From 087c8bf3303e69d79cbbfdfb8edc836587cfb820 Mon Sep 17 00:00:00 2001
From: marshmellow42 <marshmellowrf@gmail.com>
Date: Thu, 8 Jun 2017 17:07:14 -0400
Subject: [PATCH 1/1] fix compile issues on OS X 10.11
OSX 10.11 does not have clock_gettime()
clang <= 8.0.0 has a bug in __builtin_cpu_supports() and it doesn't
function.
see https://llvm.org/bugs/show_bug.cgi?id=25510
---
client/cmdhfmfhard.c | 11 +-
client/hardnested/hardnested_bf_core.c | 42 +++---
client/hardnested/hardnested_bitarray_core.c | 134 +++++++++++--------
client/util_posix.c | 100 +++++++++++---
4 files changed, 195 insertions(+), 92 deletions(-)
diff --git a/client/cmdhfmfhard.c b/client/cmdhfmfhard.c
index 6cd75af5..b5eabb82 100644
--- a/client/cmdhfmfhard.c
+++ b/client/cmdhfmfhard.c
@@ -70,16 +70,19 @@ static float brute_force_per_second;
static void get_SIMD_instruction_set(char* instruction_set) {
- #if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
+ #if !defined(__APPLE__) || (defined(__APPLE__) && (__clang_major__ > 8))
+ #if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
if (__builtin_cpu_supports("avx512f")) strcpy(instruction_set, "AVX512F");
else if (__builtin_cpu_supports("avx2")) strcpy(instruction_set, "AVX2");
- #else
+ #else
if (__builtin_cpu_supports("avx2")) strcpy(instruction_set, "AVX2");
- #endif
+ #endif
else if (__builtin_cpu_supports("avx")) strcpy(instruction_set, "AVX");
else if (__builtin_cpu_supports("sse2")) strcpy(instruction_set, "SSE2");
else if (__builtin_cpu_supports("mmx")) strcpy(instruction_set, "MMX");
- else strcpy(instruction_set, "unsupported");
+ else
+ #endif
+ strcpy(instruction_set, "unsupported");
}
diff --git a/client/hardnested/hardnested_bf_core.c b/client/hardnested/hardnested_bf_core.c
index 05246296..2388f6f5 100644
--- a/client/hardnested/hardnested_bf_core.c
+++ b/client/hardnested/hardnested_bf_core.c
@@ -550,17 +550,19 @@ bitslice_test_nonces_t *bitslice_test_nonces_function_p = &bitslice_test_nonces_
// determine the available instruction set at runtime and call the correct function
const uint64_t crack_states_bitsliced_dispatch(uint32_t cuid, uint8_t *best_first_bytes, statelist_t *p, uint32_t *keys_found, uint64_t *num_keys_tested, uint32_t nonces_to_bruteforce, uint8_t *bf_test_nonce_2nd_byte, noncelist_t *nonces) {
-#if defined (__i386__) || defined (__x86_64__)
- #if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
- if (__builtin_cpu_supports("avx512f")) crack_states_bitsliced_function_p = &crack_states_bitsliced_AVX512;
- else if (__builtin_cpu_supports("avx2")) crack_states_bitsliced_function_p = &crack_states_bitsliced_AVX2;
- #else
- if (__builtin_cpu_supports("avx2")) crack_states_bitsliced_function_p = &crack_states_bitsliced_AVX2;
+#if defined (__i386__) || defined (__x86_64__)
+ #if !defined(__APPLE__) || (defined(__APPLE__) && (__clang_major__ > 8))
+ #if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
+ if (__builtin_cpu_supports("avx512f")) crack_states_bitsliced_function_p = &crack_states_bitsliced_AVX512;
+ else if (__builtin_cpu_supports("avx2")) crack_states_bitsliced_function_p = &crack_states_bitsliced_AVX2;
+ #else
+ if (__builtin_cpu_supports("avx2")) crack_states_bitsliced_function_p = &crack_states_bitsliced_AVX2;
+ #endif
+ else if (__builtin_cpu_supports("avx")) crack_states_bitsliced_function_p = &crack_states_bitsliced_AVX;
+ else if (__builtin_cpu_supports("sse2")) crack_states_bitsliced_function_p = &crack_states_bitsliced_SSE2;
+ else if (__builtin_cpu_supports("mmx")) crack_states_bitsliced_function_p = &crack_states_bitsliced_MMX;
+ else
#endif
- else if (__builtin_cpu_supports("avx")) crack_states_bitsliced_function_p = &crack_states_bitsliced_AVX;
- else if (__builtin_cpu_supports("sse2")) crack_states_bitsliced_function_p = &crack_states_bitsliced_SSE2;
- else if (__builtin_cpu_supports("mmx")) crack_states_bitsliced_function_p = &crack_states_bitsliced_MMX;
- else
#endif
crack_states_bitsliced_function_p = &crack_states_bitsliced_NOSIMD;
@@ -570,16 +572,18 @@ const uint64_t crack_states_bitsliced_dispatch(uint32_t cuid, uint8_t *best_firs
void bitslice_test_nonces_dispatch(uint32_t nonces_to_bruteforce, uint32_t *bf_test_nonce, uint8_t *bf_test_nonce_par) {
#if defined (__i386__) || defined (__x86_64__)
- #if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
- if (__builtin_cpu_supports("avx512f")) bitslice_test_nonces_function_p = &bitslice_test_nonces_AVX512;
- else if (__builtin_cpu_supports("avx2")) bitslice_test_nonces_function_p = &bitslice_test_nonces_AVX2;
- #else
- if (__builtin_cpu_supports("avx2")) bitslice_test_nonces_function_p = &bitslice_test_nonces_AVX2;
+ #if !defined(__APPLE__) || (defined(__APPLE__) && (__clang_major__ > 8))
+ #if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
+ if (__builtin_cpu_supports("avx512f")) bitslice_test_nonces_function_p = &bitslice_test_nonces_AVX512;
+ else if (__builtin_cpu_supports("avx2")) bitslice_test_nonces_function_p = &bitslice_test_nonces_AVX2;
+ #else
+ if (__builtin_cpu_supports("avx2")) bitslice_test_nonces_function_p = &bitslice_test_nonces_AVX2;
+ #endif
+ else if (__builtin_cpu_supports("avx")) bitslice_test_nonces_function_p = &bitslice_test_nonces_AVX;
+ else if (__builtin_cpu_supports("sse2")) bitslice_test_nonces_function_p = &bitslice_test_nonces_SSE2;
+ else if (__builtin_cpu_supports("mmx")) bitslice_test_nonces_function_p = &bitslice_test_nonces_MMX;
+ else
#endif
- else if (__builtin_cpu_supports("avx")) bitslice_test_nonces_function_p = &bitslice_test_nonces_AVX;
- else if (__builtin_cpu_supports("sse2")) bitslice_test_nonces_function_p = &bitslice_test_nonces_SSE2;
- else if (__builtin_cpu_supports("mmx")) bitslice_test_nonces_function_p = &bitslice_test_nonces_MMX;
- else
#endif
bitslice_test_nonces_function_p = &bitslice_test_nonces_NOSIMD;
diff --git a/client/hardnested/hardnested_bitarray_core.c b/client/hardnested/hardnested_bitarray_core.c
index 320c4a96..5615d006 100644
--- a/client/hardnested/hardnested_bitarray_core.c
+++ b/client/hardnested/hardnested_bitarray_core.c
@@ -319,16 +319,18 @@ count_bitarray_AND4_t *count_bitarray_AND4_function_p = &count_bitarray_AND4_dis
// determine the available instruction set at runtime and call the correct function
uint32_t *malloc_bitarray_dispatch(uint32_t x) {
#if defined (__i386__) || defined (__x86_64__)
- #if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
+ #if !defined(__APPLE__) || (defined(__APPLE__) && (__clang_major__ > 8))
+ #if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
if (__builtin_cpu_supports("avx512f")) malloc_bitarray_function_p = &malloc_bitarray_AVX512;
else if (__builtin_cpu_supports("avx2")) malloc_bitarray_function_p = &malloc_bitarray_AVX2;
- #else
+ #else
if (__builtin_cpu_supports("avx2")) malloc_bitarray_function_p = &malloc_bitarray_AVX2;
- #endif
+ #endif
else if (__builtin_cpu_supports("avx")) malloc_bitarray_function_p = &malloc_bitarray_AVX;
else if (__builtin_cpu_supports("sse2")) malloc_bitarray_function_p = &malloc_bitarray_SSE2;
else if (__builtin_cpu_supports("mmx")) malloc_bitarray_function_p = &malloc_bitarray_MMX;
- else
+ else
+ #endif
#endif
malloc_bitarray_function_p = &malloc_bitarray_NOSIMD;
@@ -337,17 +339,19 @@ uint32_t *malloc_bitarray_dispatch(uint32_t x) {
}
void free_bitarray_dispatch(uint32_t *x) {
-#if defined (__i386__) || defined (__x86_64__)
- #if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
+#if defined (__i386__) || defined (__x86_64__)
+ #if !defined(__APPLE__) || (defined(__APPLE__) && (__clang_major__ > 8))
+ #if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
if (__builtin_cpu_supports("avx512f")) free_bitarray_function_p = &free_bitarray_AVX512;
else if (__builtin_cpu_supports("avx2")) free_bitarray_function_p = &free_bitarray_AVX2;
- #else
+ #else
if (__builtin_cpu_supports("avx2")) free_bitarray_function_p = &free_bitarray_AVX2;
- #endif
+ #endif
else if (__builtin_cpu_supports("avx")) free_bitarray_function_p = &free_bitarray_AVX;
else if (__builtin_cpu_supports("sse2")) free_bitarray_function_p = &free_bitarray_SSE2;
else if (__builtin_cpu_supports("mmx")) free_bitarray_function_p = &free_bitarray_MMX;
- else
+ else
+ #endif
#endif
free_bitarray_function_p = &free_bitarray_NOSIMD;
@@ -357,16 +361,18 @@ void free_bitarray_dispatch(uint32_t *x) {
uint32_t bitcount_dispatch(uint32_t a) {
#if defined (__i386__) || defined (__x86_64__)
- #if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
+ #if !defined(__APPLE__) || (defined(__APPLE__) && (__clang_major__ > 8))
+ #if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
if (__builtin_cpu_supports("avx512f")) bitcount_function_p = &bitcount_AVX512;
else if (__builtin_cpu_supports("avx2")) bitcount_function_p = &bitcount_AVX2;
- #else
+ #else
if (__builtin_cpu_supports("avx2")) bitcount_function_p = &bitcount_AVX2;
- #endif
+ #endif
else if (__builtin_cpu_supports("avx")) bitcount_function_p = &bitcount_AVX;
else if (__builtin_cpu_supports("sse2")) bitcount_function_p = &bitcount_SSE2;
else if (__builtin_cpu_supports("mmx")) bitcount_function_p = &bitcount_MMX;
- else
+ else
+ #endif
#endif
bitcount_function_p = &bitcount_NOSIMD;
@@ -376,16 +382,18 @@ uint32_t bitcount_dispatch(uint32_t a) {
uint32_t count_states_dispatch(uint32_t *bitarray) {
#if defined (__i386__) || defined (__x86_64__)
- #if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
+ #if !defined(__APPLE__) || (defined(__APPLE__) && (__clang_major__ > 8))
+ #if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
if (__builtin_cpu_supports("avx512f")) count_states_function_p = &count_states_AVX512;
else if (__builtin_cpu_supports("avx2")) count_states_function_p = &count_states_AVX2;
- #else
+ #else
if (__builtin_cpu_supports("avx2")) count_states_function_p = &count_states_AVX2;
- #endif
+ #endif
else if (__builtin_cpu_supports("avx")) count_states_function_p = &count_states_AVX;
else if (__builtin_cpu_supports("sse2")) count_states_function_p = &count_states_SSE2;
else if (__builtin_cpu_supports("mmx")) count_states_function_p = &count_states_MMX;
- else
+ else
+ #endif
#endif
count_states_function_p = &count_states_NOSIMD;
@@ -395,16 +403,18 @@ uint32_t count_states_dispatch(uint32_t *bitarray) {
void bitarray_AND_dispatch(uint32_t *A, uint32_t *B) {
#if defined (__i386__) || defined (__x86_64__)
- #if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
+ #if !defined(__APPLE__) || (defined(__APPLE__) && (__clang_major__ > 8))
+ #if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
if (__builtin_cpu_supports("avx512f")) bitarray_AND_function_p = &bitarray_AND_AVX512;
else if (__builtin_cpu_supports("avx2")) bitarray_AND_function_p = &bitarray_AND_AVX2;
- #else
+ #else
if (__builtin_cpu_supports("avx2")) bitarray_AND_function_p = &bitarray_AND_AVX2;
- #endif
+ #endif
else if (__builtin_cpu_supports("avx")) bitarray_AND_function_p = &bitarray_AND_AVX;
else if (__builtin_cpu_supports("sse2")) bitarray_AND_function_p = &bitarray_AND_SSE2;
else if (__builtin_cpu_supports("mmx")) bitarray_AND_function_p = &bitarray_AND_MMX;
- else
+ else
+ #endif
#endif
bitarray_AND_function_p = &bitarray_AND_NOSIMD;
@@ -413,17 +423,19 @@ void bitarray_AND_dispatch(uint32_t *A, uint32_t *B) {
}
void bitarray_low20_AND_dispatch(uint32_t *A, uint32_t *B) {
-#if defined (__i386__) || defined (__x86_64__)
- #if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
+#if defined (__i386__) || defined (__x86_64__)
+ #if !defined(__APPLE__) || (defined(__APPLE__) && (__clang_major__ > 8))
+ #if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
if (__builtin_cpu_supports("avx512f")) bitarray_low20_AND_function_p = &bitarray_low20_AND_AVX512;
else if (__builtin_cpu_supports("avx2")) bitarray_low20_AND_function_p = &bitarray_low20_AND_AVX2;
- #else
+ #else
if (__builtin_cpu_supports("avx2")) bitarray_low20_AND_function_p = &bitarray_low20_AND_AVX2;
- #endif
+ #endif
else if (__builtin_cpu_supports("avx")) bitarray_low20_AND_function_p = &bitarray_low20_AND_AVX;
else if (__builtin_cpu_supports("sse2")) bitarray_low20_AND_function_p = &bitarray_low20_AND_SSE2;
else if (__builtin_cpu_supports("mmx")) bitarray_low20_AND_function_p = &bitarray_low20_AND_MMX;
- else
+ else
+ #endif
#endif
bitarray_low20_AND_function_p = &bitarray_low20_AND_NOSIMD;
@@ -433,16 +445,18 @@ void bitarray_low20_AND_dispatch(uint32_t *A, uint32_t *B) {
uint32_t count_bitarray_AND_dispatch(uint32_t *A, uint32_t *B) {
#if defined (__i386__) || defined (__x86_64__)
- #if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
+ #if !defined(__APPLE__) || (defined(__APPLE__) && (__clang_major__ > 8))
+ #if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
if (__builtin_cpu_supports("avx512f")) count_bitarray_AND_function_p = &count_bitarray_AND_AVX512;
else if (__builtin_cpu_supports("avx2")) count_bitarray_AND_function_p = &count_bitarray_AND_AVX2;
- #else
+ #else
if (__builtin_cpu_supports("avx2")) count_bitarray_AND_function_p = &count_bitarray_AND_AVX2;
- #endif
+ #endif
else if (__builtin_cpu_supports("avx")) count_bitarray_AND_function_p = &count_bitarray_AND_AVX;
else if (__builtin_cpu_supports("sse2")) count_bitarray_AND_function_p = &count_bitarray_AND_SSE2;
else if (__builtin_cpu_supports("mmx")) count_bitarray_AND_function_p = &count_bitarray_AND_MMX;
- else
+ else
+ #endif
#endif
count_bitarray_AND_function_p = &count_bitarray_AND_NOSIMD;
@@ -452,16 +466,18 @@ uint32_t count_bitarray_AND_dispatch(uint32_t *A, uint32_t *B) {
uint32_t count_bitarray_low20_AND_dispatch(uint32_t *A, uint32_t *B) {
#if defined (__i386__) || defined (__x86_64__)
- #if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
+ #if !defined(__APPLE__) || (defined(__APPLE__) && (__clang_major__ > 8))
+ #if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
if (__builtin_cpu_supports("avx512f")) count_bitarray_low20_AND_function_p = &count_bitarray_low20_AND_AVX512;
else if (__builtin_cpu_supports("avx2")) count_bitarray_low20_AND_function_p = &count_bitarray_low20_AND_AVX2;
- #else
+ #else
if (__builtin_cpu_supports("avx2")) count_bitarray_low20_AND_function_p = &count_bitarray_low20_AND_AVX2;
- #endif
+ #endif
else if (__builtin_cpu_supports("avx")) count_bitarray_low20_AND_function_p = &count_bitarray_low20_AND_AVX;
else if (__builtin_cpu_supports("sse2")) count_bitarray_low20_AND_function_p = &count_bitarray_low20_AND_SSE2;
else if (__builtin_cpu_supports("mmx")) count_bitarray_low20_AND_function_p = &count_bitarray_low20_AND_MMX;
- else
+ else
+ #endif
#endif
count_bitarray_low20_AND_function_p = &count_bitarray_low20_AND_NOSIMD;
@@ -471,16 +487,18 @@ uint32_t count_bitarray_low20_AND_dispatch(uint32_t *A, uint32_t *B) {
void bitarray_AND4_dispatch(uint32_t *A, uint32_t *B, uint32_t *C, uint32_t *D) {
#if defined (__i386__) || defined (__x86_64__)
- #if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
+ #if !defined(__APPLE__) || (defined(__APPLE__) && (__clang_major__ > 8))
+ #if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
if (__builtin_cpu_supports("avx512f")) bitarray_AND4_function_p = &bitarray_AND4_AVX512;
else if (__builtin_cpu_supports("avx2")) bitarray_AND4_function_p = &bitarray_AND4_AVX2;
- #else
+ #else
if (__builtin_cpu_supports("avx2")) bitarray_AND4_function_p = &bitarray_AND4_AVX2;
- #endif
+ #endif
else if (__builtin_cpu_supports("avx")) bitarray_AND4_function_p = &bitarray_AND4_AVX;
else if (__builtin_cpu_supports("sse2")) bitarray_AND4_function_p = &bitarray_AND4_SSE2;
else if (__builtin_cpu_supports("mmx")) bitarray_AND4_function_p = &bitarray_AND4_MMX;
- else
+ else
+ #endif
#endif
bitarray_AND4_function_p = &bitarray_AND4_NOSIMD;
@@ -490,16 +508,18 @@ void bitarray_AND4_dispatch(uint32_t *A, uint32_t *B, uint32_t *C, uint32_t *D)
void bitarray_OR_dispatch(uint32_t *A, uint32_t *B) {
#if defined (__i386__) || defined (__x86_64__)
- #if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
+ #if !defined(__APPLE__) || (defined(__APPLE__) && (__clang_major__ > 8))
+ #if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
if (__builtin_cpu_supports("avx512f")) bitarray_OR_function_p = &bitarray_OR_AVX512;
else if (__builtin_cpu_supports("avx2")) bitarray_OR_function_p = &bitarray_OR_AVX2;
- #else
+ #else
if (__builtin_cpu_supports("avx2")) bitarray_OR_function_p = &bitarray_OR_AVX2;
- #endif
+ #endif
else if (__builtin_cpu_supports("avx")) bitarray_OR_function_p = &bitarray_OR_AVX;
else if (__builtin_cpu_supports("sse2")) bitarray_OR_function_p = &bitarray_OR_SSE2;
else if (__builtin_cpu_supports("mmx")) bitarray_OR_function_p = &bitarray_OR_MMX;
- else
+ else
+ #endif
#endif
bitarray_OR_function_p = &bitarray_OR_NOSIMD;
@@ -509,16 +529,18 @@ void bitarray_OR_dispatch(uint32_t *A, uint32_t *B) {
uint32_t count_bitarray_AND2_dispatch(uint32_t *A, uint32_t *B) {
#if defined (__i386__) || defined (__x86_64__)
- #if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
+ #if !defined(__APPLE__) || (defined(__APPLE__) && (__clang_major__ > 8))
+ #if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
if (__builtin_cpu_supports("avx512f")) count_bitarray_AND2_function_p = &count_bitarray_AND2_AVX512;
else if (__builtin_cpu_supports("avx2")) count_bitarray_AND2_function_p = &count_bitarray_AND2_AVX2;
- #else
+ #else
if (__builtin_cpu_supports("avx2")) count_bitarray_AND2_function_p = &count_bitarray_AND2_AVX2;
- #endif
+ #endif
else if (__builtin_cpu_supports("avx")) count_bitarray_AND2_function_p = &count_bitarray_AND2_AVX;
else if (__builtin_cpu_supports("sse2")) count_bitarray_AND2_function_p = &count_bitarray_AND2_SSE2;
else if (__builtin_cpu_supports("mmx")) count_bitarray_AND2_function_p = &count_bitarray_AND2_MMX;
- else
+ else
+ #endif
#endif
count_bitarray_AND2_function_p = &count_bitarray_AND2_NOSIMD;
@@ -528,16 +550,18 @@ uint32_t count_bitarray_AND2_dispatch(uint32_t *A, uint32_t *B) {
uint32_t count_bitarray_AND3_dispatch(uint32_t *A, uint32_t *B, uint32_t *C) {
#if defined (__i386__) || defined (__x86_64__)
- #if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
+ #if !defined(__APPLE__) || (defined(__APPLE__) && (__clang_major__ > 8))
+ #if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
if (__builtin_cpu_supports("avx512f")) count_bitarray_AND3_function_p = &count_bitarray_AND3_AVX512;
else if (__builtin_cpu_supports("avx2")) count_bitarray_AND3_function_p = &count_bitarray_AND3_AVX2;
- #else
+ #else
if (__builtin_cpu_supports("avx2")) count_bitarray_AND3_function_p = &count_bitarray_AND3_AVX2;
- #endif
+ #endif
else if (__builtin_cpu_supports("avx")) count_bitarray_AND3_function_p = &count_bitarray_AND3_AVX;
else if (__builtin_cpu_supports("sse2")) count_bitarray_AND3_function_p = &count_bitarray_AND3_SSE2;
else if (__builtin_cpu_supports("mmx")) count_bitarray_AND3_function_p = &count_bitarray_AND3_MMX;
- else
+ else
+ #endif
#endif
count_bitarray_AND3_function_p = &count_bitarray_AND3_NOSIMD;
@@ -547,16 +571,18 @@ uint32_t count_bitarray_AND3_dispatch(uint32_t *A, uint32_t *B, uint32_t *C) {
uint32_t count_bitarray_AND4_dispatch(uint32_t *A, uint32_t *B, uint32_t *C, uint32_t *D) {
#if defined (__i386__) || defined (__x86_64__)
- #if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
+ #if !defined(__APPLE__) || (defined(__APPLE__) && (__clang_major__ > 8))
+ #if (__GNUC__ >= 5) && (__GNUC__ > 5 || __GNUC_MINOR__ > 2)
if (__builtin_cpu_supports("avx512f")) count_bitarray_AND4_function_p = &count_bitarray_AND4_AVX512;
else if (__builtin_cpu_supports("avx2")) count_bitarray_AND4_function_p = &count_bitarray_AND4_AVX2;
- #else
+ #else
if (__builtin_cpu_supports("avx2")) count_bitarray_AND4_function_p = &count_bitarray_AND4_AVX2;
- #endif
+ #endif
else if (__builtin_cpu_supports("avx")) count_bitarray_AND4_function_p = &count_bitarray_AND4_AVX;
else if (__builtin_cpu_supports("sse2")) count_bitarray_AND4_function_p = &count_bitarray_AND4_SSE2;
else if (__builtin_cpu_supports("mmx")) count_bitarray_AND4_function_p = &count_bitarray_AND4_MMX;
- else
+ else
+ #endif
#endif
count_bitarray_AND4_function_p = &count_bitarray_AND4_NOSIMD;
diff --git a/client/util_posix.c b/client/util_posix.c
index 3e61b674..382f6a60 100644
--- a/client/util_posix.c
+++ b/client/util_posix.c
@@ -24,10 +24,10 @@
#include <errno.h>
static void nsleep(uint64_t n) {
- struct timespec timeout;
- timeout.tv_sec = n/1000000000;
- timeout.tv_nsec = n%1000000000;
- while (nanosleep(&timeout, &timeout) && errno == EINTR);
+ struct timespec timeout;
+ timeout.tv_sec = n/1000000000;
+ timeout.tv_nsec = n%1000000000;
+ while (nanosleep(&timeout, &timeout) && errno == EINTR);
}
void msleep(uint32_t n) {
@@ -35,21 +35,91 @@ void msleep(uint32_t n) {
}
#endif // _WIN32
+#ifdef __MACH__
+
+ #define CLOCK_MONOTONIC (1)
+ #define CLOCK_REALTIME (2)
+
+ #include <sys/time.h>
+ #include <mach/clock.h>
+ #include <mach/mach.h>
+ #include <mach/mach_time.h>
+
+ /* clock_gettime is not implemented on OSX prior to 10.12 */
+ int _civet_clock_gettime(int clk_id, struct timespec *t);
+
+ int _civet_clock_gettime(int clk_id, struct timespec *t)
+ {
+ memset(t, 0, sizeof(*t));
+ if (clk_id == CLOCK_REALTIME) {
+ struct timeval now;
+ int rv = gettimeofday(&now, NULL);
+ if (rv) {
+ return rv;
+ }
+ t->tv_sec = now.tv_sec;
+ t->tv_nsec = now.tv_usec * 1000;
+ return 0;
+
+ } else if (clk_id == CLOCK_MONOTONIC) {
+ static uint64_t clock_start_time = 0;
+ static mach_timebase_info_data_t timebase_ifo = {0, 0};
+
+ uint64_t now = mach_absolute_time();
+
+ if (clock_start_time == 0) {
+ //kern_return_t mach_status = mach_timebase_info(&timebase_ifo);
+ // appease "unused variable" warning for release builds
+ //(void)mach_status;
+ clock_start_time = now;
+ }
+
+ now = (uint64_t)((double)(now - clock_start_time)
+ * (double)timebase_ifo.numer
+ / (double)timebase_ifo.denom);
+
+ t->tv_sec = now / 1000000000;
+ t->tv_nsec = now % 1000000000;
+ return 0;
+ }
+ return -1; // EINVAL - Clock ID is unknown
+ }
+
+ /* if clock_gettime is declared, then __CLOCK_AVAILABILITY will be defined */
+ #ifdef __CLOCK_AVAILABILITY
+ /* If we compiled with Mac OSX 10.12 or later, then clock_gettime will be declared
+ * but it may be NULL at runtime. So we need to check before using it. */
+ int _civet_safe_clock_gettime(int clk_id, struct timespec *t);
+
+ int _civet_safe_clock_gettime(int clk_id, struct timespec *t) {
+ if( clock_gettime ) {
+ return clock_gettime(clk_id, t);
+ }
+ return _civet_clock_gettime(clk_id, t);
+ }
+ #define clock_gettime _civet_safe_clock_gettime
+ #else
+ #define clock_gettime _civet_clock_gettime
+ #endif
+
+#endif
+
+
// a milliseconds timer for performance measurement
uint64_t msclock() {
#if defined(_WIN32)
- #include <sys/types.h>
-
- // WORKAROUND FOR MinGW (some versions - use if normal code does not compile)
- // It has no _ftime_s and needs explicit inclusion of timeb.h
- #include <sys/timeb.h>
- struct _timeb t;
- _ftime(&t);
- return 1000 * t.time + t.millitm;
-
- // NORMAL CODE (use _ftime_s)
+ #include <sys/types.h>
+
+ // WORKAROUND FOR MinGW (some versions - use if normal code does not compile)
+ // It has no _ftime_s and needs explicit inclusion of timeb.h
+ #include <sys/timeb.h>
+ struct _timeb t;
+ _ftime(&t);
+ return 1000 * t.time + t.millitm;
+
+// NORMAL CODE (use _ftime_s)
//struct _timeb t;
- //if (_ftime_s(&t)) {
+ //if (_ftime_s(&t)) {
// return 0;
//} else {
// return 1000 * t.time + t.millitm;
--
2.39.5