#ifndef PERMUTATIONS_H_
#define PERMUTATIONS_H_

#include "ascon.h"
#include "config.h"
#include "round.h"

static const uint64_t C[12] = {
    0xffffffffffffff0full, 0xffffffffffffff1eull, 0xffffffffffffff2dull,
    0xffffffffffffff3cull, 0xffffffffffffff4bull, 0xffffffffffffff5aull,
    0xffffffffffffff69ull, 0xffffffffffffff78ull, 0xffffffffffffff87ull,
    0xffffffffffffff96ull, 0xffffffffffffffa5ull, 0xffffffffffffffb4ull,
};

#define P12() \
  __asm__ __volatile__ ( \
      ".arm \n\t" \
      ".fpu neon \n\t" \
      "vldm %[s], {d0-d4} \n\t" \
      "vmvn d2, d2 \n\t" \
      ROUND(0) \
      ROUND(8) \
      ROUND(16) \
      ROUND(24) \
      ROUND(32) \
      ROUND(40) \
      ROUND(48) \
      ROUND(56) \
      ROUND(64) \
      ROUND(72) \
      ROUND(80) \
      ROUND(88) \
      "vmvn d2, d2 \n\t" \
      "vstm %[s], {d0-d4} \n\t" \
      :: [s] "r" (&s), [C] "r" (C) \
      : "d0", "d1", "d2", "d3", "d4", \
        "d10", "d11", "d12", "d13", "d14", \
        "d20", "d21", "d22", "d23", "d24", \
        "d31", "memory")

#define P6() \
  __asm__ __volatile__ ( \
      ".arm \n\t" \
      ".fpu neon \n\t" \
      "vldm %[s], {d0-d4} \n\t" \
      "vmvn d2, d2 \n\t" \
      ROUND(48) \
      ROUND(56) \
      ROUND(64) \
      ROUND(72) \
      ROUND(80) \
      ROUND(88) \
      "vmvn d2, d2 \n\t" \
      "vstm %[s], {d0-d4} \n\t" \
      :: [s] "r" (&s), [C] "r" (C) \
      : "d0", "d1", "d2", "d3", "d4", \
        "d10", "d11", "d12", "d13", "d14", \
        "d20", "d21", "d22", "d23", "d24", \
        "d31", "memory")

#define AD()                                     \
  do {                                           \
    uint32_t adlen_hi = (uint32_t)(adlen >> 32); \
    uint32_t adlen_lo = (uint32_t)adlen;         \
    __asm__ __volatile__ ( \
        ".arm \n\t" \
        ".fpu neon \n\t" \
        "cmp %[adlen_hi], #0 \n\t" \
        "cmpeq %[adlen_lo], #7 \n\t" \
        "bls .LAD1 \n\t" \
        "vldm %[s], {d0-d4} \n\t" \
        ".LAD0: \n\t" \
        "vldm %[ad]!, {d16} \n\t" \
        "vrev64.8 d16, d16 \n\t" \
        "veor d0, d0, d16 \n\t" \
        "vmvn d2, d2 \n\t" \
        ROUND(48) \
        ROUND(56) \
        ROUND(64) \
        ROUND(72) \
        ROUND(80) \
        ROUND(88) \
        "vmvn d2, d2 \n\t" \
        "sub %[adlen_lo], %[adlen_lo], #8 \n\t" \
        "sbc %[adlen_hi], %[adlen_hi], #0 \n\t" \
        "cmp %[adlen_hi], #0 \n\t" \
        "cmpeq %[adlen_lo], #7 \n\t" \
        "bhi .LAD0 \n\t" \
        "vstm %[s], {d0-d4} \n\t" \
        ".LAD1: \n\t" \
        : [adlen_hi] "+r" (adlen_hi), [adlen_lo] "+r" (adlen_lo), \
          [ad] "+r" (ad) \
        : [s] "r" (&s), [C] "r" (C) \
        : "d0", "d1", "d2", "d3", "d4", \
          "d10", "d11", "d12", "d13", "d14", "d16", \
          "d20", "d21", "d22", "d23", "d24", \
          "d31", "memory");                     \
    adlen = (uint64_t)adlen_hi << 32 | adlen_lo; \
  } while (0)

#define PT()                                   \
  do {                                         \
    uint32_t mlen_hi = (uint32_t)(mlen >> 32); \
    uint32_t mlen_lo = (uint32_t)mlen;         \
    __asm__ __volatile__ ( \
        ".arm \n\t" \
        ".fpu neon \n\t" \
        "cmp %[mlen_hi], #0 \n\t" \
        "cmpeq %[mlen_lo], #7 \n\t" \
        "bls .LPT1 \n\t" \
        "vldm %[s], {d0-d4} \n\t" \
        ".LPT0: \n\t" \
        "vldm %[m]!, {d16} \n\t" \
        "vrev64.8 d16, d16 \n\t" \
        "veor d0, d0, d16 \n\t" \
        "vrev64.8 d26, d0 \n\t" \
        "vstm %[c]!, {d26} \n\t" \
        "vmvn d2, d2 \n\t" \
        ROUND(48) \
        ROUND(56) \
        ROUND(64) \
        ROUND(72) \
        ROUND(80) \
        ROUND(88) \
        "vmvn d2, d2 \n\t" \
        "sub %[mlen_lo], %[mlen_lo], #8 \n\t" \
        "sbc %[mlen_hi], %[mlen_hi], #0 \n\t" \
        "cmp %[mlen_hi], #0 \n\t" \
        "cmpeq %[mlen_lo], #7 \n\t" \
        "bhi .LPT0 \n\t" \
        "vstm %[s], {d0-d4} \n\t" \
        ".LPT1: \n\t" \
        : [mlen_hi] "+r" (mlen_hi), [mlen_lo] "+r" (mlen_lo), \
          [m] "+r" (m), [c] "+r" (c) \
        : [s] "r" (&s), [C] "r" (C) \
        : "d0", "d1", "d2", "d3", "d4", \
          "d10", "d11", "d12", "d13", "d14", "d16", \
          "d20", "d21", "d22", "d23", "d24", "d26", \
          "d31", "memory");                   \
    mlen = (uint64_t)mlen_hi << 32 | mlen_lo;  \
  } while (0)

#define CT()                                   \
  do {                                         \
    uint32_t clen_hi = (uint32_t)(clen >> 32); \
    uint32_t clen_lo = (uint32_t)clen;         \
    __asm__ __volatile__ ( \
        ".arm \n\t" \
        ".fpu neon \n\t" \
        "cmp %[clen_hi], #0 \n\t" \
        "cmpeq %[clen_lo], #7 \n\t" \
        "bls .LCT1 \n\t" \
        "vldm %[s], {d0-d4} \n\t" \
        ".LCT0: \n\t" \
        "vldm %[c]!, {d26} \n\t" \
        "vrev64.8 d16, d0 \n\t" \
        "veor d16, d16, d26 \n\t" \
        "vrev64.8 d0, d26 \n\t" \
        "vstm %[m]!, {d16} \n\t" \
        "vmvn d2, d2 \n\t" \
        ROUND(48) \
        ROUND(56) \
        ROUND(64) \
        ROUND(72) \
        ROUND(80) \
        ROUND(88) \
        "vmvn d2, d2 \n\t" \
        "sub %[clen_lo], %[clen_lo], #8 \n\t" \
        "sbc %[clen_hi], %[clen_hi], #0 \n\t" \
        "cmp %[clen_hi], #0 \n\t" \
        "cmpeq %[clen_lo], #7 \n\t" \
        "bhi .LCT0 \n\t" \
        "vstm %[s], {d0-d4} \n\t" \
        ".LCT1: \n\t" \
        : [clen_hi] "+r" (clen_hi), [clen_lo] "+r" (clen_lo), \
          [m] "+r" (m), [c] "+r" (c) \
        : [s] "r" (&s), [C] "r" (C) \
        : "d0", "d1", "d2", "d3", "d4", \
          "d10", "d11", "d12", "d13", "d14", "d16", \
          "d20", "d21", "d22", "d23", "d24", "d26", \
          "d31", "memory");                   \
    clen = (uint64_t)clen_hi << 32 | clen_lo;  \
  } while (0)

#endif /* PERMUTATIONS_H_ */