/* SPDX-License-Identifier: BSD-2-Clause */
/*
 * H.264 chroma DC 2x2 Hadamard pre-pass (public, in-tree CPU).
 *
 * The 4 DC coefficients of an MB's chroma 4x4 AC blocks go through
 * this 2x2 Hadamard before quant-scaling and re-injection into the
 * AC blocks' [0,0] coefficient.  Algorithm per H.264 §8.5.11.1.
 *
 * Pure CPU primitive — there's no substrate-dispatch wrapper because
 * the work is 4 adds + 4 subs.  Callers compose with QP-dependent
 * scaling themselves (the scale shape varies by slice/PPS chroma_qp
 * offset context and shouldn't be baked into the kernel).
 *
 * Bit-exact validated against tests/h264_chroma_dc_hadamard_ref.c
 * (7-case spec-derived test suite including the H·H = 4·I algebraic
 * invariant; see PR #23).  Same algorithm; this is the public
 * src-tree copy.
 */
#include "daedalus.h"

#include <stdint.h>

void daedalus_h264_chroma_dc_hadamard_2x2(int16_t c[4])
{
    int t0 = c[0] + c[1];
    int t1 = c[0] - c[1];
    int t2 = c[2] + c[3];
    int t3 = c[2] - c[3];

    c[0] = (int16_t)(t0 + t2);   /* f[0,0] = sum of all 4   */
    c[1] = (int16_t)(t1 + t3);   /* f[0,1] = col-difference */
    c[2] = (int16_t)(t0 - t2);   /* f[1,0] = row-difference */
    c[3] = (int16_t)(t1 - t3);   /* f[1,1] = anti-diagonal  */
}