diff options
| author | Jonathan Protzenko <protz@microsoft.com> | 2024-08-13 21:42:19 (GMT) |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2024-08-13 21:42:19 (GMT) |
| commit | 325e9b8ef400b86fb077aa40d5cb8cec6e4df7bb (patch) | |
| tree | 9c1677183d3a508207f05097e6751edce520d35e /Modules/_hacl/include | |
| parent | ee1b8ce26e700350e47a5f65201097121c41912e (diff) | |
| download | cpython-325e9b8ef400b86fb077aa40d5cb8cec6e4df7bb.zip cpython-325e9b8ef400b86fb077aa40d5cb8cec6e4df7bb.tar.gz cpython-325e9b8ef400b86fb077aa40d5cb8cec6e4df7bb.tar.bz2 | |
gh-99108: Add HACL* Blake2 implementation to hashlib (GH-119316)
This replaces the existing hashlib Blake2 module with a single implementation that uses HACL\*'s Blake2b/Blake2s implementations. We added support for all the modes exposed by the Python API, including tree hashing, leaf nodes, and so on. We ported and merged all of these changes upstream in HACL\*, added test vectors based on Python's existing implementation, and exposed everything needed for hashlib.
This was joint work done with @R1kM.
See the PR for much discussion and benchmarking details. TL;DR: On many systems, 8-50% faster (!) than `libb2`, on some systems it appeared 10-20% slower than `libb2`.
Diffstat (limited to 'Modules/_hacl/include')
| -rw-r--r-- | Modules/_hacl/include/krml/internal/target.h | 69 |
1 files changed, 69 insertions, 0 deletions
diff --git a/Modules/_hacl/include/krml/internal/target.h b/Modules/_hacl/include/krml/internal/target.h index c7fcc01..292adc1 100644 --- a/Modules/_hacl/include/krml/internal/target.h +++ b/Modules/_hacl/include/krml/internal/target.h @@ -69,6 +69,14 @@ # endif #endif +#ifndef KRML_ATTRIBUTE_TARGET +# if defined(__GNUC__) +# define KRML_ATTRIBUTE_TARGET(x) __attribute__((target(x))) +# else +# define KRML_ATTRIBUTE_TARGET(x) +# endif +#endif + #ifndef KRML_NOINLINE # if defined(_MSC_VER) # define KRML_NOINLINE __declspec(noinline) @@ -82,6 +90,67 @@ # endif #endif +#ifndef KRML_MUSTINLINE +# if defined(_MSC_VER) +# define KRML_MUSTINLINE inline __forceinline +# elif defined (__GNUC__) +# define KRML_MUSTINLINE inline __attribute__((always_inline)) +# else +# define KRML_MUSTINLINE inline +# warning "The KRML_MUSTINLINE macro defaults to plain inline for this toolchain!" +# warning "Please locate target.h and try to fill it out with a suitable definition for this compiler." +# endif +#endif + +#ifndef KRML_PRE_ALIGN +# ifdef _MSC_VER +# define KRML_PRE_ALIGN(X) __declspec(align(X)) +# else +# define KRML_PRE_ALIGN(X) +# endif +#endif + +#ifndef KRML_POST_ALIGN +# ifdef _MSC_VER +# define KRML_POST_ALIGN(X) +# else +# define KRML_POST_ALIGN(X) __attribute__((aligned(X))) +# endif +#endif + +/* MinGW-W64 does not support C11 aligned_alloc, but it supports + * MSVC's _aligned_malloc. + */ +#ifndef KRML_ALIGNED_MALLOC +# ifdef __MINGW32__ +# include <_mingw.h> +# endif +# if ( \ + defined(_MSC_VER) || \ + (defined(__MINGW32__) && defined(__MINGW64_VERSION_MAJOR))) +# define KRML_ALIGNED_MALLOC(X, Y) _aligned_malloc(Y, X) +# else +# define KRML_ALIGNED_MALLOC(X, Y) aligned_alloc(X, Y) +# endif +#endif + +/* Since aligned allocations with MinGW-W64 are done with + * _aligned_malloc (see above), such pointers must be freed with + * _aligned_free. + */ +#ifndef KRML_ALIGNED_FREE +# ifdef __MINGW32__ +# include <_mingw.h> +# endif +# if ( \ + defined(_MSC_VER) || \ + (defined(__MINGW32__) && defined(__MINGW64_VERSION_MAJOR))) +# define KRML_ALIGNED_FREE(X) _aligned_free(X) +# else +# define KRML_ALIGNED_FREE(X) free(X) +# endif +#endif + /* In FStar.Buffer.fst, the size of arrays is uint32_t, but it's a number of * *elements*. Do an ugly, run-time check (some of which KaRaMeL can eliminate). */ |