digest-blake3 0.37.0.1 → 1.2.0.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 4dc981436633bde6ba4fb278252d8a4a1ba58d039d0b1c8c794e36c4e47fa4a0
-  data.tar.gz: ae40be72a0252730792f3e82a00da765546c9606d91405ff69d3bad078ad307f
+  metadata.gz: ec7a77d6875b688e1cb1fbe8470cbf67278f9fe3f2f8e516bafe7abc0bf54bc4
+  data.tar.gz: 74e13b2480eccd5c2fe3fa913a0962217c1f07c95b5db80b8303086488ee5d9f
 SHA512:
-  metadata.gz: c18ca69b1f4b47ac8308ee00cc6db861eb48bd3921a85c291cb09ea595534b1a476988453c1931cb9982e9f96e7d14e4fa4356cbecaa9c584252c9b7ad30ac62
-  data.tar.gz: b9cf5f04daf5d83a797191caa2f2c30e068ddd6b771d887acb96963b93550171e77ea81128fffc316b49641391a4f245a8484b8b909d53502be3a2fb3170ad76
+  metadata.gz: de0fb7b5ccce755c313da8e547a430950d181170c64561746890ce8855ce5e09d3232b16316f36d22320ae5d23cf7904e8221a26358e96d9566ba247ef613214
+  data.tar.gz: 33e15e9469128ba227dbe6b57d9c44fe55078b9031975bf9db783a469c93342c7ccbf38b763ddfed7f09c941a42a6df89302cfda0e38b0ad4967a12acac4b18a

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    digest-blake3 (0.37.0)
+    digest-blake3 (1.2.0.0)
 
 GEM
   remote: https://rubygems.org/

data/ext/digest/blake3/blake3.c

@@ -5,6 +5,8 @@
 #include "blake3.h"
 #include "blake3_impl.h"
 
+const char *blake3_version(void) { return BLAKE3_VERSION_STRING; }
+
 INLINE void chunk_state_init(blake3_chunk_state *self, const uint32_t key[8],
                              uint8_t flags) {
   memcpy(self->cv, key, BLAKE3_KEY_LEN);
@@ -338,12 +340,18 @@ INLINE void compress_subtree_to_parent_node(
   uint8_t cv_array[MAX_SIMD_DEGREE_OR_2 * BLAKE3_OUT_LEN];
   size_t num_cvs = blake3_compress_subtree_wide(input, input_len, key,
                                                 chunk_counter, flags, cv_array);
+  assert(num_cvs <= MAX_SIMD_DEGREE_OR_2);
 
   // If MAX_SIMD_DEGREE is greater than 2 and there's enough input,
   // compress_subtree_wide() returns more than 2 chaining values. Condense
   // them into 2 by forming parent nodes repeatedly.
   uint8_t out_array[MAX_SIMD_DEGREE_OR_2 * BLAKE3_OUT_LEN / 2];
-  while (num_cvs > 2) {
+  // The second half of this loop condition is always true, and we just
+  // asserted it above. But GCC can't tell that it's always true, and if NDEBUG
+  // is set on platforms where MAX_SIMD_DEGREE_OR_2 == 2, GCC emits spurious
+  // warnings here. GCC 8.5 is particularly sensitive, so if you're changing
+  // this code, test it against that version.
+  while (num_cvs > 2 && num_cvs <= MAX_SIMD_DEGREE_OR_2) {
     num_cvs =
         compress_parents_parallel(cv_array, num_cvs, key, flags, out_array);
     memcpy(cv_array, out_array, num_cvs * BLAKE3_OUT_LEN);

data/ext/digest/blake3/blake3.h

@@ -8,12 +8,12 @@
 extern "C" {
 #endif
 
+#define BLAKE3_VERSION_STRING "1.2.0"
 #define BLAKE3_KEY_LEN 32
 #define BLAKE3_OUT_LEN 32
 #define BLAKE3_BLOCK_LEN 64
 #define BLAKE3_CHUNK_LEN 1024
 #define BLAKE3_MAX_DEPTH 54
-#define BLAKE3_MAX_SIMD_DEGREE 16
 
 // This struct is a private implementation detail. It has to be here because
 // it's part of blake3_hasher below.
@@ -38,11 +38,12 @@ typedef struct {
   uint8_t cv_stack[(BLAKE3_MAX_DEPTH + 1) * BLAKE3_OUT_LEN];
 } blake3_hasher;
 
+const char *blake3_version(void);
 void blake3_hasher_init(blake3_hasher *self);
 void blake3_hasher_init_keyed(blake3_hasher *self,
                               const uint8_t key[BLAKE3_KEY_LEN]);
 void blake3_hasher_init_derive_key(blake3_hasher *self, const char *context);
-void blake3_hasher_init_derive_key_raw(blake3_hasher *self, const void *context, 
+void blake3_hasher_init_derive_key_raw(blake3_hasher *self, const void *context,
                                        size_t context_len);
 void blake3_hasher_update(blake3_hasher *self, const void *input,
                           size_t input_len);

data/ext/digest/blake3/blake3_dispatch.c

@@ -14,6 +14,8 @@
 #endif
 #endif
 
+#define MAYBE_UNUSED(x) (void)((x))
+
 #if defined(IS_X86)
 static uint64_t xgetbv() {
 #if defined(_MSC_VER)
@@ -137,6 +139,7 @@ void blake3_compress_in_place(uint32_t cv[8],
                               uint8_t flags) {
 #if defined(IS_X86)
   const enum cpu_feature features = get_cpu_features();
+  MAYBE_UNUSED(features);
 #if !defined(BLAKE3_NO_AVX512)
   if (features & AVX512VL) {
     blake3_compress_in_place_avx512(cv, block, block_len, counter, flags);
@@ -165,6 +168,7 @@ void blake3_compress_xof(const uint32_t cv[8],
                          uint8_t out[64]) {
 #if defined(IS_X86)
   const enum cpu_feature features = get_cpu_features();
+  MAYBE_UNUSED(features);
 #if !defined(BLAKE3_NO_AVX512)
   if (features & AVX512VL) {
     blake3_compress_xof_avx512(cv, block, block_len, counter, flags, out);
@@ -193,6 +197,7 @@ void blake3_hash_many(const uint8_t *const *inputs, size_t num_inputs,
                       uint8_t flags_start, uint8_t flags_end, uint8_t *out) {
 #if defined(IS_X86)
   const enum cpu_feature features = get_cpu_features();
+  MAYBE_UNUSED(features);
 #if !defined(BLAKE3_NO_AVX512)
   if ((features & (AVX512F|AVX512VL)) == (AVX512F|AVX512VL)) {
     blake3_hash_many_avx512(inputs, num_inputs, blocks, key, counter,
@@ -227,7 +232,7 @@ void blake3_hash_many(const uint8_t *const *inputs, size_t num_inputs,
 #endif
 #endif
 
-#if defined(BLAKE3_USE_NEON)
+#if BLAKE3_USE_NEON == 1
   blake3_hash_many_neon(inputs, num_inputs, blocks, key, counter,
                         increment_counter, flags, flags_start, flags_end, out);
   return;
@@ -242,6 +247,7 @@ void blake3_hash_many(const uint8_t *const *inputs, size_t num_inputs,
 size_t blake3_simd_degree(void) {
 #if defined(IS_X86)
   const enum cpu_feature features = get_cpu_features();
+  MAYBE_UNUSED(features);
 #if !defined(BLAKE3_NO_AVX512)
   if ((features & (AVX512F|AVX512VL)) == (AVX512F|AVX512VL)) {
     return 16;
@@ -263,7 +269,7 @@ size_t blake3_simd_degree(void) {
   }
 #endif
 #endif
-#if defined(BLAKE3_USE_NEON)
+#if BLAKE3_USE_NEON == 1
   return 4;
 #endif
   return 1;

data/ext/digest/blake3/blake3_impl.h

@@ -38,6 +38,10 @@ enum blake3_flags {
 #define IS_X86_32
 #endif
 
+#if defined(__aarch64__) || defined(_M_ARM64)
+#define IS_AARCH64
+#endif
+
 #if defined(IS_X86)
 #if defined(_MSC_VER)
 #include <intrin.h>
@@ -45,9 +49,18 @@ enum blake3_flags {
 #include <immintrin.h>
 #endif
 
+#if !defined(BLAKE3_USE_NEON) 
+  // If BLAKE3_USE_NEON not manually set, autodetect based on AArch64ness
+  #if defined(IS_AARCH64)
+    #define BLAKE3_USE_NEON 1
+  #else
+    #define BLAKE3_USE_NEON 0
+  #endif
+#endif
+
 #if defined(IS_X86)
 #define MAX_SIMD_DEGREE 16
-#elif defined(BLAKE3_USE_NEON)
+#elif BLAKE3_USE_NEON == 1
 #define MAX_SIMD_DEGREE 4
 #else
 #define MAX_SIMD_DEGREE 1
@@ -257,7 +270,7 @@ void blake3_hash_many_avx512(const uint8_t *const *inputs, size_t num_inputs,
 #endif
 #endif
 
-#if defined(BLAKE3_USE_NEON)
+#if BLAKE3_USE_NEON == 1
 void blake3_hash_many_neon(const uint8_t *const *inputs, size_t num_inputs,
                            size_t blocks, const uint32_t key[8],
                            uint64_t counter, bool increment_counter,

data/ext/digest/blake3/blake3_neon.c

@@ -2,7 +2,12 @@
 
 #include <arm_neon.h>
 
-// TODO: This is probably incorrect for big-endian ARM. How should that work?
+#ifdef __ARM_BIG_ENDIAN
+#error "This implementation only supports little-endian ARM."
+// It might be that all we need for big-endian support here is to get the loads
+// and stores right, but step zero would be finding a way to test it in CI.
+#endif
+
 INLINE uint32x4_t loadu_128(const uint8_t src[16]) {
   // vld1q_u32 has alignment requirements. Don't use it.
   uint32x4_t x;

data/ext/digest/blake3/blake3_sse2_x86-64_unix.S

@@ -1704,7 +1704,7 @@ blake3_hash_many_sse2:
         pshufd  xmm15, xmm11, 0x93
         shl     rax, 0x20
         or      rax, 0x40
-        movd    xmm3, rax
+        movq    xmm3, rax
         movdqa  xmmword ptr [rsp+0x20], xmm3
         movaps  xmm3, xmmword ptr [rsp]
         movaps  xmm11, xmmword ptr [rsp+0x10]
@@ -1917,7 +1917,7 @@ blake3_hash_many_sse2:
         movaps  xmm2, xmmword ptr [BLAKE3_IV+rip]
         shl     rax, 32
         or      rax, 64
-        movd    xmm12, rax
+        movq    xmm12, rax
         movdqa  xmm3, xmm13
         punpcklqdq xmm3, xmm12
         movups  xmm4, xmmword ptr [r8+rdx-0x40]

data/ext/digest/blake3/blake3_sse2_x86-64_windows_gnu.S

@@ -1715,7 +1715,7 @@ blake3_hash_many_sse2:
         pshufd  xmm15, xmm11, 0x93
         shl     rax, 0x20
         or      rax, 0x40
-        movd    xmm3, rax
+        movq    xmm3, rax
         movdqa  xmmword ptr [rsp+0x20], xmm3
         movaps  xmm3, xmmword ptr [rsp]
         movaps  xmm11, xmmword ptr [rsp+0x10]
@@ -1928,7 +1928,7 @@ blake3_hash_many_sse2:
         movaps  xmm2, xmmword ptr [BLAKE3_IV+rip]
         shl     rax, 32
         or      rax, 64
-        movd    xmm12, rax
+        movq    xmm12, rax
         movdqa  xmm3, xmm13
         punpcklqdq xmm3, xmm12
         movups  xmm4, xmmword ptr [r8+rdx-0x40]
@@ -2137,10 +2137,10 @@ _blake3_compress_in_place_sse2:
         por     xmm9, xmm8
         movdqa  xmm8, xmm7
         punpcklqdq xmm8, xmm5
-        movdqa  xmm10, xmm6
+        movdqa  xmm14, xmm6
         pand    xmm8, xmmword ptr [PBLENDW_0x3F_MASK+rip]
-        pand    xmm10, xmmword ptr [PBLENDW_0xC0_MASK+rip]
-        por     xmm8, xmm10
+        pand    xmm14, xmmword ptr [PBLENDW_0xC0_MASK+rip]
+        por     xmm8, xmm14
         pshufd  xmm8, xmm8, 0x78
         punpckhdq xmm5, xmm7
         punpckldq xmm6, xmm5
@@ -2268,10 +2268,10 @@ blake3_compress_xof_sse2:
         por     xmm9, xmm8
         movdqa  xmm8, xmm7
         punpcklqdq xmm8, xmm5
-        movdqa  xmm10, xmm6
+        movdqa  xmm14, xmm6
         pand    xmm8, xmmword ptr [PBLENDW_0x3F_MASK+rip]
-        pand    xmm10, xmmword ptr [PBLENDW_0xC0_MASK+rip]
-        por     xmm8, xmm10
+        pand    xmm14, xmmword ptr [PBLENDW_0xC0_MASK+rip]
+        por     xmm8, xmm14
         pshufd  xmm8, xmm8, 0x78
         punpckhdq xmm5, xmm7
         punpckldq xmm6, xmm5

data/ext/digest/blake3/blake3_sse2_x86-64_windows_msvc.asm

@@ -2054,8 +2054,8 @@ _blake3_compress_in_place_sse2 PROC
         movzx   r8d, r8b
         shl     rax, 32
         add     r8, rax
-        movq    xmm3, r9
-        movq    xmm4, r8
+        movd    xmm3, r9
+        movd    xmm4, r8
         punpcklqdq xmm3, xmm4
         movups  xmm4, xmmword ptr [rdx]
         movups  xmm5, xmmword ptr [rdx+10H]
@@ -2139,10 +2139,10 @@ _blake3_compress_in_place_sse2 PROC
         por     xmm9, xmm8
         movdqa  xmm8, xmm7
         punpcklqdq xmm8, xmm5
-        movdqa  xmm10, xmm6
+        movdqa  xmm14, xmm6
         pand    xmm8, xmmword ptr [PBLENDW_0x3F_MASK]
-        pand    xmm10, xmmword ptr [PBLENDW_0xC0_MASK]
-        por     xmm8, xmm10
+        pand    xmm14, xmmword ptr [PBLENDW_0xC0_MASK]
+        por     xmm8, xmm14
         pshufd  xmm8, xmm8, 78H
         punpckhdq xmm5, xmm7
         punpckldq xmm6, xmm5
@@ -2186,8 +2186,8 @@ _blake3_compress_xof_sse2 PROC
         mov     r10, qword ptr [rsp+0A8H]
         shl     rax, 32
         add     r8, rax
-        movq    xmm3, r9
-        movq    xmm4, r8
+        movd    xmm3, r9
+        movd    xmm4, r8
         punpcklqdq xmm3, xmm4
         movups  xmm4, xmmword ptr [rdx]
         movups  xmm5, xmmword ptr [rdx+10H]
@@ -2271,10 +2271,10 @@ _blake3_compress_xof_sse2 PROC
         por     xmm9, xmm8
         movdqa  xmm8, xmm7
         punpcklqdq xmm8, xmm5
-        movdqa  xmm10, xmm6
+        movdqa  xmm14, xmm6
         pand    xmm8, xmmword ptr [PBLENDW_0x3F_MASK]
-        pand    xmm10, xmmword ptr [PBLENDW_0xC0_MASK]
-        por     xmm8, xmm10
+        pand    xmm14, xmmword ptr [PBLENDW_0xC0_MASK]
+        por     xmm8, xmm14
         pshufd  xmm8, xmm8, 78H
         punpckhdq xmm5, xmm7
         punpckldq xmm6, xmm5

data/ext/digest/blake3/blake3_sse41_x86-64_windows_msvc.asm

@@ -1817,8 +1817,8 @@ _blake3_compress_in_place_sse41 PROC
         movzx   r8d, r8b
         shl     rax, 32
         add     r8, rax
-        movq    xmm3, r9
-        movq    xmm4, r8
+        movd    xmm3, r9
+        movd    xmm4, r8
         punpcklqdq xmm3, xmm4
         movups  xmm4, xmmword ptr [rdx]
         movups  xmm5, xmmword ptr [rdx+10H]
@@ -1938,8 +1938,8 @@ _blake3_compress_xof_sse41 PROC
         mov     r10, qword ptr [rsp+0A8H]
         shl     rax, 32
         add     r8, rax
-        movq    xmm3, r9
-        movq    xmm4, r8
+        movd    xmm3, r9
+        movd    xmm4, r8
         punpcklqdq xmm3, xmm4
         movups  xmm4, xmmword ptr [rdx]
         movups  xmm5, xmmword ptr [rdx+10H]

data/lib/digest/blake3/version.rb

@@ -2,6 +2,6 @@ require 'digest'
 
 module Digest
   class BLAKE3 < Base
-    VERSION = "0.37.0.1"
+    VERSION = "1.2.0.0"
   end
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: digest-blake3
 version: !ruby/object:Gem::Version
-  version: 0.37.0.1
+  version: 1.2.0.0
 platform: ruby
 authors:
 - Will Bryant
 autorequire:
 bindir: exe
 cert_chain: []
-date: 2020-10-18 00:00:00.000000000 Z
+date: 2022-01-07 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bundler