/*********************************************************************
* Filename:   sha256.c
* Author:     Brad Conte (brad AT bradconte.com)
* Copyright:
* Disclaimer: This code is presented "as is" without any guarantees.
* Details:    Implementation of the SHA-256 hashing algorithm.
              SHA-256 is one of the three algorithms in the SHA2
              specification. The others, SHA-384 and SHA-512, are not
              offered in this implementation.
              Algorithm specification can be found here:
               * http://csrc.nist.gov/publications/fips/fips180-2/fips180-2withchangenotice.pdf
              This implementation uses little endian byte order.
*********************************************************************/

/*************************** HEADER FILES ***************************/
#include <stdlib.h>
#include <stdio.h>
#include <memory.h>
#include "sha256.h"

static const uint32_t K[] =
{
    0x428A2F98, 0x71374491, 0xB5C0FBCF, 0xE9B5DBA5,
    0x3956C25B, 0x59F111F1, 0x923F82A4, 0xAB1C5ED5,
    0xD807AA98, 0x12835B01, 0x243185BE, 0x550C7DC3,
    0x72BE5D74, 0x80DEB1FE, 0x9BDC06A7, 0xC19BF174,
    0xE49B69C1, 0xEFBE4786, 0x0FC19DC6, 0x240CA1CC,
    0x2DE92C6F, 0x4A7484AA, 0x5CB0A9DC, 0x76F988DA,
    0x983E5152, 0xA831C66D, 0xB00327C8, 0xBF597FC7,
    0xC6E00BF3, 0xD5A79147, 0x06CA6351, 0x14292967,
    0x27B70A85, 0x2E1B2138, 0x4D2C6DFC, 0x53380D13,
    0x650A7354, 0x766A0ABB, 0x81C2C92E, 0x92722C85,
    0xA2BFE8A1, 0xA81A664B, 0xC24B8B70, 0xC76C51A3,
    0xD192E819, 0xD6990624, 0xF40E3585, 0x106AA070,
    0x19A4C116, 0x1E376C08, 0x2748774C, 0x34B0BCB5,
    0x391C0CB3, 0x4ED8AA4A, 0x5B9CCA4F, 0x682E6FF3,
    0x748F82EE, 0x78A5636F, 0x84C87814, 0x8CC70208,
    0x90BEFFFA, 0xA4506CEB, 0xBEF9A3F7, 0xC67178F2
};

#if defined(__arm__) || defined(__aarch32__) || defined(__arm64__) || defined(__aarch64__) || defined(_M_ARM)
// ============== ARM64 begin =======================
// All the ARM servers supports SHA256 instructions
# if defined(__GNUC__)
#  include <stdint.h>
# endif
# if defined(__ARM_NEON) || defined(_MSC_VER) || defined(__GNUC__)
#  include <arm_neon.h>
# endif
/* GCC and LLVM Clang, but not Apple Clang */
# if defined(__GNUC__) && !defined(__apple_build_version__)
#  if defined(__ARM_ACLE) || defined(__ARM_FEATURE_CRYPTO)
#   include <arm_acle.h>
#  endif
# endif
void sha256_process(uint32_t state[8], const uint8_t data[], uint32_t length)
{
    uint32x4_t STATE0, STATE1, ABEF_SAVE, CDGH_SAVE;
    uint32x4_t MSG0, MSG1, MSG2, MSG3;
    uint32x4_t TMP0, TMP1, TMP2;

    /* Load state */
    STATE0 = vld1q_u32(&state[0]);
    STATE1 = vld1q_u32(&state[4]);

    while (length >= 64)
    {
        /* Save state */
        ABEF_SAVE = STATE0;
        CDGH_SAVE = STATE1;

        /* Load message */
        MSG0 = vld1q_u32((const uint32_t *)(data +  0));
        MSG1 = vld1q_u32((const uint32_t *)(data + 16));
        MSG2 = vld1q_u32((const uint32_t *)(data + 32));
        MSG3 = vld1q_u32((const uint32_t *)(data + 48));

        /* Reverse for little endian */
        MSG0 = vreinterpretq_u32_u8(vrev32q_u8(vreinterpretq_u8_u32(MSG0)));
        MSG1 = vreinterpretq_u32_u8(vrev32q_u8(vreinterpretq_u8_u32(MSG1)));
        MSG2 = vreinterpretq_u32_u8(vrev32q_u8(vreinterpretq_u8_u32(MSG2)));
        MSG3 = vreinterpretq_u32_u8(vrev32q_u8(vreinterpretq_u8_u32(MSG3)));

        TMP0 = vaddq_u32(MSG0, vld1q_u32(&K[0x00]));

        /* Rounds 0-3 */
        MSG0 = vsha256su0q_u32(MSG0, MSG1);
        TMP2 = STATE0;
        TMP1 = vaddq_u32(MSG1, vld1q_u32(&K[0x04]));
        STATE0 = vsha256hq_u32(STATE0, STATE1, TMP0);
        STATE1 = vsha256h2q_u32(STATE1, TMP2, TMP0);
        MSG0 = vsha256su1q_u32(MSG0, MSG2, MSG3);

        /* Rounds 4-7 */
        MSG1 = vsha256su0q_u32(MSG1, MSG2);
        TMP2 = STATE0;
        TMP0 = vaddq_u32(MSG2, vld1q_u32(&K[0x08]));
        STATE0 = vsha256hq_u32(STATE0, STATE1, TMP1);
        STATE1 = vsha256h2q_u32(STATE1, TMP2, TMP1);
        MSG1 = vsha256su1q_u32(MSG1, MSG3, MSG0);

        /* Rounds 8-11 */
        MSG2 = vsha256su0q_u32(MSG2, MSG3);
        TMP2 = STATE0;
        TMP1 = vaddq_u32(MSG3, vld1q_u32(&K[0x0c]));
        STATE0 = vsha256hq_u32(STATE0, STATE1, TMP0);
        STATE1 = vsha256h2q_u32(STATE1, TMP2, TMP0);
        MSG2 = vsha256su1q_u32(MSG2, MSG0, MSG1);

        /* Rounds 12-15 */
        MSG3 = vsha256su0q_u32(MSG3, MSG0);
        TMP2 = STATE0;
        TMP0 = vaddq_u32(MSG0, vld1q_u32(&K[0x10]));
        STATE0 = vsha256hq_u32(STATE0, STATE1, TMP1);
        STATE1 = vsha256h2q_u32(STATE1, TMP2, TMP1);
        MSG3 = vsha256su1q_u32(MSG3, MSG1, MSG2);

        /* Rounds 16-19 */
        MSG0 = vsha256su0q_u32(MSG0, MSG1);
        TMP2 = STATE0;
        TMP1 = vaddq_u32(MSG1, vld1q_u32(&K[0x14]));
        STATE0 = vsha256hq_u32(STATE0, STATE1, TMP0);
        STATE1 = vsha256h2q_u32(STATE1, TMP2, TMP0);
        MSG0 = vsha256su1q_u32(MSG0, MSG2, MSG3);

        /* Rounds 20-23 */
        MSG1 = vsha256su0q_u32(MSG1, MSG2);
        TMP2 = STATE0;
        TMP0 = vaddq_u32(MSG2, vld1q_u32(&K[0x18]));
        STATE0 = vsha256hq_u32(STATE0, STATE1, TMP1);
        STATE1 = vsha256h2q_u32(STATE1, TMP2, TMP1);
        MSG1 = vsha256su1q_u32(MSG1, MSG3, MSG0);

        /* Rounds 24-27 */
        MSG2 = vsha256su0q_u32(MSG2, MSG3);
        TMP2 = STATE0;
        TMP1 = vaddq_u32(MSG3, vld1q_u32(&K[0x1c]));
        STATE0 = vsha256hq_u32(STATE0, STATE1, TMP0);
        STATE1 = vsha256h2q_u32(STATE1, TMP2, TMP0);
        MSG2 = vsha256su1q_u32(MSG2, MSG0, MSG1);

        /* Rounds 28-31 */
        MSG3 = vsha256su0q_u32(MSG3, MSG0);
        TMP2 = STATE0;
        TMP0 = vaddq_u32(MSG0, vld1q_u32(&K[0x20]));
        STATE0 = vsha256hq_u32(STATE0, STATE1, TMP1);
        STATE1 = vsha256h2q_u32(STATE1, TMP2, TMP1);
        MSG3 = vsha256su1q_u32(MSG3, MSG1, MSG2);

        /* Rounds 32-35 */
        MSG0 = vsha256su0q_u32(MSG0, MSG1);
        TMP2 = STATE0;
        TMP1 = vaddq_u32(MSG1, vld1q_u32(&K[0x24]));
        STATE0 = vsha256hq_u32(STATE0, STATE1, TMP0);
        STATE1 = vsha256h2q_u32(STATE1, TMP2, TMP0);
        MSG0 = vsha256su1q_u32(MSG0, MSG2, MSG3);

        /* Rounds 36-39 */
        MSG1 = vsha256su0q_u32(MSG1, MSG2);
        TMP2 = STATE0;
        TMP0 = vaddq_u32(MSG2, vld1q_u32(&K[0x28]));
        STATE0 = vsha256hq_u32(STATE0, STATE1, TMP1);
        STATE1 = vsha256h2q_u32(STATE1, TMP2, TMP1);
        MSG1 = vsha256su1q_u32(MSG1, MSG3, MSG0);

        /* Rounds 40-43 */
        MSG2 = vsha256su0q_u32(MSG2, MSG3);
        TMP2 = STATE0;
        TMP1 = vaddq_u32(MSG3, vld1q_u32(&K[0x2c]));
        STATE0 = vsha256hq_u32(STATE0, STATE1, TMP0);
        STATE1 = vsha256h2q_u32(STATE1, TMP2, TMP0);
        MSG2 = vsha256su1q_u32(MSG2, MSG0, MSG1);

        /* Rounds 44-47 */
        MSG3 = vsha256su0q_u32(MSG3, MSG0);
        TMP2 = STATE0;
        TMP0 = vaddq_u32(MSG0, vld1q_u32(&K[0x30]));
        STATE0 = vsha256hq_u32(STATE0, STATE1, TMP1);
        STATE1 = vsha256h2q_u32(STATE1, TMP2, TMP1);
        MSG3 = vsha256su1q_u32(MSG3, MSG1, MSG2);

        /* Rounds 48-51 */
        TMP2 = STATE0;
        TMP1 = vaddq_u32(MSG1, vld1q_u32(&K[0x34]));
        STATE0 = vsha256hq_u32(STATE0, STATE1, TMP0);
        STATE1 = vsha256h2q_u32(STATE1, TMP2, TMP0);

        /* Rounds 52-55 */
        TMP2 = STATE0;
        TMP0 = vaddq_u32(MSG2, vld1q_u32(&K[0x38]));
        STATE0 = vsha256hq_u32(STATE0, STATE1, TMP1);
        STATE1 = vsha256h2q_u32(STATE1, TMP2, TMP1);

        /* Rounds 56-59 */
        TMP2 = STATE0;
        TMP1 = vaddq_u32(MSG3, vld1q_u32(&K[0x3c]));
        STATE0 = vsha256hq_u32(STATE0, STATE1, TMP0);
        STATE1 = vsha256h2q_u32(STATE1, TMP2, TMP0);

        /* Rounds 60-63 */
        TMP2 = STATE0;
        STATE0 = vsha256hq_u32(STATE0, STATE1, TMP1);
        STATE1 = vsha256h2q_u32(STATE1, TMP2, TMP1);

        /* Combine state */
        STATE0 = vaddq_u32(STATE0, ABEF_SAVE);
        STATE1 = vaddq_u32(STATE1, CDGH_SAVE);

        data += 64;
        length -= 64;
    }

    /* Save state */
    vst1q_u32(&state[0], STATE0);
    vst1q_u32(&state[4], STATE1);
}

// ============== ARM64 end =======================
#else
// ============== x86-64 begin =======================
/* Include the GCC super header */
#if defined(__GNUC__)
# include <stdint.h>
# include <x86intrin.h>
#endif

/* Microsoft supports Intel SHA ACLE extensions as of Visual Studio 2015 */
#if defined(_MSC_VER)
# include <immintrin.h>
# define WIN32_LEAN_AND_MEAN
# include <Windows.h>
#endif
#define ROTATE(x,y)  (((x)>>(y)) | ((x)<<(32-(y))))
#define Sigma0(x)    (ROTATE((x), 2) ^ ROTATE((x),13) ^ ROTATE((x),22))
#define Sigma1(x)    (ROTATE((x), 6) ^ ROTATE((x),11) ^ ROTATE((x),25))
#define sigma0(x)    (ROTATE((x), 7) ^ ROTATE((x),18) ^ ((x)>> 3))
#define sigma1(x)    (ROTATE((x),17) ^ ROTATE((x),19) ^ ((x)>>10))

#define Ch(x,y,z)    (((x) & (y)) ^ ((~(x)) & (z)))
#define Maj(x,y,z)   (((x) & (y)) ^ ((x) & (z)) ^ ((y) & (z)))

/* Avoid undefined behavior                    */
/* https://stackoverflow.com/q/29538935/608639 */
uint32_t B2U32(uint8_t val, uint8_t sh)
{
    return ((uint32_t)val) << sh;
}

void sha256_process_c(uint32_t state[8], const uint8_t data[], size_t length)
{
    uint32_t a, b, c, d, e, f, g, h, s0, s1, T1, T2;
    uint32_t X[16], i;

    size_t blocks = length / 64;
    while (blocks--)
    {
        a = state[0];
        b = state[1];
        c = state[2];
        d = state[3];
        e = state[4];
        f = state[5];
        g = state[6];
        h = state[7];

        for (i = 0; i < 16; i++)
        {
            X[i] = B2U32(data[0], 24) | B2U32(data[1], 16) | B2U32(data[2], 8) | B2U32(data[3], 0);
            data += 4;

            T1 = h;
            T1 += Sigma1(e);
            T1 += Ch(e, f, g);
            T1 += K[i];
            T1 += X[i];

            T2 = Sigma0(a);
            T2 += Maj(a, b, c);

            h = g;
            g = f;
            f = e;
            e = d + T1;
            d = c;
            c = b;
            b = a;
            a = T1 + T2;
        }

        for (; i < 64; i++)
        {
            s0 = X[(i + 1) & 0x0f];
            s0 = sigma0(s0);
            s1 = X[(i + 14) & 0x0f];
            s1 = sigma1(s1);

            T1 = X[i & 0xf] += s0 + s1 + X[(i + 9) & 0xf];
            T1 += h + Sigma1(e) + Ch(e, f, g) + K[i];
            T2 = Sigma0(a) + Maj(a, b, c);
            h = g;
            g = f;
            f = e;
            e = d + T1;
            d = c;
            c = b;
            b = a;
            a = T1 + T2;
        }

        state[0] += a;
        state[1] += b;
        state[2] += c;
        state[3] += d;
        state[4] += e;
        state[5] += f;
        state[6] += g;
        state[7] += h;
    }
}

/* Process multiple blocks. The caller is responsible for setting the initial */
/*  state, and the caller is responsible for padding the final block.        */
void sha256_process_asm(uint32_t state[8], const uint8_t data[], size_t length)
{
    __m128i STATE0, STATE1;
    __m128i MSG, TMP;
    __m128i MSG0, MSG1, MSG2, MSG3;
    __m128i ABEF_SAVE, CDGH_SAVE;
    const __m128i MASK = _mm_set_epi64x(0x0c0d0e0f08090a0bULL, 0x0405060700010203ULL);

    /* Load initial values */
    TMP = _mm_loadu_si128((const __m128i*) &state[0]);
    STATE1 = _mm_loadu_si128((const __m128i*) &state[4]);


    TMP = _mm_shuffle_epi32(TMP, 0xB1);          /* CDAB */
    STATE1 = _mm_shuffle_epi32(STATE1, 0x1B);    /* EFGH */
    STATE0 = _mm_alignr_epi8(TMP, STATE1, 8);    /* ABEF */
    STATE1 = _mm_blend_epi16(STATE1, TMP, 0xF0); /* CDGH */

    while (length >= 64)
    {
        /* Save current state */
        ABEF_SAVE = STATE0;
        CDGH_SAVE = STATE1;

        /* Rounds 0-3 */
        MSG = _mm_loadu_si128((const __m128i*) (data+0));
        MSG0 = _mm_shuffle_epi8(MSG, MASK);
        MSG = _mm_add_epi32(MSG0, _mm_set_epi64x(0xE9B5DBA5B5C0FBCFULL, 0x71374491428A2F98ULL));
        STATE1 = _mm_sha256rnds2_epu32(STATE1, STATE0, MSG);
        MSG = _mm_shuffle_epi32(MSG, 0x0E);
        STATE0 = _mm_sha256rnds2_epu32(STATE0, STATE1, MSG);

        /* Rounds 4-7 */
        MSG1 = _mm_loadu_si128((const __m128i*) (data+16));
        MSG1 = _mm_shuffle_epi8(MSG1, MASK);
        MSG = _mm_add_epi32(MSG1, _mm_set_epi64x(0xAB1C5ED5923F82A4ULL, 0x59F111F13956C25BULL));
        STATE1 = _mm_sha256rnds2_epu32(STATE1, STATE0, MSG);
        MSG = _mm_shuffle_epi32(MSG, 0x0E);
        STATE0 = _mm_sha256rnds2_epu32(STATE0, STATE1, MSG);
        MSG0 = _mm_sha256msg1_epu32(MSG0, MSG1);

        /* Rounds 8-11 */
        MSG2 = _mm_loadu_si128((const __m128i*) (data+32));
        MSG2 = _mm_shuffle_epi8(MSG2, MASK);
        MSG = _mm_add_epi32(MSG2, _mm_set_epi64x(0x550C7DC3243185BEULL, 0x12835B01D807AA98ULL));
        STATE1 = _mm_sha256rnds2_epu32(STATE1, STATE0, MSG);
        MSG = _mm_shuffle_epi32(MSG, 0x0E);
        STATE0 = _mm_sha256rnds2_epu32(STATE0, STATE1, MSG);
        MSG1 = _mm_sha256msg1_epu32(MSG1, MSG2);

        /* Rounds 12-15 */
        MSG3 = _mm_loadu_si128((const __m128i*) (data+48));
        MSG3 = _mm_shuffle_epi8(MSG3, MASK);
        MSG = _mm_add_epi32(MSG3, _mm_set_epi64x(0xC19BF1749BDC06A7ULL, 0x80DEB1FE72BE5D74ULL));
        STATE1 = _mm_sha256rnds2_epu32(STATE1, STATE0, MSG);
        TMP = _mm_alignr_epi8(MSG3, MSG2, 4);
        MSG0 = _mm_add_epi32(MSG0, TMP);
        MSG0 = _mm_sha256msg2_epu32(MSG0, MSG3);
        MSG = _mm_shuffle_epi32(MSG, 0x0E);
        STATE0 = _mm_sha256rnds2_epu32(STATE0, STATE1, MSG);
        MSG2 = _mm_sha256msg1_epu32(MSG2, MSG3);

        /* Rounds 16-19 */
        MSG = _mm_add_epi32(MSG0, _mm_set_epi64x(0x240CA1CC0FC19DC6ULL, 0xEFBE4786E49B69C1ULL));
        STATE1 = _mm_sha256rnds2_epu32(STATE1, STATE0, MSG);
        TMP = _mm_alignr_epi8(MSG0, MSG3, 4);
        MSG1 = _mm_add_epi32(MSG1, TMP);
        MSG1 = _mm_sha256msg2_epu32(MSG1, MSG0);
        MSG = _mm_shuffle_epi32(MSG, 0x0E);
        STATE0 = _mm_sha256rnds2_epu32(STATE0, STATE1, MSG);
        MSG3 = _mm_sha256msg1_epu32(MSG3, MSG0);

        /* Rounds 20-23 */
        MSG = _mm_add_epi32(MSG1, _mm_set_epi64x(0x76F988DA5CB0A9DCULL, 0x4A7484AA2DE92C6FULL));
        STATE1 = _mm_sha256rnds2_epu32(STATE1, STATE0, MSG);
        TMP = _mm_alignr_epi8(MSG1, MSG0, 4);
        MSG2 = _mm_add_epi32(MSG2, TMP);
        MSG2 = _mm_sha256msg2_epu32(MSG2, MSG1);
        MSG = _mm_shuffle_epi32(MSG, 0x0E);
        STATE0 = _mm_sha256rnds2_epu32(STATE0, STATE1, MSG);
        MSG0 = _mm_sha256msg1_epu32(MSG0, MSG1);

        /* Rounds 24-27 */
        MSG = _mm_add_epi32(MSG2, _mm_set_epi64x(0xBF597FC7B00327C8ULL, 0xA831C66D983E5152ULL));
        STATE1 = _mm_sha256rnds2_epu32(STATE1, STATE0, MSG);
        TMP = _mm_alignr_epi8(MSG2, MSG1, 4);
        MSG3 = _mm_add_epi32(MSG3, TMP);
        MSG3 = _mm_sha256msg2_epu32(MSG3, MSG2);
        MSG = _mm_shuffle_epi32(MSG, 0x0E);
        STATE0 = _mm_sha256rnds2_epu32(STATE0, STATE1, MSG);
        MSG1 = _mm_sha256msg1_epu32(MSG1, MSG2);

        /* Rounds 28-31 */
        MSG = _mm_add_epi32(MSG3, _mm_set_epi64x(0x1429296706CA6351ULL,  0xD5A79147C6E00BF3ULL));
        STATE1 = _mm_sha256rnds2_epu32(STATE1, STATE0, MSG);
        TMP = _mm_alignr_epi8(MSG3, MSG2, 4);
        MSG0 = _mm_add_epi32(MSG0, TMP);
        MSG0 = _mm_sha256msg2_epu32(MSG0, MSG3);
        MSG = _mm_shuffle_epi32(MSG, 0x0E);
        STATE0 = _mm_sha256rnds2_epu32(STATE0, STATE1, MSG);
        MSG2 = _mm_sha256msg1_epu32(MSG2, MSG3);

        /* Rounds 32-35 */
        MSG = _mm_add_epi32(MSG0, _mm_set_epi64x(0x53380D134D2C6DFCULL, 0x2E1B213827B70A85ULL));
        STATE1 = _mm_sha256rnds2_epu32(STATE1, STATE0, MSG);
        TMP = _mm_alignr_epi8(MSG0, MSG3, 4);
        MSG1 = _mm_add_epi32(MSG1, TMP);
        MSG1 = _mm_sha256msg2_epu32(MSG1, MSG0);
        MSG = _mm_shuffle_epi32(MSG, 0x0E);
        STATE0 = _mm_sha256rnds2_epu32(STATE0, STATE1, MSG);
        MSG3 = _mm_sha256msg1_epu32(MSG3, MSG0);

        /* Rounds 36-39 */
        MSG = _mm_add_epi32(MSG1, _mm_set_epi64x(0x92722C8581C2C92EULL, 0x766A0ABB650A7354ULL));
        STATE1 = _mm_sha256rnds2_epu32(STATE1, STATE0, MSG);
        TMP = _mm_alignr_epi8(MSG1, MSG0, 4);
        MSG2 = _mm_add_epi32(MSG2, TMP);
        MSG2 = _mm_sha256msg2_epu32(MSG2, MSG1);
        MSG = _mm_shuffle_epi32(MSG, 0x0E);
        STATE0 = _mm_sha256rnds2_epu32(STATE0, STATE1, MSG);
        MSG0 = _mm_sha256msg1_epu32(MSG0, MSG1);

        /* Rounds 40-43 */
        MSG = _mm_add_epi32(MSG2, _mm_set_epi64x(0xC76C51A3C24B8B70ULL, 0xA81A664BA2BFE8A1ULL));
        STATE1 = _mm_sha256rnds2_epu32(STATE1, STATE0, MSG);
        TMP = _mm_alignr_epi8(MSG2, MSG1, 4);
        MSG3 = _mm_add_epi32(MSG3, TMP);
        MSG3 = _mm_sha256msg2_epu32(MSG3, MSG2);
        MSG = _mm_shuffle_epi32(MSG, 0x0E);
        STATE0 = _mm_sha256rnds2_epu32(STATE0, STATE1, MSG);
        MSG1 = _mm_sha256msg1_epu32(MSG1, MSG2);

        /* Rounds 44-47 */
        MSG = _mm_add_epi32(MSG3, _mm_set_epi64x(0x106AA070F40E3585ULL, 0xD6990624D192E819ULL));
        STATE1 = _mm_sha256rnds2_epu32(STATE1, STATE0, MSG);
        TMP = _mm_alignr_epi8(MSG3, MSG2, 4);
        MSG0 = _mm_add_epi32(MSG0, TMP);
        MSG0 = _mm_sha256msg2_epu32(MSG0, MSG3);
        MSG = _mm_shuffle_epi32(MSG, 0x0E);
        STATE0 = _mm_sha256rnds2_epu32(STATE0, STATE1, MSG);
        MSG2 = _mm_sha256msg1_epu32(MSG2, MSG3);

        /* Rounds 48-51 */
        MSG = _mm_add_epi32(MSG0, _mm_set_epi64x(0x34B0BCB52748774CULL, 0x1E376C0819A4C116ULL));
        STATE1 = _mm_sha256rnds2_epu32(STATE1, STATE0, MSG);
        TMP = _mm_alignr_epi8(MSG0, MSG3, 4);
        MSG1 = _mm_add_epi32(MSG1, TMP);
        MSG1 = _mm_sha256msg2_epu32(MSG1, MSG0);
        MSG = _mm_shuffle_epi32(MSG, 0x0E);
        STATE0 = _mm_sha256rnds2_epu32(STATE0, STATE1, MSG);
        MSG3 = _mm_sha256msg1_epu32(MSG3, MSG0);

        /* Rounds 52-55 */
        MSG = _mm_add_epi32(MSG1, _mm_set_epi64x(0x682E6FF35B9CCA4FULL, 0x4ED8AA4A391C0CB3ULL));
        STATE1 = _mm_sha256rnds2_epu32(STATE1, STATE0, MSG);
        TMP = _mm_alignr_epi8(MSG1, MSG0, 4);
        MSG2 = _mm_add_epi32(MSG2, TMP);
        MSG2 = _mm_sha256msg2_epu32(MSG2, MSG1);
        MSG = _mm_shuffle_epi32(MSG, 0x0E);
        STATE0 = _mm_sha256rnds2_epu32(STATE0, STATE1, MSG);

        /* Rounds 56-59 */
        MSG = _mm_add_epi32(MSG2, _mm_set_epi64x(0x8CC7020884C87814ULL, 0x78A5636F748F82EEULL));
        STATE1 = _mm_sha256rnds2_epu32(STATE1, STATE0, MSG);
        TMP = _mm_alignr_epi8(MSG2, MSG1, 4);
        MSG3 = _mm_add_epi32(MSG3, TMP);
        MSG3 = _mm_sha256msg2_epu32(MSG3, MSG2);
        MSG = _mm_shuffle_epi32(MSG, 0x0E);
        STATE0 = _mm_sha256rnds2_epu32(STATE0, STATE1, MSG);

        /* Rounds 60-63 */
        MSG = _mm_add_epi32(MSG3, _mm_set_epi64x(0xC67178F2BEF9A3F7ULL, 0xA4506CEB90BEFFFAULL));
        STATE1 = _mm_sha256rnds2_epu32(STATE1, STATE0, MSG);
        MSG = _mm_shuffle_epi32(MSG, 0x0E);
        STATE0 = _mm_sha256rnds2_epu32(STATE0, STATE1, MSG);

        /* Combine state  */
        STATE0 = _mm_add_epi32(STATE0, ABEF_SAVE);
        STATE1 = _mm_add_epi32(STATE1, CDGH_SAVE);

        data += 64;
        length -= 64;
    }

    TMP = _mm_shuffle_epi32(STATE0, 0x1B);       /* FEBA */
    STATE1 = _mm_shuffle_epi32(STATE1, 0xB1);    /* DCHG */
    STATE0 = _mm_blend_epi16(TMP, STATE1, 0xF0); /* DCBA */
    STATE1 = _mm_alignr_epi8(STATE1, TMP, 8);    /* ABEF */

    /* Save state */
    _mm_storeu_si128((__m128i*) &state[0], STATE0);
    _mm_storeu_si128((__m128i*) &state[4], STATE1);
}

#if defined(__clang__) || defined(__GNUC__) || defined(__INTEL_COMPILER)

#include <cpuid.h>
int supports_sha_ni(void)
{
    unsigned int CPUInfo[4];
    __cpuid(0, CPUInfo[0], CPUInfo[1], CPUInfo[2], CPUInfo[3]);
    if (CPUInfo[0] < 7)
        return 0;

    __cpuid_count(7, 0, CPUInfo[0], CPUInfo[1], CPUInfo[2], CPUInfo[3]);
    return CPUInfo[1] & (1 << 29); /* SHA */
}

#else /* defined(__clang__) || defined(__GNUC__) */

int supports_sha_ni(void)
{
    unsigned int CPUInfo[4];
    __cpuid(CPUInfo, 0);  
    if (CPUInfo[0] < 7)
        return 0;

    __cpuidex(CPUInfo, 7, 0);
    return CPUInfo[1] & (1 << 29); /* Check SHA */
}

#endif /* defined(__clang__) || defined(__GNUC__) */

void sha256_process(uint32_t state[8], const uint8_t data[], size_t length) {
    static int has_sha_ni = -1;
    if(has_sha_ni == -1 ) {
        has_sha_ni = supports_sha_ni();
    }

    if(has_sha_ni) {
        sha256_process_asm(state, data, length);
	//printf("In sha256_process_asm length %zu\n", length);
    } else {
        sha256_process_c(state, data, length);
	//printf("In sha256_process_c length %zu\n", length);
    }
}
// ============== x86-64 end =======================
#endif

void sha256_init(SHA256_CTX *ctx)
{
	ctx->datalen = 0;
	ctx->bitlen = 0;
	ctx->state[0] = 0x6a09e667;
	ctx->state[1] = 0xbb67ae85;
	ctx->state[2] = 0x3c6ef372;
	ctx->state[3] = 0xa54ff53a;
	ctx->state[4] = 0x510e527f;
	ctx->state[5] = 0x9b05688c;
	ctx->state[6] = 0x1f83d9ab;
	ctx->state[7] = 0x5be0cd19;
}

void sha256_update(SHA256_CTX *ctx, const BYTE data[], size_t len)
{
	WORD i;

	size_t rounded = 64*(len/64);
	if(rounded != 0) {
		sha256_process(ctx->state, data, rounded);
	}

	ctx->bitlen = rounded*8;
	ctx->datalen = 0;
	for (i = rounded; i < len; ++i) {
		ctx->data[ctx->datalen] = data[i];
		ctx->datalen++;
	}
}

void sha256_final(SHA256_CTX *ctx, BYTE hash[])
{
	WORD i;

	i = ctx->datalen;

	// Pad whatever data is left in the buffer.
	if (ctx->datalen < 56) {
		ctx->data[i++] = 0x80;
		while (i < 56)
			ctx->data[i++] = 0x00;
	}
	else {
		ctx->data[i++] = 0x80;
		while (i < 64)
			ctx->data[i++] = 0x00;
		sha256_process(ctx->state, ctx->data, 64);
		memset(ctx->data, 0, 56);
	}

	// Append to the padding the total message's length in bits and transform.
	ctx->bitlen += ctx->datalen * 8;
	ctx->data[63] = ctx->bitlen;
	ctx->data[62] = ctx->bitlen >> 8;
	ctx->data[61] = ctx->bitlen >> 16;
	ctx->data[60] = ctx->bitlen >> 24;
	ctx->data[59] = ctx->bitlen >> 32;
	ctx->data[58] = ctx->bitlen >> 40;
	ctx->data[57] = ctx->bitlen >> 48;
	ctx->data[56] = ctx->bitlen >> 56;
	sha256_process(ctx->state, ctx->data, 64);

	// Since this implementation uses little endian byte ordering and SHA uses big endian,
	// reverse all the bytes when copying the final state to the output hash.
	for (i = 0; i < 4; ++i) {
		hash[i]      = (ctx->state[0] >> (24 - i * 8)) & 0x000000ff;
		hash[i + 4]  = (ctx->state[1] >> (24 - i * 8)) & 0x000000ff;
		hash[i + 8]  = (ctx->state[2] >> (24 - i * 8)) & 0x000000ff;
		hash[i + 12] = (ctx->state[3] >> (24 - i * 8)) & 0x000000ff;
		hash[i + 16] = (ctx->state[4] >> (24 - i * 8)) & 0x000000ff;
		hash[i + 20] = (ctx->state[5] >> (24 - i * 8)) & 0x000000ff;
		hash[i + 24] = (ctx->state[6] >> (24 - i * 8)) & 0x000000ff;
		hash[i + 28] = (ctx->state[7] >> (24 - i * 8)) & 0x000000ff;
	}
}