/*
 * Copyright (c) Facebook, Inc. and its affiliates.
 * All rights reserved.
 *
 * This source code is licensed under the BSD-style license found in the
 * LICENSE file in the root directory of this source tree.
 */

#pragma once

#include <limits.h>
#include <stdint.h>
#include <stdbool.h>
#include <assert.h>

#include <fp16/bitcasts.h>

#if defined(__clang__)
  #if __clang_major__ == 3 && __clang_minor__ >= 7 || __clang_major__ > 3
    #define QNNP_IGNORE_SHIFT_BASE_UB __attribute__((__no_sanitize__("shift-base")))
  #else
    #define QNNP_IGNORE_SHIFT_BASE_UB
  #endif
#elif defined(__GNUC__)
  #if __GNUC__ >= 8
    #define QNNP_IGNORE_SHIFT_BASE_UB __attribute__((__no_sanitize__("shift-base")))
  #elif __GNUC__ == 4 && __GNUC_MINOR__ >= 9 || __GNUC__ > 4
    /* 4.9 <= gcc < 8 support ubsan, but doesn't support no_sanitize attribute */
    #define QNNP_IGNORE_SHIFT_BASE_UB
    #ifndef QNNP_USE_SHIFT_BASE_UB_WORKAROUND
      #define QNNP_USE_SHIFT_BASE_UB_WORKAROUND 1
    #endif
  #else
    #define QNNP_IGNORE_SHIFT_BASE_UB
  #endif
#else
  #define QNNP_IGNORE_SHIFT_BASE_UB
#endif

QNNP_IGNORE_SHIFT_BASE_UB
inline static int32_t asr_s32(int32_t x, uint32_t n) {
  #ifdef QNNP_USE_SHIFT_BASE_UB_WORKAROUND
    #if defined(__x86_64__) || defined(__aarch64__)
      return (int32_t) ((uint64_t) (int64_t) x >> n);
    #else
      return x >= 0 ? x >> n : ~(~x >> n);
    #endif
  #else
    return x >> n;
  #endif
}

QNNP_IGNORE_SHIFT_BASE_UB
inline static int64_t asr_s64(int64_t x, uint32_t n) {
  #ifdef QNNP_USE_SHIFT_BASE_UB_WORKAROUND
    return x >= 0 ? x >> n : ~(~x >> n);
  #else
    return x >> n;
  #endif
}

inline static uint8_t scalar_requantize_precise(
  int32_t value,
  float scale,
  uint8_t zero_point,
  uint8_t qmin,
  uint8_t qmax)
{
  assert(scale < 1.0f);
  assert(scale >= 0x1.0p-32f);

  const uint32_t scale_bits = fp32_to_bits(scale);
  const uint32_t multiplier = (scale_bits & UINT32_C(0x007FFFFF)) | UINT32_C(0x00800000);
  const uint32_t shift = 127 + 23 - (scale_bits >> 23);
  assert(shift >= 24);
  assert(shift < 56);

  /*
   * Compute absolute value of input as unsigned 32-bit int.
   * All further computations will work with unsigned values to avoid undefined behaviour on signed operations.
   */
  const uint32_t abs_value = (value >= 0) ? (uint32_t) value : -(uint32_t) value;

  /* Compute full 64-bit product of 32-bit factors */
  const uint64_t product = (uint64_t) abs_value * (uint64_t) multiplier;

  /*
   * Shift the full 64-bit product right with rounding.
   * Rounding is performed towards closest integer, with midpoints rounded up (same as away from zero).
   */
  const uint64_t rounding = UINT64_C(1) << (shift - 1);
  const uint32_t abs_scaled_value = (uint32_t) ((product + rounding) >> shift);

  /*
   * Copy the sign of input to scaled absolute input value.
   */
  const int32_t scaled_value = (int32_t) (value >= 0 ? abs_scaled_value : -abs_scaled_value);

  /* Clamp scaled value with zero point between smin and smax */
  int32_t clamped_value = scaled_value;
  const int32_t smin = (int32_t) (uint32_t) qmin - (int32_t) (uint32_t) zero_point;
  if (clamped_value < smin) {
    clamped_value = smin;
  }
  const int32_t smax = (int32_t) (uint32_t) qmax - (int32_t) (uint32_t) zero_point;
  if (clamped_value > smax) {
    clamped_value = smax;
  }

  /* Add zero point to clamped value */
  const int32_t biased_value = clamped_value + (int32_t) (uint32_t) zero_point;

  return biased_value;
}
