5;
  • encryption mechanisms ChaCha20-Poly1305 and XChaCha20-Poly1305.

In July 2018, the expert on cryptography Daniel J. Bernstein (Daniel J. Bernstein) published the djbsort library. It is designed for cryptographic systems and encryption algorithms and speeds up the sorting of arrays of integers.

LLVM 10.0.0 to be Released

New version of the popular development toolkit brings, among other things, support for the C++ Concepts
26 March 2020   969

After six months of development, the release of the LLVM 10.0 project, a GCC-compatible toolkit (compilers, optimizers, and code generators), compiling programs into an intermediate bitcode of RISC-like virtual instructions (a low-level virtual machine with a multi-level optimization system), is presented. The generated pseudo-code can be converted using the JIT compiler into machine instructions directly at the time of program execution.

Among the new features of LLVM 10.0, there are support for C ++ Concepts (C ++ Concepts), termination of the launch of Clang in the form of a separate process, support for CFG checks (control flow guard) for Windows, and support for new CPU features.

The main innovations of LLVM 10.0:

  • New interprocedural optimizations and analyzers have been added to the Attributor framework. The prediction of the state of 19 different attributes, including 12 attributes of 12 LLVM IR and 7 abstract attributes such as liveness, is provided.
  • New built-in compiler matrix mathematical functions (Intrinsics) have been added, which, when compiled, are replaced by effective vector instructions.
  • Numerous improvements to the backends for the X86, AArch64, ARM, SystemZ, MIPS, AMDGPU, and PowerPC architectures. Added support for Cortex-A65, Cortex-A65AE, Neoverse E1 and Neoverse N1 CPUs. For ARMv8.1-M, ​​the code generation process has been optimized (for example, support for loops with minimal overhead has appeared) and support for auto-vectorization using the MVE extension has been added. Improved support for CPU MIPS Octeon. PowerPC includes vectorization of mathematical routines using the MASSV (Mathematical Acceleration SubSystem) library, improved code generation, and optimized memory access from loops. For x86, the processing of vector types v2i32, v4i16, v2i16, v8i8, v4i8 and v2i8 has been changed.
  • Improved code generator for WebAssembly. Added support for TLS (Thread-Local Storage) and atomic.fence instructions. Significantly expanded support for SIMD. WebAssembly object files added the ability to use function signatures with multiple values.
  • When processing cycles, the MemorySSA analyzer is used to determine the dependencies between different memory operations. MemorySSA can reduce compilation and execution time, or can be used instead of AliasSetTracker without sacrificing performance.
  • The LLDB debugger has significantly improved support for the DWARF v5 format. Improved build support with MinGW and added the initial ability to debug Windows executable files for ARM and ARM64 architectures. Added descriptions of options offered when autocompleting input by pressing tabs.
  • Enhanced LLD Linker Features. Improved support for the ELF format, including full compatibility of glob templates with the GNU linker, added support for the compressed debug sections ".zdebug", added the PT_GNU_PROPERTY property to determine the .note.gnu.property section (can be used in future Linux kernels), implemented modes "-z noseparate-code", "-z separate-code" and "-z separate-loadable-segments". Improved support for MinGW and WebAssembly.

Get more at the release notes.