generic-library/isa-l
1
0
Fork 0
mirror of https://github.com/intel/isa-l.git synced 2024-12-12 09:23:50 +01:00
Code Issues Projects Releases Wiki Activity

include: reformat using new code style

Browse Source 
Signed-off-by: Marcel Cornu <marcel.d.cornu@intel.com>
This commit is contained in:
Marcel Cornu 2024-04-19 17:09:27 +01:00 committed by Pablo de Lara
parent 55fbfabfc6
commit fa5b8baf84
10 changed files with 1013 additions and 890 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
include/aarch64_label.h
View File

@ -3,16 +3,16 @@
#ifdef __USER_LABEL_PREFIX__
#define CONCAT1(a, b) CONCAT2(a, b)
#define CONCAT2(a, b) a ## b
#define cdecl(x) CONCAT1 (__USER_LABEL_PREFIX__, x)
#define CONCAT2(a, b) a##b
#define cdecl(x) CONCAT1(__USER_LABEL_PREFIX__, x)
#else
#define cdecl(x) x
#endif
#ifdef __APPLE__
#define ASM_DEF_RODATA .section __TEXT,__const
#define ASM_DEF_RODATA .section __TEXT, __const
#else
#define ASM_DEF_RODATA .section .rodata
#define ASM_DEF_RODATA .section.rodata
#endif
#endif

71
include/crc.h
View File

@ -27,13 +27,11 @@
OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
**********************************************************************/
/**
* @file crc.h
* @brief CRC functions.
*/
#ifndef _CRC_H_
#define _CRC_H_
@ -43,7 +41,6 @@
extern "C" {
#endif
/* Multi-binary functions */
/**
@ -54,12 +51,11 @@ extern "C" {
*
* @returns 16 bit CRC
*/
uint16_t crc16_t10dif(
uint16_t init_crc, //!< initial CRC value, 16 bits
uint16_t
crc16_t10dif(uint16_t init_crc, //!< initial CRC value, 16 bits
const unsigned char *buf, //!< buffer to calculate CRC on
uint64_t len //!< buffer length in bytes (64-bit data)
);
);
/**
* @brief Generate CRC and copy T10 standard, runs appropriate version.
@ -68,13 +64,12 @@ uint16_t crc16_t10dif(
*
* @returns 16 bit CRC
*/
uint16_t crc16_t10dif_copy(
uint16_t init_crc, //!< initial CRC value, 16 bits
uint16_t
crc16_t10dif_copy(uint16_t init_crc, //!< initial CRC value, 16 bits
uint8_t *dst, //!< buffer destination for copy
uint8_t *src, //!< buffer source to crc + copy
uint64_t len //!< buffer length in bytes (64-bit data)
);
);
/**
* @brief Generate CRC from the IEEE standard, runs appropriate version.
@ -96,11 +91,11 @@ uint16_t crc16_t10dif_copy(
* @returns 32 bit CRC
*/
uint32_t crc32_ieee(
uint32_t init_crc, //!< initial CRC value, 32 bits
uint32_t
crc32_ieee(uint32_t init_crc, //!< initial CRC value, 32 bits
const unsigned char *buf, //!< buffer to calculate CRC on
uint64_t len //!< buffer length in bytes (64-bit data)
);
);
/**
* @brief Generate the customized CRC
@ -124,12 +119,11 @@ uint32_t crc32_ieee(
*
* @returns 32 bit CRC
*/
uint32_t crc32_gzip_refl(
uint32_t init_crc, //!< initial CRC value, 32 bits
uint32_t
crc32_gzip_refl(uint32_t init_crc, //!< initial CRC value, 32 bits
const unsigned char *buf, //!< buffer to calculate CRC on
uint64_t len //!< buffer length in bytes (64-bit data)
);
);
/**
* @brief ISCSI CRC function, runs appropriate version.
@ -139,12 +133,11 @@ uint32_t crc32_gzip_refl(
*
* @returns 32 bit CRC
*/
unsigned int crc32_iscsi(
unsigned char *buffer, //!< buffer to calculate CRC on
unsigned int
crc32_iscsi(unsigned char *buffer, //!< buffer to calculate CRC on
int len, //!< buffer length in bytes
unsigned int init_crc //!< initial CRC value
);
);
/* Base functions */
@ -152,45 +145,42 @@ unsigned int crc32_iscsi(
* @brief ISCSI CRC function, baseline version
* @returns 32 bit CRC
*/
unsigned int crc32_iscsi_base(
unsigned char *buffer, //!< buffer to calculate CRC on
unsigned int
crc32_iscsi_base(unsigned char *buffer, //!< buffer to calculate CRC on
int len, //!< buffer length in bytes
unsigned int crc_init //!< initial CRC value
);
);
/**
* @brief Generate CRC from the T10 standard, runs baseline version
* @returns 16 bit CRC
*/
uint16_t crc16_t10dif_base(
uint16_t seed, //!< initial CRC value, 16 bits
uint16_t
crc16_t10dif_base(uint16_t seed, //!< initial CRC value, 16 bits
uint8_t *buf, //!< buffer to calculate CRC on
uint64_t len //!< buffer length in bytes (64-bit data)
);
);
/**
* @brief Generate CRC and copy T10 standard, runs baseline version.
* @returns 16 bit CRC
*/
uint16_t crc16_t10dif_copy_base(
uint16_t init_crc, //!< initial CRC value, 16 bits
uint16_t
crc16_t10dif_copy_base(uint16_t init_crc, //!< initial CRC value, 16 bits
uint8_t *dst, //!< buffer destination for copy
uint8_t *src, //!< buffer source to crc + copy
uint64_t len //!< buffer length in bytes (64-bit data)
);
);
/**
* @brief Generate CRC from the IEEE standard, runs baseline version
* @returns 32 bit CRC
*/
uint32_t crc32_ieee_base(
uint32_t seed, //!< initial CRC value, 32 bits
uint32_t
crc32_ieee_base(uint32_t seed, //!< initial CRC value, 32 bits
uint8_t *buf, //!< buffer to calculate CRC on
uint64_t len //!< buffer length in bytes (64-bit data)
);
);
/**
* @brief Generate the customized CRC
@ -198,12 +188,11 @@ uint32_t crc32_ieee_base(
* runs baseline version
* @returns 32 bit CRC
*/
uint32_t crc32_gzip_refl_base(
uint32_t seed, //!< initial CRC value, 32 bits
uint32_t
crc32_gzip_refl_base(uint32_t seed, //!< initial CRC value, 32 bits
uint8_t *buf, //!< buffer to calculate CRC on
uint64_t len //!< buffer length in bytes (64-bit data)
);
);
#ifdef __cplusplus
}

147
include/crc64.h
View File

@ -27,13 +27,11 @@
OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
**********************************************************************/
/**
* @file crc64.h
* @brief CRC64 functions.
*/
#ifndef _CRC64_H_
#define _CRC64_H_
@ -43,7 +41,6 @@
extern "C" {
#endif
/* Multi-binary functions */
/**
@ -54,11 +51,11 @@ extern "C" {
* selects the appropriate version at runtime.
* @returns 64 bit CRC
*/
uint64_t crc64_ecma_refl(
uint64_t init_crc, //!< initial CRC value, 64 bits
uint64_t
crc64_ecma_refl(uint64_t init_crc, //!< initial CRC value, 64 bits
const unsigned char *buf, //!< buffer to calculate CRC on
uint64_t len //!< buffer length in bytes (64-bit data)
);
);
/**
* @brief Generate CRC from ECMA-182 standard in normal format, runs
@ -68,11 +65,11 @@ uint64_t crc64_ecma_refl(
* selects the appropriate version at runtime.
* @returns 64 bit CRC
*/
uint64_t crc64_ecma_norm(
uint64_t init_crc, //!< initial CRC value, 64 bits
uint64_t
crc64_ecma_norm(uint64_t init_crc, //!< initial CRC value, 64 bits
const unsigned char *buf, //!< buffer to calculate CRC on
uint64_t len //!< buffer length in bytes (64-bit data)
);
);
/**
* @brief Generate CRC from ISO standard in reflected format, runs
@ -82,11 +79,11 @@ uint64_t crc64_ecma_norm(
* selects the appropriate version at runtime.
* @returns 64 bit CRC
*/
uint64_t crc64_iso_refl(
uint64_t init_crc, //!< initial CRC value, 64 bits
uint64_t
crc64_iso_refl(uint64_t init_crc, //!< initial CRC value, 64 bits
const unsigned char *buf, //!< buffer to calculate CRC on
uint64_t len //!< buffer length in bytes (64-bit data)
);
);
/**
* @brief Generate CRC from ISO standard in normal format, runs
@ -96,11 +93,11 @@ uint64_t crc64_iso_refl(
* selects the appropriate version at runtime.
* @returns 64 bit CRC
*/
uint64_t crc64_iso_norm(
uint64_t init_crc, //!< initial CRC value, 64 bits
uint64_t
crc64_iso_norm(uint64_t init_crc, //!< initial CRC value, 64 bits
const unsigned char *buf, //!< buffer to calculate CRC on
uint64_t len //!< buffer length in bytes (64-bit data)
);
);
/**
* @brief Generate CRC from "Jones" coefficients in reflected format, runs
@ -110,11 +107,11 @@ uint64_t crc64_iso_norm(
* selects the appropriate version at runtime.
* @returns 64 bit CRC
*/
uint64_t crc64_jones_refl(
uint64_t init_crc, //!< initial CRC value, 64 bits
uint64_t
crc64_jones_refl(uint64_t init_crc, //!< initial CRC value, 64 bits
const unsigned char *buf, //!< buffer to calculate CRC on
uint64_t len //!< buffer length in bytes (64-bit data)
);
);
/**
* @brief Generate CRC from "Jones" coefficients in normal format, runs
@ -124,11 +121,11 @@ uint64_t crc64_jones_refl(
* selects the appropriate version at runtime.
* @returns 64 bit CRC
*/
uint64_t crc64_jones_norm(
uint64_t init_crc, //!< initial CRC value, 64 bits
uint64_t
crc64_jones_norm(uint64_t init_crc, //!< initial CRC value, 64 bits
const unsigned char *buf, //!< buffer to calculate CRC on
uint64_t len //!< buffer length in bytes (64-bit data)
);
);
/**
* @brief Generate CRC from "Rocksoft" coefficients in reflected format, runs
@ -138,11 +135,11 @@ uint64_t crc64_jones_norm(
* selects the appropriate version at runtime.
* @returns 64 bit CRC
*/
uint64_t crc64_rocksoft_refl(
uint64_t init_crc, //!< initial CRC value, 64 bits
uint64_t
crc64_rocksoft_refl(uint64_t init_crc, //!< initial CRC value, 64 bits
const unsigned char *buf, //!< buffer to calculate CRC on
uint64_t len //!< buffer length in bytes (64-bit data)
);
);
/**
* @brief Generate CRC from "Rocksoft" coefficients in normal format, runs
@ -152,11 +149,11 @@ uint64_t crc64_rocksoft_refl(
* selects the appropriate version at runtime.
* @returns 64 bit CRC
*/
uint64_t crc64_rocksoft_norm(
uint64_t init_crc, //!< initial CRC value, 64 bits
uint64_t
crc64_rocksoft_norm(uint64_t init_crc, //!< initial CRC value, 64 bits
const unsigned char *buf, //!< buffer to calculate CRC on
uint64_t len //!< buffer length in bytes (64-bit data)
);
);
/* Arch specific versions */
@ -167,11 +164,11 @@ uint64_t crc64_rocksoft_norm(
* @returns 64 bit CRC
*/
uint64_t crc64_ecma_refl_by8(
uint64_t init_crc, //!< initial CRC value, 64 bits
uint64_t
crc64_ecma_refl_by8(uint64_t init_crc, //!< initial CRC value, 64 bits
const unsigned char *buf, //!< buffer to calculate CRC on
uint64_t len //!< buffer length in bytes (64-bit data)
);
);
/**
* @brief Generate CRC from ECMA-182 standard in normal format.
@ -180,31 +177,31 @@ uint64_t crc64_ecma_refl_by8(
* @returns 64 bit CRC
*/
uint64_t crc64_ecma_norm_by8(
uint64_t init_crc, //!< initial CRC value, 64 bits
uint64_t
crc64_ecma_norm_by8(uint64_t init_crc, //!< initial CRC value, 64 bits
const unsigned char *buf, //!< buffer to calculate CRC on
uint64_t len //!< buffer length in bytes (64-bit data)
);
);
/**
* @brief Generate CRC from ECMA-182 standard in reflected format, runs baseline version
* @returns 64 bit CRC
*/
uint64_t crc64_ecma_refl_base(
uint64_t init_crc, //!< initial CRC value, 64 bits
uint64_t
crc64_ecma_refl_base(uint64_t init_crc, //!< initial CRC value, 64 bits
const unsigned char *buf, //!< buffer to calculate CRC on
uint64_t len //!< buffer length in bytes (64-bit data)
);
);
/**
* @brief Generate CRC from ECMA-182 standard in normal format, runs baseline version
* @returns 64 bit CRC
*/
uint64_t crc64_ecma_norm_base(
uint64_t init_crc, //!< initial CRC value, 64 bits
uint64_t
crc64_ecma_norm_base(uint64_t init_crc, //!< initial CRC value, 64 bits
const unsigned char *buf, //!< buffer to calculate CRC on
uint64_t len //!< buffer length in bytes (64-bit data)
);
);
/**
* @brief Generate CRC from ISO standard in reflected format.
@ -213,11 +210,11 @@ uint64_t crc64_ecma_norm_base(
* @returns 64 bit CRC
*/
uint64_t crc64_iso_refl_by8(
uint64_t init_crc, //!< initial CRC value, 64 bits
uint64_t
crc64_iso_refl_by8(uint64_t init_crc, //!< initial CRC value, 64 bits
const unsigned char *buf, //!< buffer to calculate CRC on
uint64_t len //!< buffer length in bytes (64-bit data)
);
);
/**
* @brief Generate CRC from ISO standard in normal format.
@ -226,31 +223,31 @@ uint64_t crc64_iso_refl_by8(
* @returns 64 bit CRC
*/
uint64_t crc64_iso_norm_by8(
uint64_t init_crc, //!< initial CRC value, 64 bits
uint64_t
crc64_iso_norm_by8(uint64_t init_crc, //!< initial CRC value, 64 bits
const unsigned char *buf, //!< buffer to calculate CRC on
uint64_t len //!< buffer length in bytes (64-bit data)
);
);
/**
* @brief Generate CRC from ISO standard in reflected format, runs baseline version
* @returns 64 bit CRC
*/
uint64_t crc64_iso_refl_base(
uint64_t init_crc, //!< initial CRC value, 64 bits
uint64_t
crc64_iso_refl_base(uint64_t init_crc, //!< initial CRC value, 64 bits
const unsigned char *buf, //!< buffer to calculate CRC on
uint64_t len //!< buffer length in bytes (64-bit data)
);
);
/**
* @brief Generate CRC from ISO standard in normal format, runs baseline version
* @returns 64 bit CRC
*/
uint64_t crc64_iso_norm_base(
uint64_t init_crc, //!< initial CRC value, 64 bits
uint64_t
crc64_iso_norm_base(uint64_t init_crc, //!< initial CRC value, 64 bits
const unsigned char *buf, //!< buffer to calculate CRC on
uint64_t len //!< buffer length in bytes (64-bit data)
);
);
/**
* @brief Generate CRC from "Jones" coefficients in reflected format.
@ -259,11 +256,11 @@ uint64_t crc64_iso_norm_base(
* @returns 64 bit CRC
*/
uint64_t crc64_jones_refl_by8(
uint64_t init_crc, //!< initial CRC value, 64 bits
uint64_t
crc64_jones_refl_by8(uint64_t init_crc, //!< initial CRC value, 64 bits
const unsigned char *buf, //!< buffer to calculate CRC on
uint64_t len //!< buffer length in bytes (64-bit data)
);
);
/**
* @brief Generate CRC from "Jones" coefficients in normal format.
@ -272,31 +269,31 @@ uint64_t crc64_jones_refl_by8(
* @returns 64 bit CRC
*/
uint64_t crc64_jones_norm_by8(
uint64_t init_crc, //!< initial CRC value, 64 bits
uint64_t
crc64_jones_norm_by8(uint64_t init_crc, //!< initial CRC value, 64 bits
const unsigned char *buf, //!< buffer to calculate CRC on
uint64_t len //!< buffer length in bytes (64-bit data)
);
);
/**
* @brief Generate CRC from "Jones" coefficients in reflected format, runs baseline version
* @returns 64 bit CRC
*/
uint64_t crc64_jones_refl_base(
uint64_t init_crc, //!< initial CRC value, 64 bits
uint64_t
crc64_jones_refl_base(uint64_t init_crc, //!< initial CRC value, 64 bits
const unsigned char *buf, //!< buffer to calculate CRC on
uint64_t len //!< buffer length in bytes (64-bit data)
);
);
/**
* @brief Generate CRC from "Jones" coefficients in normal format, runs baseline version
* @returns 64 bit CRC
*/
uint64_t crc64_jones_norm_base(
uint64_t init_crc, //!< initial CRC value, 64 bits
uint64_t
crc64_jones_norm_base(uint64_t init_crc, //!< initial CRC value, 64 bits
const unsigned char *buf, //!< buffer to calculate CRC on
uint64_t len //!< buffer length in bytes (64-bit data)
);
);
/**
* @brief Generate CRC from "Rocksoft" coefficients in reflected format.
@ -305,21 +302,21 @@ uint64_t crc64_jones_norm_base(
* @returns 64 bit CRC
*/
uint64_t crc64_rocksoft_refl_by8(
uint64_t init_crc, //!< initial CRC value, 64 bits
uint64_t
crc64_rocksoft_refl_by8(uint64_t init_crc, //!< initial CRC value, 64 bits
const unsigned char *buf, //!< buffer to calculate CRC on
uint64_t len //!< buffer length in bytes (64-bit data)
);
);
/**
* @brief Generate CRC from "Rocksoft" coefficients in reflected format, runs baseline version
* @returns 64 bit CRC
*/
uint64_t crc64_rocksoft_refl_base(
uint64_t init_crc, //!< initial CRC value, 64 bits
uint64_t
crc64_rocksoft_refl_base(uint64_t init_crc, //!< initial CRC value, 64 bits
const unsigned char *buf, //!< buffer to calculate CRC on
uint64_t len //!< buffer length in bytes (64-bit data)
);
);
/**
* @brief Generate CRC from "Rocksoft" coefficients in normal format.
@ -328,21 +325,21 @@ uint64_t crc64_rocksoft_refl_base(
* @returns 64 bit CRC
*/
uint64_t crc64_rocksoft_norm_by8(
uint64_t init_crc, //!< initial CRC value, 64 bits
uint64_t
crc64_rocksoft_norm_by8(uint64_t init_crc, //!< initial CRC value, 64 bits
const unsigned char *buf, //!< buffer to calculate CRC on
uint64_t len //!< buffer length in bytes (64-bit data)
);
);
/**
* @brief Generate CRC from "Rocksoft" coefficients in normal format, runs baseline version
* @returns 64 bit CRC
*/
uint64_t crc64_rocksoft_norm_base(
uint64_t init_crc, //!< initial CRC value, 64 bits
uint64_t
crc64_rocksoft_norm_base(uint64_t init_crc, //!< initial CRC value, 64 bits
const unsigned char *buf, //!< buffer to calculate CRC on
uint64_t len //!< buffer length in bytes (64-bit data)
);
);
#ifdef __cplusplus
}

218
include/erasure_code.h
View File

@ -27,7 +27,6 @@
OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
**********************************************************************/
#ifndef _ERASURE_CODE_H_
#define _ERASURE_CODE_H_
@ -71,7 +70,8 @@ extern "C" {
* @returns none
*/
void ec_init_tables(int k, int rows, unsigned char* a, unsigned char* gftbls);
void
ec_init_tables(int k, int rows, unsigned char *a, unsigned char *gftbls);
/**
* @brief Initialize tables for fast Erasure Code encode and decode, runs baseline version.
@ -79,7 +79,8 @@ void ec_init_tables(int k, int rows, unsigned char* a, unsigned char* gftbls);
* Baseline version of ec_encode_data() with same parameters.
*/
void ec_init_tables_base(int k, int rows, unsigned char* a, unsigned char* gftbls);
void
ec_init_tables_base(int k, int rows, unsigned char *a, unsigned char *gftbls);
/**
* @brief Generate or decode erasure codes on blocks of data, runs appropriate version.
@ -103,7 +104,8 @@ void ec_init_tables_base(int k, int rows, unsigned char* a, unsigned char* gftbl
* @returns none
*/
void ec_encode_data(int len, int k, int rows, unsigned char *gftbls, unsigned char **data,
void
ec_encode_data(int len, int k, int rows, unsigned char *gftbls, unsigned char **data,
unsigned char **coding);
/**
@ -111,11 +113,13 @@ void ec_encode_data(int len, int k, int rows, unsigned char *gftbls, unsigned ch
*
* Baseline version of ec_encode_data() with same parameters.
*/
void ec_encode_data_base(int len, int srcs, int dests, unsigned char *v, unsigned char **src,
void
ec_encode_data_base(int len, int srcs, int dests, unsigned char *v, unsigned char **src,
unsigned char **dest);
/**
* @brief Generate update for encode or decode of erasure codes from single source, runs appropriate version.
* @brief Generate update for encode or decode of erasure codes from single source, runs appropriate
* version.
*
* Given one source data block, update one or multiple blocks of encoded data as
* specified by a matrix of GF(2^8) coefficients. When given a suitable set of
@ -136,7 +140,8 @@ void ec_encode_data_base(int len, int srcs, int dests, unsigned char *v, unsigne
* @param coding Array of pointers to coded output buffers.
* @returns none
*/
void ec_encode_data_update(int len, int k, int rows, int vec_i, unsigned char *g_tbls,
void
ec_encode_data_update(int len, int k, int rows, int vec_i, unsigned char *g_tbls,
unsigned char *data, unsigned char **coding);
/**
@ -145,7 +150,8 @@ void ec_encode_data_update(int len, int k, int rows, int vec_i, unsigned char *g
* Baseline version of ec_encode_data_update().
*/
void ec_encode_data_update_base(int len, int k, int rows, int vec_i, unsigned char *v,
void
ec_encode_data_update_base(int len, int k, int rows, int vec_i, unsigned char *v,
unsigned char *data, unsigned char **dest);
/**
@ -168,9 +174,9 @@ void ec_encode_data_update_base(int len, int k, int rows, int vec_i, unsigned ch
* @returns none
*/
void gf_vect_dot_prod_base(int len, int vlen, unsigned char *gftbls,
unsigned char **src, unsigned char *dest);
void
gf_vect_dot_prod_base(int len, int vlen, unsigned char *gftbls, unsigned char **src,
unsigned char *dest);
/**
* @brief GF(2^8) vector dot product, runs appropriate version.
@ -192,8 +198,9 @@ void gf_vect_dot_prod_base(int len, int vlen, unsigned char *gftbls,
* @returns none
*/
void gf_vect_dot_prod(int len, int vlen, unsigned char *gftbls,
unsigned char **src, unsigned char *dest);
void
gf_vect_dot_prod(int len, int vlen, unsigned char *gftbls, unsigned char **src,
unsigned char *dest);
/**
* @brief GF(2^8) vector multiply accumulate, runs appropriate version.
@ -218,7 +225,8 @@ void gf_vect_dot_prod(int len, int vlen, unsigned char *gftbls,
* @returns none
*/
void gf_vect_mad(int len, int vec, int vec_i, unsigned char *gftbls, unsigned char *src,
void
gf_vect_mad(int len, int vec, int vec_i, unsigned char *gftbls, unsigned char *src,
unsigned char *dest);
/**
@ -227,7 +235,8 @@ void gf_vect_mad(int len, int vec, int vec_i, unsigned char *gftbls, unsigned ch
* Baseline version of gf_vect_mad() with same parameters.
*/
void gf_vect_mad_base(int len, int vec, int vec_i, unsigned char *v, unsigned char *src,
void
gf_vect_mad_base(int len, int vec, int vec_i, unsigned char *v, unsigned char *src,
unsigned char *dest);
// x86 only
@ -239,7 +248,8 @@ void gf_vect_mad_base(int len, int vec, int vec_i, unsigned char *v, unsigned ch
* Arch specific version of ec_encode_data() with same parameters.
* @requires SSE4.1
*/
void ec_encode_data_sse(int len, int k, int rows, unsigned char *gftbls, unsigned char **data,
void
ec_encode_data_sse(int len, int k, int rows, unsigned char *gftbls, unsigned char **data,
unsigned char **coding);
/**
@ -248,7 +258,8 @@ void ec_encode_data_sse(int len, int k, int rows, unsigned char *gftbls, unsigne
* Arch specific version of ec_encode_data() with same parameters.
* @requires AVX
*/
void ec_encode_data_avx(int len, int k, int rows, unsigned char *gftbls, unsigned char **data,
void
ec_encode_data_avx(int len, int k, int rows, unsigned char *gftbls, unsigned char **data,
unsigned char **coding);
/**
@ -257,7 +268,8 @@ void ec_encode_data_avx(int len, int k, int rows, unsigned char *gftbls, unsigne
* Arch specific version of ec_encode_data() with same parameters.
* @requires AVX2
*/
void ec_encode_data_avx2(int len, int k, int rows, unsigned char *gftbls, unsigned char **data,
void
ec_encode_data_avx2(int len, int k, int rows, unsigned char *gftbls, unsigned char **data,
unsigned char **coding);
/**
@ -267,7 +279,8 @@ void ec_encode_data_avx2(int len, int k, int rows, unsigned char *gftbls, unsign
* @requires SSE4.1
*/
void ec_encode_data_update_sse(int len, int k, int rows, int vec_i, unsigned char *g_tbls,
void
ec_encode_data_update_sse(int len, int k, int rows, int vec_i, unsigned char *g_tbls,
unsigned char *data, unsigned char **coding);
/**
@ -277,7 +290,8 @@ void ec_encode_data_update_sse(int len, int k, int rows, int vec_i, unsigned cha
* @requires AVX
*/
void ec_encode_data_update_avx(int len, int k, int rows, int vec_i, unsigned char *g_tbls,
void
ec_encode_data_update_avx(int len, int k, int rows, int vec_i, unsigned char *g_tbls,
unsigned char *data, unsigned char **coding);
/**
@ -287,7 +301,8 @@ void ec_encode_data_update_avx(int len, int k, int rows, int vec_i, unsigned cha
* @requires AVX2
*/
void ec_encode_data_update_avx2(int len, int k, int rows, int vec_i, unsigned char *g_tbls,
void
ec_encode_data_update_avx2(int len, int k, int rows, int vec_i, unsigned char *g_tbls,
unsigned char *data, unsigned char **coding);
/**
@ -308,8 +323,9 @@ void ec_encode_data_update_avx2(int len, int k, int rows, int vec_i, unsigned ch
* @returns none
*/
void gf_vect_dot_prod_sse(int len, int vlen, unsigned char *gftbls,
unsigned char **src, unsigned char *dest);
void
gf_vect_dot_prod_sse(int len, int vlen, unsigned char *gftbls, unsigned char **src,
unsigned char *dest);
/**
* @brief GF(2^8) vector dot product.
@ -329,8 +345,9 @@ void gf_vect_dot_prod_sse(int len, int vlen, unsigned char *gftbls,
* @returns none
*/
void gf_vect_dot_prod_avx(int len, int vlen, unsigned char *gftbls,
unsigned char **src, unsigned char *dest);
void
gf_vect_dot_prod_avx(int len, int vlen, unsigned char *gftbls, unsigned char **src,
unsigned char *dest);
/**
* @brief GF(2^8) vector dot product.
@ -350,8 +367,9 @@ void gf_vect_dot_prod_avx(int len, int vlen, unsigned char *gftbls,
* @returns none
*/
void gf_vect_dot_prod_avx2(int len, int vlen, unsigned char *gftbls,
unsigned char **src, unsigned char *dest);
void
gf_vect_dot_prod_avx2(int len, int vlen, unsigned char *gftbls, unsigned char **src,
unsigned char *dest);
/**
* @brief GF(2^8) vector dot product with two outputs.
@ -372,8 +390,9 @@ void gf_vect_dot_prod_avx2(int len, int vlen, unsigned char *gftbls,
* @returns none
*/
void gf_2vect_dot_prod_sse(int len, int vlen, unsigned char *gftbls,
unsigned char **src, unsigned char **dest);
void
gf_2vect_dot_prod_sse(int len, int vlen, unsigned char *gftbls, unsigned char **src,
unsigned char **dest);
/**
* @brief GF(2^8) vector dot product with two outputs.
@ -394,8 +413,9 @@ void gf_2vect_dot_prod_sse(int len, int vlen, unsigned char *gftbls,
* @returns none
*/
void gf_2vect_dot_prod_avx(int len, int vlen, unsigned char *gftbls,
unsigned char **src, unsigned char **dest);
void
gf_2vect_dot_prod_avx(int len, int vlen, unsigned char *gftbls, unsigned char **src,
unsigned char **dest);
/**
* @brief GF(2^8) vector dot product with two outputs.
@ -416,8 +436,9 @@ void gf_2vect_dot_prod_avx(int len, int vlen, unsigned char *gftbls,
* @returns none
*/
void gf_2vect_dot_prod_avx2(int len, int vlen, unsigned char *gftbls,
unsigned char **src, unsigned char **dest);
void
gf_2vect_dot_prod_avx2(int len, int vlen, unsigned char *gftbls, unsigned char **src,
unsigned char **dest);
/**
* @brief GF(2^8) vector dot product with three outputs.
@ -438,8 +459,9 @@ void gf_2vect_dot_prod_avx2(int len, int vlen, unsigned char *gftbls,
* @returns none
*/
void gf_3vect_dot_prod_sse(int len, int vlen, unsigned char *gftbls,
unsigned char **src, unsigned char **dest);
void
gf_3vect_dot_prod_sse(int len, int vlen, unsigned char *gftbls, unsigned char **src,
unsigned char **dest);
/**
* @brief GF(2^8) vector dot product with three outputs.
@ -460,8 +482,9 @@ void gf_3vect_dot_prod_sse(int len, int vlen, unsigned char *gftbls,
* @returns none
*/
void gf_3vect_dot_prod_avx(int len, int vlen, unsigned char *gftbls,
unsigned char **src, unsigned char **dest);
void
gf_3vect_dot_prod_avx(int len, int vlen, unsigned char *gftbls, unsigned char **src,
unsigned char **dest);
/**
* @brief GF(2^8) vector dot product with three outputs.
@ -482,8 +505,9 @@ void gf_3vect_dot_prod_avx(int len, int vlen, unsigned char *gftbls,
* @returns none
*/
void gf_3vect_dot_prod_avx2(int len, int vlen, unsigned char *gftbls,
unsigned char **src, unsigned char **dest);
void
gf_3vect_dot_prod_avx2(int len, int vlen, unsigned char *gftbls, unsigned char **src,
unsigned char **dest);
/**
* @brief GF(2^8) vector dot product with four outputs.
@ -504,8 +528,9 @@ void gf_3vect_dot_prod_avx2(int len, int vlen, unsigned char *gftbls,
* @returns none
*/
void gf_4vect_dot_prod_sse(int len, int vlen, unsigned char *gftbls,
unsigned char **src, unsigned char **dest);
void
gf_4vect_dot_prod_sse(int len, int vlen, unsigned char *gftbls, unsigned char **src,
unsigned char **dest);
/**
* @brief GF(2^8) vector dot product with four outputs.
@ -526,8 +551,9 @@ void gf_4vect_dot_prod_sse(int len, int vlen, unsigned char *gftbls,
* @returns none
*/
void gf_4vect_dot_prod_avx(int len, int vlen, unsigned char *gftbls,
unsigned char **src, unsigned char **dest);
void
gf_4vect_dot_prod_avx(int len, int vlen, unsigned char *gftbls, unsigned char **src,
unsigned char **dest);
/**
* @brief GF(2^8) vector dot product with four outputs.
@ -548,8 +574,9 @@ void gf_4vect_dot_prod_avx(int len, int vlen, unsigned char *gftbls,
* @returns none
*/
void gf_4vect_dot_prod_avx2(int len, int vlen, unsigned char *gftbls,
unsigned char **src, unsigned char **dest);
void
gf_4vect_dot_prod_avx2(int len, int vlen, unsigned char *gftbls, unsigned char **src,
unsigned char **dest);
/**
* @brief GF(2^8) vector dot product with five outputs.
@ -570,8 +597,9 @@ void gf_4vect_dot_prod_avx2(int len, int vlen, unsigned char *gftbls,
* @returns none
*/
void gf_5vect_dot_prod_sse(int len, int vlen, unsigned char *gftbls,
unsigned char **src, unsigned char **dest);
void
gf_5vect_dot_prod_sse(int len, int vlen, unsigned char *gftbls, unsigned char **src,
unsigned char **dest);
/**
* @brief GF(2^8) vector dot product with five outputs.
@ -592,8 +620,9 @@ void gf_5vect_dot_prod_sse(int len, int vlen, unsigned char *gftbls,
* @returns none
*/
void gf_5vect_dot_prod_avx(int len, int vlen, unsigned char *gftbls,
unsigned char **src, unsigned char **dest);
void
gf_5vect_dot_prod_avx(int len, int vlen, unsigned char *gftbls, unsigned char **src,
unsigned char **dest);
/**
* @brief GF(2^8) vector dot product with five outputs.
@ -614,8 +643,9 @@ void gf_5vect_dot_prod_avx(int len, int vlen, unsigned char *gftbls,
* @returns none
*/
void gf_5vect_dot_prod_avx2(int len, int vlen, unsigned char *gftbls,
unsigned char **src, unsigned char **dest);
void
gf_5vect_dot_prod_avx2(int len, int vlen, unsigned char *gftbls, unsigned char **src,
unsigned char **dest);
/**
* @brief GF(2^8) vector dot product with six outputs.
@ -636,8 +666,9 @@ void gf_5vect_dot_prod_avx2(int len, int vlen, unsigned char *gftbls,
* @returns none
*/
void gf_6vect_dot_prod_sse(int len, int vlen, unsigned char *gftbls,
unsigned char **src, unsigned char **dest);
void
gf_6vect_dot_prod_sse(int len, int vlen, unsigned char *gftbls, unsigned char **src,
unsigned char **dest);
/**
* @brief GF(2^8) vector dot product with six outputs.
@ -658,8 +689,9 @@ void gf_6vect_dot_prod_sse(int len, int vlen, unsigned char *gftbls,
* @returns none
*/
void gf_6vect_dot_prod_avx(int len, int vlen, unsigned char *gftbls,
unsigned char **src, unsigned char **dest);
void
gf_6vect_dot_prod_avx(int len, int vlen, unsigned char *gftbls, unsigned char **src,
unsigned char **dest);
/**
* @brief GF(2^8) vector dot product with six outputs.
@ -680,8 +712,9 @@ void gf_6vect_dot_prod_avx(int len, int vlen, unsigned char *gftbls,
* @returns none
*/
void gf_6vect_dot_prod_avx2(int len, int vlen, unsigned char *gftbls,
unsigned char **src, unsigned char **dest);
void
gf_6vect_dot_prod_avx2(int len, int vlen, unsigned char *gftbls, unsigned char **src,
unsigned char **dest);
/**
* @brief GF(2^8) vector multiply accumulate, arch specific version.
@ -690,7 +723,8 @@ void gf_6vect_dot_prod_avx2(int len, int vlen, unsigned char *gftbls,
* @requires SSE4.1
*/
void gf_vect_mad_sse(int len, int vec, int vec_i, unsigned char *gftbls, unsigned char *src,
void
gf_vect_mad_sse(int len, int vec, int vec_i, unsigned char *gftbls, unsigned char *src,
unsigned char *dest);
/**
* @brief GF(2^8) vector multiply accumulate, arch specific version.
@ -699,7 +733,8 @@ void gf_vect_mad_sse(int len, int vec, int vec_i, unsigned char *gftbls, unsigne
* @requires AVX
*/
void gf_vect_mad_avx(int len, int vec, int vec_i, unsigned char *gftbls, unsigned char *src,
void
gf_vect_mad_avx(int len, int vec, int vec_i, unsigned char *gftbls, unsigned char *src,
unsigned char *dest);
/**
@ -709,10 +744,10 @@ void gf_vect_mad_avx(int len, int vec, int vec_i, unsigned char *gftbls, unsigne
* @requires AVX2
*/
void gf_vect_mad_avx2(int len, int vec, int vec_i, unsigned char *gftbls, unsigned char *src,
void
gf_vect_mad_avx2(int len, int vec, int vec_i, unsigned char *gftbls, unsigned char *src,
unsigned char *dest);
/**
* @brief GF(2^8) vector multiply with 2 accumulate. SSE version.
*
@ -734,20 +769,23 @@ void gf_vect_mad_avx2(int len, int vec, int vec_i, unsigned char *gftbls, unsign
* @returns none
*/
void gf_2vect_mad_sse(int len, int vec, int vec_i, unsigned char *gftbls, unsigned char *src,
void
gf_2vect_mad_sse(int len, int vec, int vec_i, unsigned char *gftbls, unsigned char *src,
unsigned char **dest);
/**
* @brief GF(2^8) vector multiply with 2 accumulate. AVX version of gf_2vect_mad_sse().
* @requires AVX
*/
void gf_2vect_mad_avx(int len, int vec, int vec_i, unsigned char *gftbls, unsigned char *src,
void
gf_2vect_mad_avx(int len, int vec, int vec_i, unsigned char *gftbls, unsigned char *src,
unsigned char **dest);
/**
* @brief GF(2^8) vector multiply with 2 accumulate. AVX2 version of gf_2vect_mad_sse().
* @requires AVX2
*/
void gf_2vect_mad_avx2(int len, int vec, int vec_i, unsigned char *gftbls, unsigned char *src,
void
gf_2vect_mad_avx2(int len, int vec, int vec_i, unsigned char *gftbls, unsigned char *src,
unsigned char **dest);
/**
@ -771,21 +809,24 @@ void gf_2vect_mad_avx2(int len, int vec, int vec_i, unsigned char *gftbls, unsig
* @returns none
*/
void gf_3vect_mad_sse(int len, int vec, int vec_i, unsigned char *gftbls, unsigned char *src,
void
gf_3vect_mad_sse(int len, int vec, int vec_i, unsigned char *gftbls, unsigned char *src,
unsigned char **dest);
/**
* @brief GF(2^8) vector multiply with 3 accumulate. AVX version of gf_3vect_mad_sse().
* @requires AVX
*/
void gf_3vect_mad_avx(int len, int vec, int vec_i, unsigned char *gftbls, unsigned char *src,
void
gf_3vect_mad_avx(int len, int vec, int vec_i, unsigned char *gftbls, unsigned char *src,
unsigned char **dest);
/**
* @brief GF(2^8) vector multiply with 3 accumulate. AVX2 version of gf_3vect_mad_sse().
* @requires AVX2
*/
void gf_3vect_mad_avx2(int len, int vec, int vec_i, unsigned char *gftbls, unsigned char *src,
void
gf_3vect_mad_avx2(int len, int vec, int vec_i, unsigned char *gftbls, unsigned char *src,
unsigned char **dest);
/**
@ -809,60 +850,69 @@ void gf_3vect_mad_avx2(int len, int vec, int vec_i, unsigned char *gftbls, unsig
* @returns none
*/
void gf_4vect_mad_sse(int len, int vec, int vec_i, unsigned char *gftbls, unsigned char *src,
void
gf_4vect_mad_sse(int len, int vec, int vec_i, unsigned char *gftbls, unsigned char *src,
unsigned char **dest);
/**
* @brief GF(2^8) vector multiply with 4 accumulate. AVX version of gf_4vect_mad_sse().
* @requires AVX
*/
void gf_4vect_mad_avx(int len, int vec, int vec_i, unsigned char *gftbls, unsigned char *src,
void
gf_4vect_mad_avx(int len, int vec, int vec_i, unsigned char *gftbls, unsigned char *src,
unsigned char **dest);
/**
* @brief GF(2^8) vector multiply with 4 accumulate. AVX2 version of gf_4vect_mad_sse().
* @requires AVX2
*/
void gf_4vect_mad_avx2(int len, int vec, int vec_i, unsigned char *gftbls, unsigned char *src,
void
gf_4vect_mad_avx2(int len, int vec, int vec_i, unsigned char *gftbls, unsigned char *src,
unsigned char **dest);
/**
* @brief GF(2^8) vector multiply with 5 accumulate. SSE version.
* @requires SSE4.1
*/
void gf_5vect_mad_sse(int len, int vec, int vec_i, unsigned char *gftbls, unsigned char *src,
void
gf_5vect_mad_sse(int len, int vec, int vec_i, unsigned char *gftbls, unsigned char *src,
unsigned char **dest);
/**
* @brief GF(2^8) vector multiply with 5 accumulate. AVX version.
* @requires AVX
*/
void gf_5vect_mad_avx(int len, int vec, int vec_i, unsigned char *gftbls, unsigned char *src,
void
gf_5vect_mad_avx(int len, int vec, int vec_i, unsigned char *gftbls, unsigned char *src,
unsigned char **dest);
/**
* @brief GF(2^8) vector multiply with 5 accumulate. AVX2 version.
* @requires AVX2
*/
void gf_5vect_mad_avx2(int len, int vec, int vec_i, unsigned char *gftbls, unsigned char *src,
void
gf_5vect_mad_avx2(int len, int vec, int vec_i, unsigned char *gftbls, unsigned char *src,
unsigned char **dest);
/**
* @brief GF(2^8) vector multiply with 6 accumulate. SSE version.
* @requires SSE4.1
*/
void gf_6vect_mad_sse(int len, int vec, int vec_i, unsigned char *gftbls, unsigned char *src,
void
gf_6vect_mad_sse(int len, int vec, int vec_i, unsigned char *gftbls, unsigned char *src,
unsigned char **dest);
/**
* @brief GF(2^8) vector multiply with 6 accumulate. AVX version.
* @requires AVX
*/
void gf_6vect_mad_avx(int len, int vec, int vec_i, unsigned char *gftbls, unsigned char *src,
void
gf_6vect_mad_avx(int len, int vec, int vec_i, unsigned char *gftbls, unsigned char *src,
unsigned char **dest);
/**
* @brief GF(2^8) vector multiply with 6 accumulate. AVX2 version.
* @requires AVX2
*/
void gf_6vect_mad_avx2(int len, int vec, int vec_i, unsigned char *gftbls, unsigned char *src,
void
gf_6vect_mad_avx2(int len, int vec, int vec_i, unsigned char *gftbls, unsigned char *src,
unsigned char **dest);
#endif
@ -879,7 +929,8 @@ void gf_6vect_mad_avx2(int len, int vec, int vec_i, unsigned char *gftbls, unsig
* @returns Product of a and b in GF(2^8)
*/
unsigned char gf_mul(unsigned char a, unsigned char b);
unsigned char
gf_mul(unsigned char a, unsigned char b);
/**
* @brief Single element GF(2^8) inverse.
@ -888,7 +939,8 @@ unsigned char gf_mul(unsigned char a, unsigned char b);
* @returns Field element b such that a x b = {1}
*/
unsigned char gf_inv(unsigned char a);
unsigned char
gf_inv(unsigned char a);
/**
* @brief Generate a matrix of coefficients to be used for encoding.
@ -914,7 +966,8 @@ unsigned char gf_inv(unsigned char a);
* @returns none
*/
void gf_gen_rs_matrix(unsigned char *a, int m, int k);
void
gf_gen_rs_matrix(unsigned char *a, int m, int k);
/**
* @brief Generate a Cauchy matrix of coefficients to be used for encoding.
@ -929,7 +982,8 @@ void gf_gen_rs_matrix(unsigned char *a, int m, int k);
* @returns none
*/
void gf_gen_cauchy1_matrix(unsigned char *a, int m, int k);
void
gf_gen_cauchy1_matrix(unsigned char *a, int m, int k);
/**
* @brief Invert a matrix in GF(2^8)
@ -943,8 +997,8 @@ void gf_gen_cauchy1_matrix(unsigned char *a, int m, int k);
* @returns 0 successful, other fail on singular input matrix
*/
int gf_invert_matrix(unsigned char *in, unsigned char *out, const int n);
int
gf_invert_matrix(unsigned char *in, unsigned char *out, const int n);
/*************************************************************/

24
include/gf_vect_mul.h
View File

@ -27,7 +27,6 @@
OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
**********************************************************************/
#ifndef _GF_VECT_MUL_H
#define _GF_VECT_MUL_H
@ -46,7 +45,7 @@ extern "C" {
// x86 only
#if defined(__i386__) || defined(__x86_64__)
/**
/**
* @brief GF(2^8) vector multiply by constant.
*
* Does a GF(2^8) vector multiply b = Ca where a and b are arrays and C
@ -64,10 +63,10 @@ extern "C" {
* @returns 0 pass, other fail
*/
int gf_vect_mul_sse(int len, unsigned char *gftbl, void *src, void *dest);
int
gf_vect_mul_sse(int len, unsigned char *gftbl, void *src, void *dest);
/**
/**
* @brief GF(2^8) vector multiply by constant.
*
* Does a GF(2^8) vector multiply b = Ca where a and b are arrays and C
@ -85,7 +84,8 @@ int gf_vect_mul_sse(int len, unsigned char *gftbl, void *src, void *dest);
* @returns 0 pass, other fail
*/
int gf_vect_mul_avx(int len, unsigned char *gftbl, void *src, void *dest);
int
gf_vect_mul_avx(int len, unsigned char *gftbl, void *src, void *dest);
#endif
@ -109,8 +109,8 @@ int gf_vect_mul_avx(int len, unsigned char *gftbl, void *src, void *dest);
* @returns 0 pass, other fail
*/
int gf_vect_mul(int len, unsigned char *gftbl, void *src, void *dest);
int
gf_vect_mul(int len, unsigned char *gftbl, void *src, void *dest);
/**
* @brief Initialize 32-byte constant array for GF(2^8) vector multiply
@ -122,8 +122,8 @@ int gf_vect_mul(int len, unsigned char *gftbl, void *src, void *dest);
* @param gftbl Table output.
*/
void gf_vect_mul_init(unsigned char c, unsigned char* gftbl);
void
gf_vect_mul_init(unsigned char c, unsigned char *gftbl);
/**
* @brief GF(2^8) vector multiply by constant, runs baseline version.
@ -143,8 +143,8 @@ void gf_vect_mul_init(unsigned char c, unsigned char* gftbl);
* @returns 0 pass, other fail
*/
int gf_vect_mul_base(int len, unsigned char *a, unsigned char *src,
unsigned char *dest);
int
gf_vect_mul_base(int len, unsigned char *a, unsigned char *src, unsigned char *dest);
#ifdef __cplusplus
}

133
include/igzip_lib.h
View File

@ -83,7 +83,7 @@ extern "C" {
#define IGZIP_K 1024
#define ISAL_DEF_MAX_HDR_SIZE 328
#define ISAL_DEF_MAX_CODE_LEN 15
#define ISAL_DEF_HIST_SIZE (32*IGZIP_K)
#define ISAL_DEF_HIST_SIZE (32 * IGZIP_K)
#define ISAL_DEF_MAX_HIST_BITS 15
#define ISAL_DEF_MAX_MATCH 258
#define ISAL_DEF_MIN_MATCH 3
@ -128,17 +128,17 @@ extern "C" {
#define IGZIP_LVL3_HASH_SIZE IGZIP_HASH_MAP_HASH_SIZE
#ifdef LONGER_HUFFTABLE
enum {IGZIP_DIST_TABLE_SIZE = 8*1024};
enum { IGZIP_DIST_TABLE_SIZE = 8 * 1024 };
/* DECODE_OFFSET is dist code index corresponding to DIST_TABLE_SIZE + 1 */
enum { IGZIP_DECODE_OFFSET = 26 };
#else
enum {IGZIP_DIST_TABLE_SIZE = 2};
enum { IGZIP_DIST_TABLE_SIZE = 2 };
/* DECODE_OFFSET is dist code index corresponding to DIST_TABLE_SIZE + 1 */
enum { IGZIP_DECODE_OFFSET = 0 };
#endif
enum {IGZIP_LEN_TABLE_SIZE = 256};
enum {IGZIP_LIT_TABLE_SIZE = ISAL_DEF_LIT_SYMBOLS};
enum { IGZIP_LEN_TABLE_SIZE = 256 };
enum { IGZIP_LIT_TABLE_SIZE = ISAL_DEF_LIT_SYMBOLS };
#define IGZIP_HUFFTABLE_CUSTOM 0
#define IGZIP_HUFFTABLE_DEFAULT 1
@ -172,7 +172,6 @@ enum {IGZIP_LIT_TABLE_SIZE = ISAL_DEF_LIT_SYMBOLS};
* @brief Compression State please note ZSTATE_TRL only applies for GZIP compression
*/
/* When the state is set to ZSTATE_NEW_HDR or TMP_ZSTATE_NEW_HEADER, the
* hufftable being used for compression may be swapped
*/
@ -219,7 +218,8 @@ enum isal_block_state {
ISAL_BLOCK_TYPE0, /* Decoding a type 0 block */
ISAL_BLOCK_CODED, /* Decoding a huffman coded block */
ISAL_BLOCK_INPUT_DONE, /* Decompression of input is completed */
ISAL_BLOCK_FINISH, /* Decompression of input is completed and all data has been flushed to output */
ISAL_BLOCK_FINISH, /* Decompression of input is completed and all data has been flushed to
output */
ISAL_GZIP_EXTRA_LEN,
ISAL_GZIP_EXTRA,
ISAL_GZIP_NAME,
@ -229,7 +229,6 @@ enum isal_block_state {
ISAL_CHECKSUM_CHECK,
};
/* Inflate Flags */
#define ISAL_DEFLATE 0 /* Default */
#define ISAL_GZIP 1
@ -259,7 +258,8 @@ enum isal_block_state {
/******************************************************************************/
/** @brief Holds histogram of deflate symbols*/
struct isal_huff_histogram {
uint64_t lit_len_histogram[ISAL_DEF_LIT_LEN_SYMBOLS]; //!< Histogram of Literal/Len symbols seen
uint64_t lit_len_histogram[ISAL_DEF_LIT_LEN_SYMBOLS]; //!< Histogram of Literal/Len symbols
//!< seen
uint64_t dist_histogram[ISAL_DEF_DIST_SYMBOLS]; //!< Histogram of Distance Symbols seen
uint16_t hash_table[IGZIP_LVL0_HASH_SIZE]; //!< Tmp space used as a hash table
};
@ -370,7 +370,8 @@ struct isal_zstate {
uint8_t has_eob_hdr; //!< keeps track of eob hdr (with BFINAL set)
uint8_t has_eob; //!< keeps track of eob on the last deflate block
uint8_t has_hist; //!< flag to track if there is match history
uint16_t has_level_buf_init; //!< flag to track if user supplied memory has been initialized.
uint16_t
has_level_buf_init; //!< flag to track if user supplied memory has been initialized.
uint32_t count; //!< used for partial header/trailer writes
uint8_t tmp_out_buff[16]; //!< temporary array
uint32_t tmp_out_start; //!< temporary variable
@ -389,13 +390,14 @@ struct isal_hufftables {
uint8_t deflate_hdr[ISAL_DEF_MAX_HDR_SIZE]; //!< deflate huffman tree header
uint32_t deflate_hdr_count; //!< Number of whole bytes in deflate_huff_hdr
uint32_t deflate_hdr_extra_bits; //!< Number of bits in the partial byte in header
uint32_t dist_table[IGZIP_DIST_TABLE_SIZE]; //!< bits 4:0 are the code length, bits 31:5 are the code
uint32_t len_table[IGZIP_LEN_TABLE_SIZE]; //!< bits 4:0 are the code length, bits 31:5 are the code
uint32_t dist_table[IGZIP_DIST_TABLE_SIZE]; //!< bits 4:0 are the code length, bits 31:5 are
//!< the code
uint32_t len_table[IGZIP_LEN_TABLE_SIZE]; //!< bits 4:0 are the code length, bits 31:5 are
//!< the code
uint16_t lit_table[IGZIP_LIT_TABLE_SIZE]; //!< literal code
uint8_t lit_table_sizes[IGZIP_LIT_TABLE_SIZE]; //!< literal code length
uint16_t dcodes[30 - IGZIP_DECODE_OFFSET]; //!< distance code
uint8_t dcodes_sizes[30 - IGZIP_DECODE_OFFSET]; //!< distance code length
};
/** @brief Holds stream information*/
@ -411,7 +413,7 @@ struct isal_zstream {
struct isal_hufftables *hufftables; //!< Huffman encoding used when compressing
uint32_t level; //!< Compression level to use
uint32_t level_buf_size; //!< Size of level_buf
uint8_t * level_buf; //!< User allocated buffer required for different compression levels
uint8_t *level_buf; //!< User allocated buffer required for different compression levels
uint16_t end_of_stream; //!< non-zero if this is the last input buffer
uint16_t flush; //!< Flush type can be NO_FLUSH, SYNC_FLUSH or FULL_FLUSH
uint16_t gzip_flag; //!< Indicate if gzip compression is to be performed
@ -482,8 +484,10 @@ struct isal_zstream {
#define ISAL_L_DUP ((1 << ISAL_L_REM) - (ISAL_L_REM + 1))
#define ISAL_S_DUP ((1 << ISAL_S_REM) - (ISAL_S_REM + 1))
#define ISAL_L_UNUSED ((1 << ISAL_L_REM) - (1 << ((ISAL_L_REM)/2)) - (1 << ((ISAL_L_REM + 1)/2)) + 1)
#define ISAL_S_UNUSED ((1 << ISAL_S_REM) - (1 << ((ISAL_S_REM)/2)) - (1 << ((ISAL_S_REM + 1)/2)) + 1)
#define ISAL_L_UNUSED \
((1 << ISAL_L_REM) - (1 << ((ISAL_L_REM) / 2)) - (1 << ((ISAL_L_REM + 1) / 2)) + 1)
#define ISAL_S_UNUSED \
((1 << ISAL_S_REM) - (1 << ((ISAL_S_REM) / 2)) - (1 << ((ISAL_S_REM + 1) / 2)) + 1)
#define ISAL_L_SIZE (ISAL_DEF_LIT_LEN_SYMBOLS + ISAL_L_DUP + ISAL_L_UNUSED)
#define ISAL_S_SIZE (ISAL_DEF_DIST_SYMBOLS + ISAL_S_DUP + ISAL_S_UNUSED)
@ -499,7 +503,7 @@ struct inflate_huff_code_large {
/** @brief Small lookup table for decoding huffman codes */
struct inflate_huff_code_small {
uint16_t short_code_lookup[1 << (ISAL_DECODE_SHORT_BITS)]; //!<Short code lookup table
uint16_t short_code_lookup[1 << (ISAL_DECODE_SHORT_BITS)]; //!< Short code lookup table
uint16_t long_code_lookup[ISAL_HUFF_CODE_SMALL_LONG_ALIGNED]; //!< Long code lookup table
};
@ -521,7 +525,8 @@ struct inflate_state {
uint32_t crc; //!< Contains crc or adler32 of output if crc_flag is set
uint32_t hist_bits; //!< Log base 2 of maximum lookback distance
union {
int32_t type0_block_len; //!< Length left to read of type 0 block when outbuffer overflow occurred
int32_t type0_block_len; //!< Length left to read of type 0 block when outbuffer
//!< overflow occurred
int32_t count; //!< Count of bytes remaining to be parsed
uint32_t dict_id;
};
@ -533,8 +538,11 @@ struct inflate_state {
int16_t tmp_in_size; //!< Number of bytes in tmp_in_buffer
int32_t tmp_out_valid; //!< Number of bytes in tmp_out_buffer
int32_t tmp_out_processed; //!< Number of bytes processed in tmp_out_buffer
uint8_t tmp_in_buffer[ISAL_DEF_MAX_HDR_SIZE]; //!< Temporary buffer containing data from the input stream
uint8_t tmp_out_buffer[2 * ISAL_DEF_HIST_SIZE + ISAL_LOOK_AHEAD]; //!< Temporary buffer containing data from the output stream
uint8_t tmp_in_buffer[ISAL_DEF_MAX_HDR_SIZE]; //!< Temporary buffer containing data from the
//!< input stream
uint8_t tmp_out_buffer[2 * ISAL_DEF_HIST_SIZE +
ISAL_LOOK_AHEAD]; //!< Temporary buffer containing data from the
//!< output stream
};
/******************************************************************************/
@ -551,8 +559,8 @@ struct inflate_state {
* @param length: The length of start_stream.
* @param histogram: The returned histogram of lit/len/dist symbols.
*/
void isal_update_histogram(uint8_t * in_stream, int length, struct isal_huff_histogram * histogram);
void
isal_update_histogram(uint8_t *in_stream, int length, struct isal_huff_histogram *histogram);
/**
* @brief Creates a custom huffman code for the given histograms in which
@ -564,8 +572,8 @@ void isal_update_histogram(uint8_t * in_stream, int length, struct isal_huff_his
* repeat lengths and lookback distances
* @returns Returns a non zero value if an invalid huffman code was created.
*/
int isal_create_hufftables(struct isal_hufftables * hufftables,
struct isal_huff_histogram * histogram);
int
isal_create_hufftables(struct isal_hufftables *hufftables, struct isal_huff_histogram *histogram);
/**
* @brief Creates a custom huffman code for the given histograms like
@ -577,8 +585,9 @@ int isal_create_hufftables(struct isal_hufftables * hufftables,
* repeat lengths and lookback distances
* @returns Returns a non zero value if an invalid huffman code was created.
*/
int isal_create_hufftables_subset(struct isal_hufftables * hufftables,
struct isal_huff_histogram * histogram);
int
isal_create_hufftables_subset(struct isal_hufftables *hufftables,
struct isal_huff_histogram *histogram);
/**
* @brief Initialize compression stream data structure
@ -586,7 +595,8 @@ int isal_create_hufftables_subset(struct isal_hufftables * hufftables,
* @param stream Structure holding state information on the compression streams.
* @returns none
*/
void isal_deflate_init(struct isal_zstream *stream);
void
isal_deflate_init(struct isal_zstream *stream);
/**
* @brief Reinitialize compression stream data structure. Performs the same
@ -597,22 +607,24 @@ void isal_deflate_init(struct isal_zstream *stream);
* @param stream Structure holding state information on the compression streams.
* @returns none
*/
void isal_deflate_reset(struct isal_zstream *stream);
void
isal_deflate_reset(struct isal_zstream *stream);
/**
* @brief Set gzip header default values
*
* @param gz_hdr: Gzip header to initialize.
*/
void isal_gzip_header_init(struct isal_gzip_header *gz_hdr);
void
isal_gzip_header_init(struct isal_gzip_header *gz_hdr);
/**
* @brief Set zlib header default values
*
* @param z_hdr: zlib header to initialize.
*/
void isal_zlib_header_init(struct isal_zlib_header *z_hdr);
void
isal_zlib_header_init(struct isal_zlib_header *z_hdr);
/**
* @brief Write gzip header to output stream
@ -629,7 +641,8 @@ void isal_zlib_header_init(struct isal_zlib_header *z_hdr);
* the minimum size required to successfully write the gzip header to the output
* buffer.
*/
uint32_t isal_write_gzip_header(struct isal_zstream * stream, struct isal_gzip_header *gz_hdr);
uint32_t
isal_write_gzip_header(struct isal_zstream *stream, struct isal_gzip_header *gz_hdr);
/**
* @brief Write zlib header to output stream
@ -646,7 +659,8 @@ uint32_t isal_write_gzip_header(struct isal_zstream * stream, struct isal_gzip_h
* the minimum size required to successfully write the zlib header to the output
* buffer.
*/
uint32_t isal_write_zlib_header(struct isal_zstream * stream, struct isal_zlib_header *z_hdr);
uint32_t
isal_write_zlib_header(struct isal_zstream *stream, struct isal_zlib_header *z_hdr);
/**
* @brief Set stream to use a new Huffman code
@ -668,8 +682,9 @@ uint32_t isal_write_zlib_header(struct isal_zstream * stream, struct isal_zlib_h
* due to the stream being in a state where changing the huffman code is not
* allowed or an invalid input is provided.
*/
int isal_deflate_set_hufftables(struct isal_zstream *stream,
struct isal_hufftables *hufftables, int type);
int
isal_deflate_set_hufftables(struct isal_zstream *stream, struct isal_hufftables *hufftables,
int type);
/**
* @brief Initialize compression stream data structure
@ -677,8 +692,8 @@ int isal_deflate_set_hufftables(struct isal_zstream *stream,
* @param stream Structure holding state information on the compression streams.
* @returns none
*/
void isal_deflate_stateless_init(struct isal_zstream *stream);
void
isal_deflate_stateless_init(struct isal_zstream *stream);
/**
* @brief Set compression dictionary to use
@ -694,7 +709,8 @@ void isal_deflate_stateless_init(struct isal_zstream *stream);
* @returns COMP_OK,
* ISAL_INVALID_STATE (dictionary could not be set)
*/
int isal_deflate_set_dict(struct isal_zstream *stream, uint8_t *dict, uint32_t dict_len);
int
isal_deflate_set_dict(struct isal_zstream *stream, uint8_t *dict, uint32_t dict_len);
/** @brief Structure for holding processed dictionary information */
@ -725,8 +741,9 @@ struct isal_dict {
* @returns COMP_OK,
* ISAL_INVALID_STATE (dictionary could not be processed)
*/
int isal_deflate_process_dict(struct isal_zstream *stream, struct isal_dict *dict_str,
uint8_t *dict, uint32_t dict_len);
int
isal_deflate_process_dict(struct isal_zstream *stream, struct isal_dict *dict_str, uint8_t *dict,
uint32_t dict_len);
/**
* @brief Reset compression dictionary to use
@ -745,8 +762,8 @@ int isal_deflate_process_dict(struct isal_zstream *stream, struct isal_dict *dic
* @returns COMP_OK,
* ISAL_INVALID_STATE or other (dictionary could not be reset)
*/
int isal_deflate_reset_dict(struct isal_zstream *stream, struct isal_dict *dict_str);
int
isal_deflate_reset_dict(struct isal_zstream *stream, struct isal_dict *dict_str);
/**
* @brief Fast data (deflate) compression for storage applications.
@ -802,8 +819,8 @@ int isal_deflate_reset_dict(struct isal_zstream *stream, struct isal_dict *dict_
* ISAL_INVALID_LEVEL (if an invalid compression level is selected),
* ISAL_INVALID_LEVEL_BUF (if the level buffer is not large enough).
*/
int isal_deflate(struct isal_zstream *stream);
int
isal_deflate(struct isal_zstream *stream);
/**
* @brief Fast data (deflate) stateless compression for storage applications.
@ -832,8 +849,8 @@ int isal_deflate(struct isal_zstream *stream);
* ISAL_INVALID_LEVEL_BUF (if the level buffer is not large enough),
* STATELESS_OVERFLOW (if output buffer will not fit output).
*/
int isal_deflate_stateless(struct isal_zstream *stream);
int
isal_deflate_stateless(struct isal_zstream *stream);
/******************************************************************************/
/* Inflate functions */
@ -844,7 +861,8 @@ int isal_deflate_stateless(struct isal_zstream *stream);
* @param state Structure holding state information on the compression streams.
* @returns none
*/
void isal_inflate_init(struct inflate_state *state);
void
isal_inflate_init(struct inflate_state *state);
/**
* @brief Reinitialize decompression state data structure
@ -852,7 +870,8 @@ void isal_inflate_init(struct inflate_state *state);
* @param state Structure holding state information on the compression streams.
* @returns none
*/
void isal_inflate_reset(struct inflate_state *state);
void
isal_inflate_reset(struct inflate_state *state);
/**
* @brief Set decompression dictionary to use
@ -867,7 +886,8 @@ void isal_inflate_reset(struct inflate_state *state);
* @returns COMP_OK,
* ISAL_INVALID_STATE (dictionary could not be set)
*/
int isal_inflate_set_dict(struct inflate_state *state, uint8_t *dict, uint32_t dict_len);
int
isal_inflate_set_dict(struct inflate_state *state, uint8_t *dict, uint32_t dict_len);
/**
* @brief Read and return gzip header information
@ -890,7 +910,8 @@ int isal_inflate_set_dict(struct inflate_state *state, uint8_t *dict, uint32_t d
* ISAL_UNSUPPORTED_METHOD (deflate is not the compression method),
* ISAL_INCORRECT_CHECKSUM (gzip header checksum was incorrect)
*/
int isal_read_gzip_header (struct inflate_state *state, struct isal_gzip_header *gz_hdr);
int
isal_read_gzip_header(struct inflate_state *state, struct isal_gzip_header *gz_hdr);
/**
* @brief Read and return zlib header information
@ -905,7 +926,8 @@ int isal_read_gzip_header (struct inflate_state *state, struct isal_gzip_header
* ISAL_UNSUPPORTED_METHOD (deflate is not the compression method),
* ISAL_INCORRECT_CHECKSUM (zlib header checksum was incorrect)
*/
int isal_read_zlib_header (struct inflate_state *state, struct isal_zlib_header *zlib_hdr);
int
isal_read_zlib_header(struct inflate_state *state, struct isal_zlib_header *zlib_hdr);
/**
* @brief Fast data (deflate) decompression for storage applications.
@ -953,7 +975,8 @@ int isal_read_zlib_header (struct inflate_state *state, struct isal_zlib_header
* ISAL_INCORRECT_CHECKSUM.
*/
int isal_inflate(struct inflate_state *state);
int
isal_inflate(struct inflate_state *state);
/**
* @brief Fast data (deflate) stateless decompression for storage applications.
@ -975,7 +998,8 @@ int isal_inflate(struct inflate_state *state);
* ISAL_UNSUPPORTED_METHOD,
* ISAL_INCORRECT_CHECKSUM.
*/
int isal_inflate_stateless(struct inflate_state *state);
int
isal_inflate_stateless(struct inflate_state *state);
/******************************************************************************/
/* Other functions */
@ -992,7 +1016,8 @@ int isal_inflate_stateless(struct inflate_state *state);
*
* @returns 32-bit Adler-32 checksum
*/
uint32_t isal_adler32(uint32_t init, const unsigned char *buf, uint64_t len);
uint32_t
isal_adler32(uint32_t init, const unsigned char *buf, uint64_t len);
#ifdef __cplusplus
}

5
include/mem_routines.h
View File

@ -36,7 +36,6 @@
* Defines the interface for vector versions of common memory functions.
*/
#ifndef _MEM_ROUTINES_H_
#define _MEM_ROUTINES_H_
@ -54,11 +53,11 @@ extern "C" {
* @returns 0 - region is all zeros
* other - region has non zero bytes
*/
int isal_zero_detect(void *mem, size_t len);
int
isal_zero_detect(void *mem, size_t len);
#ifdef __cplusplus
}
#endif
#endif // _MEM_ROUTINES_H_

65
include/raid.h
View File

@ -27,7 +27,6 @@
OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
**********************************************************************/
#ifndef _RAID_H_
#define _RAID_H_
@ -61,8 +60,8 @@ extern "C" {
* @returns 0 pass, other fail
*/
int xor_gen(int vects, int len, void **array);
int
xor_gen(int vects, int len, void **array);
/**
* @brief Checks that array has XOR parity sum of 0 across all vectors, runs appropriate version.
@ -78,8 +77,8 @@ int xor_gen(int vects, int len, void **array);
* @returns 0 pass, other fail
*/
int xor_check(int vects, int len, void **array);
int
xor_check(int vects, int len, void **array);
/**
* @brief Generate P+Q parity vectors from N sources, runs appropriate version.
@ -98,11 +97,12 @@ int xor_check(int vects, int len, void **array);
* @returns 0 pass, other fail
*/
int pq_gen(int vects, int len, void **array);
int
pq_gen(int vects, int len, void **array);
/**
* @brief Checks that array of N sources, P and Q are consistent across all vectors, runs appropriate version.
* @brief Checks that array of N sources, P and Q are consistent across all vectors, runs
* appropriate version.
*
* This function determines what instruction sets are enabled and
* selects the appropriate version at runtime.
@ -116,8 +116,8 @@ int pq_gen(int vects, int len, void **array);
* @returns 0 pass, other fail
*/
int pq_check(int vects, int len, void **array);
int
pq_check(int vects, int len, void **array);
/* Arch specific versions */
// x86 only
@ -136,8 +136,8 @@ int pq_check(int vects, int len, void **array);
* @returns 0 pass, other fail
*/
int xor_gen_sse(int vects, int len, void **array);
int
xor_gen_sse(int vects, int len, void **array);
/**
* @brief Generate XOR parity vector from N sources.
@ -152,8 +152,8 @@ int xor_gen_sse(int vects, int len, void **array);
* @returns 0 pass, other fail
*/
int xor_gen_avx(int vects, int len, void **array);
int
xor_gen_avx(int vects, int len, void **array);
/**
* @brief Checks that array has XOR parity sum of 0 across all vectors.
@ -167,8 +167,8 @@ int xor_gen_avx(int vects, int len, void **array);
* @returns 0 pass, other fail
*/
int xor_check_sse(int vects, int len, void **array);
int
xor_check_sse(int vects, int len, void **array);
/**
* @brief Generate P+Q parity vectors from N sources.
@ -185,8 +185,8 @@ int xor_check_sse(int vects, int len, void **array);
* @returns 0 pass, other fail
*/
int pq_gen_sse(int vects, int len, void **array);
int
pq_gen_sse(int vects, int len, void **array);
/**
* @brief Generate P+Q parity vectors from N sources.
@ -203,8 +203,8 @@ int pq_gen_sse(int vects, int len, void **array);
* @returns 0 pass, other fail
*/
int pq_gen_avx(int vects, int len, void **array);
int
pq_gen_avx(int vects, int len, void **array);
/**
* @brief Generate P+Q parity vectors from N sources.
@ -221,8 +221,8 @@ int pq_gen_avx(int vects, int len, void **array);
* @returns 0 pass, other fail
*/
int pq_gen_avx2(int vects, int len, void **array);
int
pq_gen_avx2(int vects, int len, void **array);
/**
* @brief Checks that array of N sources, P and Q are consistent across all vectors.
@ -236,7 +236,8 @@ int pq_gen_avx2(int vects, int len, void **array);
* @returns 0 pass, other fail
*/
int pq_check_sse(int vects, int len, void **array);
int
pq_check_sse(int vects, int len, void **array);
#endif
@ -253,8 +254,8 @@ int pq_check_sse(int vects, int len, void **array);
* @returns 0 pass, other fail
*/
int pq_gen_base(int vects, int len, void **array);
int
pq_gen_base(int vects, int len, void **array);
/**
* @brief Generate XOR parity vector from N sources, runs baseline version.
@ -267,8 +268,8 @@ int pq_gen_base(int vects, int len, void **array);
* @returns 0 pass, other fail
*/
int xor_gen_base(int vects, int len, void **array);
int
xor_gen_base(int vects, int len, void **array);
/**
* @brief Checks that array has XOR parity sum of 0 across all vectors, runs baseline version.
@ -281,11 +282,12 @@ int xor_gen_base(int vects, int len, void **array);
* @returns 0 pass, other fail
*/
int xor_check_base(int vects, int len, void **array);
int
xor_check_base(int vects, int len, void **array);
/**
* @brief Checks that array of N sources, P and Q are consistent across all vectors, runs baseline version.
* @brief Checks that array of N sources, P and Q are consistent across all vectors, runs baseline
* version.
*
* @param vects Number of vectors in array including P&Q. Must be > 3.
* @param len Length of each vector in bytes. Must be 16B aligned.
@ -296,7 +298,8 @@ int xor_check_base(int vects, int len, void **array);
* @returns 0 pass, other fail
*/
int pq_check_base(int vects, int len, void **array);
int
pq_check_base(int vects, int len, void **array);
#ifdef __cplusplus
}

174
include/test.h
View File

@ -47,30 +47,34 @@ extern "C" {
#include <stdint.h>
#ifdef _MSC_VER
# define inline __inline
#define inline __inline
#endif
/* Make os-independent alignment attribute, alloc and free. */
#if defined __unix__ || defined __APPLE__
# define DECLARE_ALIGNED(decl, alignval) decl __attribute__((aligned(alignval)))
# define __forceinline static inline
# define aligned_free(x) free(x)
#define DECLARE_ALIGNED(decl, alignval) decl __attribute__((aligned(alignval)))
#define __forceinline static inline
#define aligned_free(x) free(x)
#else
# ifdef __MINGW32__
# define DECLARE_ALIGNED(decl, alignval) decl __attribute__((aligned(alignval)))
# define posix_memalign(p, algn, len) (NULL == (*((char**)(p)) = (void*) _aligned_malloc(len, algn)))
# define aligned_free(x) _aligned_free(x)
# else
# define DECLARE_ALIGNED(decl, alignval) __declspec(align(alignval)) decl
# define posix_memalign(p, algn, len) (NULL == (*((char**)(p)) = (void*) _aligned_malloc(len, algn)))
# define aligned_free(x) _aligned_free(x)
# endif
#ifdef __MINGW32__
#define DECLARE_ALIGNED(decl, alignval) decl __attribute__((aligned(alignval)))
#define posix_memalign(p, algn, len) \
(NULL == (*((char **) (p)) = (void *) _aligned_malloc(len, algn)))
#define aligned_free(x) _aligned_free(x)
#else
#define DECLARE_ALIGNED(decl, alignval) __declspec(align(alignval)) decl
#define posix_memalign(p, algn, len) \
(NULL == (*((char **) (p)) = (void *) _aligned_malloc(len, algn)))
#define aligned_free(x) _aligned_free(x)
#endif
#endif
#ifdef DEBUG
# define DEBUG_PRINT(x) printf x
#define DEBUG_PRINT(x) printf x
#else
# define DEBUG_PRINT(x) do {} while (0)
#define DEBUG_PRINT(x) \
do { \
} while (0)
#endif
/* Decide whether to use benchmark time as an approximation or a minimum. Fewer
@ -86,58 +90,68 @@ extern "C" {
* standardized clock source. To obtain a meaningful result it may be
* necessary to fix the CPU clock to match the rtdsc tick rate.
*/
# include <inttypes.h>
# include <x86intrin.h>
# define USE_CYCLES
#include <inttypes.h>
#include <x86intrin.h>
#define USE_CYCLES
#else
# include <time.h>
#include <time.h>
#define USE_SECONDS
#endif
#ifdef USE_RDTSC
#ifndef BENCHMARK_TIME
# define BENCHMARK_TIME 6
#define BENCHMARK_TIME 6
#endif
# define GHZ 1000000000
# define UNIT_SCALE (GHZ)
# define CALIBRATE_TIME (UNIT_SCALE / 2)
static inline long long get_time(void) {
#define GHZ 1000000000
#define UNIT_SCALE (GHZ)
#define CALIBRATE_TIME (UNIT_SCALE / 2)
static inline long long
get_time(void)
{
unsigned int dummy;
return __rdtscp(&dummy);
}
static inline long long get_res(void) {
static inline long long
get_res(void)
{
return 1;
}
#else
#ifndef BENCHMARK_TIME
# define BENCHMARK_TIME 3
#define BENCHMARK_TIME 3
#endif
#ifdef _MSC_VER
#define UNIT_SCALE get_res()
#define CALIBRATE_TIME (UNIT_SCALE / 4)
static inline long long get_time(void) {
static inline long long
get_time(void)
{
long long ret = 0;
QueryPerformanceCounter(&ret);
return ret;
}
static inline long long get_res(void) {
static inline long long
get_res(void)
{
long long ret = 0;
QueryPerformanceFrequency(&ret);
return ret;
}
#else
# define NANO_SCALE 1000000000
# define UNIT_SCALE NANO_SCALE
# define CALIBRATE_TIME (UNIT_SCALE / 4)
#define NANO_SCALE 1000000000
#define UNIT_SCALE NANO_SCALE
#define CALIBRATE_TIME (UNIT_SCALE / 4)
#ifdef __FreeBSD__
# define CLOCK_ID CLOCK_MONOTONIC_PRECISE
#define CLOCK_ID CLOCK_MONOTONIC_PRECISE
#else
# define CLOCK_ID CLOCK_MONOTONIC
#define CLOCK_ID CLOCK_MONOTONIC
#endif
static inline long long get_time(void) {
static inline long long
get_time(void)
{
struct timespec time;
long long nano_total;
clock_gettime(CLOCK_ID, &time);
@ -147,7 +161,9 @@ static inline long long get_time(void) {
return nano_total;
}
static inline long long get_res(void) {
static inline long long
get_res(void)
{
struct timespec time;
long long nano_total;
clock_getres(CLOCK_ID, &time);
@ -165,42 +181,56 @@ struct perf {
long long iterations;
};
static inline void perf_init(struct perf *p) {
static inline void
perf_init(struct perf *p)
{
p->start = 0;
p->stop = 0;
p->run_total = 0;
}
static inline void perf_continue(struct perf *p) {
static inline void
perf_continue(struct perf *p)
{
p->start = get_time();
}
static inline void perf_pause(struct perf *p) {
static inline void
perf_pause(struct perf *p)
{
p->stop = get_time();
p->run_total = p->run_total + p->stop - p->start;
p->start = p->stop;
}
static inline void perf_start(struct perf *p) {
static inline void
perf_start(struct perf *p)
{
perf_init(p);
perf_continue(p);
}
static inline void perf_stop(struct perf *p) {
static inline void
perf_stop(struct perf *p)
{
perf_pause(p);
}
static inline double get_time_elapsed(struct perf *p) {
static inline double
get_time_elapsed(struct perf *p)
{
return 1.0 * p->run_total / UNIT_SCALE;
}
static inline long long get_base_elapsed(struct perf *p) {
static inline long long
get_base_elapsed(struct perf *p)
{
return p->run_total;
}
static inline unsigned long long estimate_perf_iterations(struct perf *p,
unsigned long long runs,
unsigned long long total) {
static inline unsigned long long
estimate_perf_iterations(struct perf *p, unsigned long long runs, unsigned long long total)
{
total = total * runs;
if (get_base_elapsed(p) > 0)
return (total + get_base_elapsed(p) - 1) / get_base_elapsed(p);
@ -208,29 +238,30 @@ static inline unsigned long long estimate_perf_iterations(struct perf *p,
return (total + get_res() - 1) / get_res();
}
#define CALIBRATE(PERF, FUNC_CALL) { \
#define CALIBRATE(PERF, FUNC_CALL) \
{ \
unsigned long long _i, _iter = 1; \
perf_start(PERF); \
FUNC_CALL; \
perf_pause(PERF); \
\
while (get_base_elapsed(PERF) < CALIBRATE_TIME) { \
_iter = estimate_perf_iterations(PERF, _iter, \
2 * CALIBRATE_TIME); \
_iter = estimate_perf_iterations(PERF, _iter, 2 * CALIBRATE_TIME); \
perf_start(PERF); \
for (_i = 0; _i < _iter; _i++) { \
FUNC_CALL; \
} \
perf_stop(PERF); \
} \
(PERF)->iterations=_iter; \
}
(PERF)->iterations = _iter; \
}
#define PERFORMANCE_TEST(PERF, RUN_TIME, FUNC_CALL) { \
#define PERFORMANCE_TEST(PERF, RUN_TIME, FUNC_CALL) \
{ \
unsigned long long _i, _iter = (PERF)->iterations; \
unsigned long long _run_total = RUN_TIME; \
_run_total *= UNIT_SCALE; \
_iter = estimate_perf_iterations(PERF, _iter, _run_total);\
_iter = estimate_perf_iterations(PERF, _iter, _run_total); \
(PERF)->iterations = 0; \
perf_start(PERF); \
for (_i = 0; _i < _iter; _i++) { \
@ -239,8 +270,7 @@ static inline unsigned long long estimate_perf_iterations(struct perf *p,
perf_pause(PERF); \
(PERF)->iterations += _iter; \
\
if(get_base_elapsed(PERF) < _run_total && \
BENCHMARK_TYPE == BENCHMARK_MIN_TIME) { \
if (get_base_elapsed(PERF) < _run_total && BENCHMARK_TYPE == BENCHMARK_MIN_TIME) { \
_iter = estimate_perf_iterations(PERF, _iter, \
_run_total - get_base_elapsed(PERF) + \
(UNIT_SCALE / 16)); \
@ -251,10 +281,11 @@ static inline unsigned long long estimate_perf_iterations(struct perf *p,
perf_pause(PERF); \
(PERF)->iterations += _iter; \
} \
}
}
#define BENCHMARK(PERF, RUN_TIME, FUNC_CALL) { \
if((RUN_TIME) > 0) { \
#define BENCHMARK(PERF, RUN_TIME, FUNC_CALL) \
{ \
if ((RUN_TIME) > 0) { \
CALIBRATE(PERF, FUNC_CALL); \
PERFORMANCE_TEST(PERF, RUN_TIME, FUNC_CALL); \
\
@ -264,36 +295,41 @@ static inline unsigned long long estimate_perf_iterations(struct perf *p,
FUNC_CALL; \
perf_stop(PERF); \
} \
}
}
#ifdef USE_CYCLES
static inline void perf_print(struct perf p, long long unit_count) {
static inline void
perf_print(struct perf p, long long unit_count)
{
long long total_units = p.iterations * unit_count;
printf("runtime = %10lld ticks", get_base_elapsed(&p));
if (total_units != 0) {
printf(", bandwidth %lld MB in %.4f GC = %.2f ticks/byte",
total_units / (1000000), get_time_elapsed(&p),
get_base_elapsed(&p) / (double)total_units);
printf(", bandwidth %lld MB in %.4f GC = %.2f ticks/byte", total_units / (1000000),
get_time_elapsed(&p), get_base_elapsed(&p) / (double) total_units);
}
printf("\n");
}
#else
static inline void perf_print(struct perf p, double unit_count) {
static inline void
perf_print(struct perf p, double unit_count)
{
long long total_units = p.iterations * unit_count;
long long usecs = (long long)(get_time_elapsed(&p) * 1000000);
long long usecs = (long long) (get_time_elapsed(&p) * 1000000);
printf("runtime = %10lld usecs", usecs);
if (total_units != 0) {
printf(", bandwidth %lld MB in %.4f sec = %.2f MB/s",
total_units / (1000000), get_time_elapsed(&p),
((double)total_units) / (1000000 * get_time_elapsed(&p)));
printf(", bandwidth %lld MB in %.4f sec = %.2f MB/s", total_units / (1000000),
get_time_elapsed(&p),
((double) total_units) / (1000000 * get_time_elapsed(&p)));
}
printf("\n");
}
#endif
static inline uint64_t get_filesize(FILE * fp) {
static inline uint64_t
get_filesize(FILE *fp)
{
uint64_t file_size;
fpos_t pos, pos_curr;
@ -304,7 +340,7 @@ static inline uint64_t get_filesize(FILE * fp) {
fseeko(fp, 0, SEEK_END);
#endif
fgetpos(fp, &pos);
file_size = *(uint64_t *) & pos;
file_size = *(uint64_t *) &pos;
fsetpos(fp, &pos_curr); /* Restore position */
return file_size;

138
include/unaligned.h
View File

@ -37,105 +37,115 @@
#ifdef __FreeBSD__
#include <sys/types.h>
#include <sys/endian.h>
# define isal_bswap16(x) bswap16(x)
# define isal_bswap32(x) bswap32(x)
# define isal_bswap64(x) bswap64(x)
#elif defined (__APPLE__)
#define isal_bswap16(x) bswap16(x)
#define isal_bswap32(x) bswap32(x)
#define isal_bswap64(x) bswap64(x)
#elif defined(__APPLE__)
#include <libkern/OSByteOrder.h>
# define isal_bswap16(x) OSSwapInt16(x)
# define isal_bswap32(x) OSSwapInt32(x)
# define isal_bswap64(x) OSSwapInt64(x)
#elif defined (__GNUC__) && !defined (__MINGW32__)
# include <byteswap.h>
# define isal_bswap16(x) bswap_16(x)
# define isal_bswap32(x) bswap_32(x)
# define isal_bswap64(x) bswap_64(x)
#define isal_bswap16(x) OSSwapInt16(x)
#define isal_bswap32(x) OSSwapInt32(x)
#define isal_bswap64(x) OSSwapInt64(x)
#elif defined(__GNUC__) && !defined(__MINGW32__)
#include <byteswap.h>
#define isal_bswap16(x) bswap_16(x)
#define isal_bswap32(x) bswap_32(x)
#define isal_bswap64(x) bswap_64(x)
#elif defined _WIN64
# define isal_bswap16(x) _byteswap_ushort(x)
# define isal_bswap32(x) _byteswap_ulong(x)
# define isal_bswap64(x) _byteswap_uint64(x)
#define isal_bswap16(x) _byteswap_ushort(x)
#define isal_bswap32(x) _byteswap_ulong(x)
#define isal_bswap64(x) _byteswap_uint64(x)
#endif
#if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
# define to_be16(x) isal_bswap16(x)
# define from_be16(x) isal_bswap16(x)
# define to_be32(x) isal_bswap32(x)
# define from_be32(x) isal_bswap32(x)
# define to_be64(x) isal_bswap64(x)
# define from_be64(x) isal_bswap64(x)
# define to_le16(x) (x)
# define from_le16(x) (x)
# define to_le32(x) (x)
# define from_le32(x) (x)
# define to_le64(x) (x)
# define from_le64(x) (x)
#define to_be16(x) isal_bswap16(x)
#define from_be16(x) isal_bswap16(x)
#define to_be32(x) isal_bswap32(x)
#define from_be32(x) isal_bswap32(x)
#define to_be64(x) isal_bswap64(x)
#define from_be64(x) isal_bswap64(x)
#define to_le16(x) (x)
#define from_le16(x) (x)
#define to_le32(x) (x)
#define from_le32(x) (x)
#define to_le64(x) (x)
#define from_le64(x) (x)
#else
# define to_be16(x) (x)
# define from_be16(x) (x)
# define to_be32(x) (x)
# define from_be32(x) (x)
# define to_be64(x) (x)
# define from_be64(x) (x)
# define to_le16(x) isal_bswap16(x)
# define from_le16(x) isal_bswap16(x)
# define to_le32(x) isal_bswap32(x)
# define from_le32(x) isal_bswap32(x)
# define to_le64(x) isal_bswap64(x)
# define from_le64(x) isal_bswap64(x)
#define to_be16(x) (x)
#define from_be16(x) (x)
#define to_be32(x) (x)
#define from_be32(x) (x)
#define to_be64(x) (x)
#define from_be64(x) (x)
#define to_le16(x) isal_bswap16(x)
#define from_le16(x) isal_bswap16(x)
#define to_le32(x) isal_bswap32(x)
#define from_le32(x) isal_bswap32(x)
#define to_le64(x) isal_bswap64(x)
#define from_le64(x) isal_bswap64(x)
#endif
static inline uint16_t load_native_u16(uint8_t * buf)
static inline uint16_t
load_native_u16(uint8_t *buf)
{
uint16_t ret;
memcpy(&ret, buf, sizeof(ret));
return ret;
}
static inline uint16_t load_le_u16(uint8_t * buf)
static inline uint16_t
load_le_u16(uint8_t *buf)
{
return from_le16(load_native_u16(buf));
}
static inline uint16_t load_be_u16(uint8_t * buf)
static inline uint16_t
load_be_u16(uint8_t *buf)
{
return from_be16(load_native_u16(buf));
}
static inline uint32_t load_native_u32(uint8_t * buf)
static inline uint32_t
load_native_u32(uint8_t *buf)
{
uint32_t ret;
memcpy(&ret, buf, sizeof(ret));
return ret;
}
static inline uint32_t load_le_u32(uint8_t * buf)
static inline uint32_t
load_le_u32(uint8_t *buf)
{
return from_le32(load_native_u32(buf));
}
static inline uint32_t load_be_u32(uint8_t * buf)
static inline uint32_t
load_be_u32(uint8_t *buf)
{
return from_be32(load_native_u32(buf));
}
static inline uint64_t load_native_u64(uint8_t * buf)
static inline uint64_t
load_native_u64(uint8_t *buf)
{
uint64_t ret;
memcpy(&ret, buf, sizeof(ret));
return ret;
}
static inline uint64_t load_le_u64(uint8_t * buf)
static inline uint64_t
load_le_u64(uint8_t *buf)
{
return from_le64(load_native_u64(buf));
}
static inline uint64_t load_be_u64(uint8_t * buf)
static inline uint64_t
load_be_u64(uint8_t *buf)
{
return from_be64(load_native_u64(buf));
}
static inline uintmax_t load_le_umax(uint8_t * buf)
static inline uintmax_t
load_le_umax(uint8_t *buf)
{
switch (sizeof(uintmax_t)) {
case sizeof(uint32_t):
@ -147,22 +157,26 @@ static inline uintmax_t load_le_umax(uint8_t * buf)
}
}
static inline void store_native_u16(uint8_t * buf, uint16_t val)
static inline void
store_native_u16(uint8_t *buf, uint16_t val)
{
memcpy(buf, &val, sizeof(val));
}
static inline void store_le_u16(uint8_t * buf, uint16_t val)
static inline void
store_le_u16(uint8_t *buf, uint16_t val)
{
store_native_u16(buf, to_le16(val));
}
static inline void store_be_u16(uint8_t * buf, uint16_t val)
static inline void
store_be_u16(uint8_t *buf, uint16_t val)
{
store_native_u16(buf, to_be16(val));
}
static inline void store_native_u16_to_u64(uint64_t * buf, uint16_t val)
static inline void
store_native_u16_to_u64(uint64_t *buf, uint16_t val)
{
#if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
store_native_u16((uint8_t *) buf, val);
@ -171,32 +185,38 @@ static inline void store_native_u16_to_u64(uint64_t * buf, uint16_t val)
#endif
}
static inline void store_native_u32(uint8_t * buf, uint32_t val)
static inline void
store_native_u32(uint8_t *buf, uint32_t val)
{
memcpy(buf, &val, sizeof(val));
}
static inline void store_le_u32(uint8_t * buf, uint32_t val)
static inline void
store_le_u32(uint8_t *buf, uint32_t val)
{
store_native_u32(buf, to_le32(val));
}
static inline void store_be_u32(uint8_t * buf, uint32_t val)
static inline void
store_be_u32(uint8_t *buf, uint32_t val)
{
store_native_u32(buf, to_be32(val));
}
static inline void store_native_u64(uint8_t * buf, uint64_t val)
static inline void
store_native_u64(uint8_t *buf, uint64_t val)
{
memcpy(buf, &val, sizeof(val));
}
static inline void store_le_u64(uint8_t * buf, uint64_t val)
static inline void
store_le_u64(uint8_t *buf, uint64_t val)
{
store_native_u64(buf, to_le64(val));
}
static inline void store_be_u64(uint8_t * buf, uint64_t val)
static inline void
store_be_u64(uint8_t *buf, uint64_t val)
{
store_native_u64(buf, to_be64(val));
}