Kbuild repository template for userland

The same code base is used for a different range of computing systems, from supercomputers to very tiny devices.

• Much simpler makefiles without the glue code that are hard to read and maintain
• Reduce the burden of code dependency management from developers
• Easy and efficient way to manage all compilation and configuration options on top of platform capabilities
• Precise dependency tracking with strong support of parallelism
• Tasks are split up and run simultaneously on multiple processors with different input in order to obtain results faster.
• Build versioning in continuous delivery
• Readable and configurable log

• Easy to change/browse configuration
• Clear dependency between features and capabilities
• Help docs in Kconfig rather than a README

scripts/cpucap.c is a dependency-free host probe that reports the CPU crypto-acceleration extensions of the build host. It is compiled with $(HOSTCC) on every make (at parse time) to $(objtree)/cpucap, and its --env output is exported so arch/*/Kconfig can read it via option env:

Enabling CPU_NATIVE then auto-selects the processor family matching the build host, pulling in its predefined capabilities (AES/PMULL/SHA-2, SHA-3, ...), and points the compiler at -march=native / -mcpu=native.

make cpucap          # report the host's capabilities
./obj/cpucap --env   # just the detected model

Consuming projects get this for free — no per-project Makefile glue. A project that wants to probe differently can export _CPUCAP_DONE plus the HOST_* values from its own top-level Makefile before including kbuild's; the built-in probe then stands down.

git clone git@github.com:n13l/crypto.git
cd crypto/
git submodule update --init