Mark crc32_c and crc32_braid functions as internal, and remove prefix.
Reorder contents of generic_functions, and remove Z_INTERNAL hints from declarations.
Add test/benchmark output to indicate whether Chorba is used.
#include "zbuild.h"
#include "crc32.h"
#include "crc32_braid_p.h"
-#include "crc32_c.h"
+#include "generic_functions.h"
-uint32_t PREFIX(crc32_c)(uint32_t crc, const uint8_t *buf, size_t len) {
+Z_INTERNAL uint32_t crc32_c(uint32_t crc, const uint8_t *buf, size_t len) {
uint32_t c;
uint64_t* aligned_buf;
size_t aligned_len;
#include "zbuild.h"
#include "crc32_braid_p.h"
#include "crc32_braid_tbl.h"
-#include "crc32_c.h"
+#include "generic_functions.h"
/* Implement Chorba algorithm from https://arxiv.org/abs/2412.16398 */
#define bitbuffersizebytes (16 * 1024 * sizeof(z_word_t))
#define GENERIC_FUNCTIONS_H_
#include "zendian.h"
-
-Z_INTERNAL uint32_t crc32_fold_reset_c(crc32_fold *crc);
-Z_INTERNAL void crc32_fold_copy_c(crc32_fold *crc, uint8_t *dst, const uint8_t *src, size_t len);
-Z_INTERNAL void crc32_fold_c(crc32_fold *crc, const uint8_t *src, size_t len, uint32_t init_crc);
-Z_INTERNAL uint32_t crc32_fold_final_c(crc32_fold *crc);
-
-Z_INTERNAL uint32_t adler32_fold_copy_c(uint32_t adler, uint8_t *dst, const uint8_t *src, size_t len);
-
+#include "deflate.h"
+#include "crc32_braid_p.h"
typedef uint32_t (*adler32_func)(uint32_t adler, const uint8_t *buf, size_t len);
typedef uint32_t (*compare256_func)(const uint8_t *src0, const uint8_t *src1);
typedef uint32_t (*crc32_func)(uint32_t crc32, const uint8_t *buf, size_t len);
+typedef void (*slide_hash_func)(deflate_state *s);
+
uint32_t adler32_c(uint32_t adler, const uint8_t *buf, size_t len);
+uint32_t adler32_fold_copy_c(uint32_t adler, uint8_t *dst, const uint8_t *src, size_t len);
-uint32_t chunksize_c(void);
uint8_t* chunkmemset_safe_c(uint8_t *out, uint8_t *from, unsigned len, unsigned left);
-void inflate_fast_c(PREFIX3(stream) *strm, uint32_t start);
-
-uint32_t PREFIX(crc32_c)(uint32_t crc, const uint8_t *buf, size_t len);
-uint32_t PREFIX(crc32_braid)(uint32_t c, const uint8_t *buf, size_t len);
-Z_INTERNAL uint32_t crc32_braid_internal(uint32_t c, const uint8_t *buf, size_t len);
+uint32_t chunksize_c(void);
uint32_t compare256_c(const uint8_t *src0, const uint8_t *src1);
-typedef void (*slide_hash_func)(deflate_state *s);
+uint32_t crc32_c(uint32_t crc, const uint8_t *buf, size_t len);
+uint32_t crc32_braid(uint32_t c, const uint8_t *buf, size_t len);
+uint32_t crc32_braid_internal(uint32_t c, const uint8_t *buf, size_t len);
-void slide_hash_c(deflate_state *s);
+#ifndef WITHOUT_CHORBA
+ uint32_t crc32_chorba_118960_nondestructive (uint32_t crc, const z_word_t* input, size_t len);
+ uint32_t crc32_chorba_32768_nondestructive (uint32_t crc, const uint64_t* buf, size_t len);
+ uint32_t crc32_chorba_small_nondestructive (uint32_t crc, const uint64_t* buf, size_t len);
+ uint32_t crc32_chorba_small_nondestructive_32bit (uint32_t crc, const uint32_t* buf, size_t len);
+#endif
+
+uint32_t crc32_fold_reset_c(crc32_fold *crc);
+void crc32_fold_copy_c(crc32_fold *crc, uint8_t *dst, const uint8_t *src, size_t len);
+void crc32_fold_c(crc32_fold *crc, const uint8_t *src, size_t len, uint32_t init_crc);
+uint32_t crc32_fold_final_c(crc32_fold *crc);
+
+void inflate_fast_c(PREFIX3(stream) *strm, uint32_t start);
uint32_t longest_match_c(deflate_state *const s, Pos cur_match);
uint32_t longest_match_slow_c(deflate_state *const s, Pos cur_match);
+void slide_hash_c(deflate_state *s);
+
#ifdef DISABLE_RUNTIME_CPU_DETECTION
// Generic code
# define native_adler32 adler32_c
size_t prealign, aligned, remaining;
if (len < VX_MIN_LEN + VX_ALIGN_MASK)
- return PREFIX(crc32_c)(crc, buf, len);
+ return crc32_c(crc, buf, len);
if ((uintptr_t)buf & VX_ALIGN_MASK) {
prealign = VX_ALIGNMENT - ((uintptr_t)buf & VX_ALIGN_MASK);
len -= prealign;
- crc = PREFIX(crc32_c)(crc, buf, prealign);
+ crc = crc32_c(crc, buf, prealign);
buf += prealign;
}
aligned = len & ~VX_ALIGN_MASK;
crc = crc32_le_vgfm_16(crc ^ 0xffffffff, buf, aligned) ^ 0xffffffff;
if (remaining)
- crc = PREFIX(crc32_c)(crc, buf + aligned, remaining);
+ crc = crc32_c(crc, buf + aligned, remaining);
return crc;
}
+++ /dev/null
-Z_INTERNAL uint32_t crc32_chorba_118960_nondestructive (uint32_t crc, const z_word_t* input, size_t len);
-Z_INTERNAL uint32_t crc32_chorba_32768_nondestructive (uint32_t crc, const uint64_t* buf, size_t len);
-Z_INTERNAL uint32_t crc32_chorba_small_nondestructive (uint32_t crc, const uint64_t* buf, size_t len);
-Z_INTERNAL uint32_t crc32_chorba_small_nondestructive_32bit (uint32_t crc, const uint32_t* buf, size_t len);
ft.adler32_fold_copy = &adler32_fold_copy_c;
ft.chunkmemset_safe = &chunkmemset_safe_c;
ft.chunksize = &chunksize_c;
- ft.crc32 = &PREFIX(crc32_c);
+ ft.crc32 = &crc32_c;
ft.crc32_fold = &crc32_fold_c;
ft.crc32_fold_copy = &crc32_fold_copy_c;
ft.crc32_fold_final = &crc32_fold_final_c;
} \
BENCHMARK_REGISTER_F(crc32, name)->Arg(1)->Arg(8)->Arg(12)->Arg(16)->Arg(32)->Arg(64)->Arg(512)->Arg(4<<10)->Arg(32<<10)->Arg(256<<10)->Arg(4096<<10);
-BENCHMARK_CRC32(braid, PREFIX(crc32_braid), 1);
-BENCHMARK_CRC32(generic, PREFIX(crc32_c), 1);
+#ifndef WITHOUT_CHORBA
+BENCHMARK_CRC32(generic_chorba, crc32_c, 1);
+#else
+BENCHMARK_CRC32(generic, crc32_c, 1);
+#endif
+
+BENCHMARK_CRC32(braid, crc32_braid, 1);
#ifdef DISABLE_RUNTIME_CPU_DETECTION
BENCHMARK_CRC32(native, native_crc32, 1);
hash(GetParam(), func); \
}
-TEST_CRC32(generic, PREFIX(crc32_c), 1)
-TEST_CRC32(braid, PREFIX(crc32_braid), 1)
+#ifndef WITHOUT_CHORBA
+TEST_CRC32(generic_chorba, crc32_c, 1)
+#else
+TEST_CRC32(generic, crc32_c, 1)
+#endif
+
+TEST_CRC32(braid, crc32_braid, 1)
#ifdef DISABLE_RUNTIME_CPU_DETECTION
TEST_CRC32(native, native_crc32, 1)