/****************************************************************************** * * Copyright (C) 2015 The Android Open Source Project * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at: * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. * ***************************************************************************** * Originally developed and contributed by Ittiam Systems Pvt. Ltd, Bangalore */ #include <string.h> #include "iv_datatypedef.h" #include "iv.h" #include "impeg2_buf_mgr.h" #include "impeg2_disp_mgr.h" #include "impeg2_defs.h" #include "impeg2_platform_macros.h" #include "impeg2_inter_pred.h" #include "impeg2_idct.h" #include "impeg2_globals.h" #include "impeg2_mem_func.h" #include "impeg2_format_conv.h" #include "impeg2_macros.h" #include "ivd.h" #include "impeg2d.h" #include "impeg2d_bitstream.h" #include "impeg2d_structs.h" #include "impeg2d_vld_tables.h" #include "impeg2d_vld.h" #include "impeg2d_pic_proc.h" #include "impeg2d_debug.h" /******************************************************************************* * Function name : impeg2d_dec_vld_symbol * * Description : Performs decoding of VLD symbol. It performs decoding by * processing 1 bit at a time * * Arguments : * stream : Bitstream * ai2_code_table : Table used for decoding * maxLen : Maximum Length of the decoded symbol in bits * * Value Returned: Decoded symbol *******************************************************************************/ WORD16 impeg2d_dec_vld_symbol(stream_t *ps_stream,const WORD16 ai2_code_table[][2], UWORD16 u2_max_len) { UWORD16 u2_data; WORD16 u2_end = 0; UWORD16 u2_org_max_len = u2_max_len; UWORD16 u2_i_bit; /* Get the maximum number of bits needed to decode a symbol */ u2_data = impeg2d_bit_stream_nxt(ps_stream,u2_max_len); do { u2_max_len--; /* Read one bit at a time from the variable to decode the huffman code */ u2_i_bit = (UWORD8)((u2_data >> u2_max_len) & 0x1); /* Get the next node pointer or the symbol from the tree */ u2_end = ai2_code_table[u2_end][u2_i_bit]; }while(u2_end > 0); /* Flush the appropriate number of bits from the ps_stream */ impeg2d_bit_stream_flush(ps_stream,(UWORD8)(u2_org_max_len - u2_max_len)); return(u2_end); } /******************************************************************************* * Function name : impeg2d_fast_dec_vld_symbol * * Description : Performs decoding of VLD symbol. It performs decoding by * processing n bits at a time * * Arguments : * stream : Bitstream * ai2_code_table : Code table containing huffman value * indexTable : Index table containing index * maxLen : Maximum Length of the decoded symbol in bits * * Value Returned: Decoded symbol *******************************************************************************/ WORD16 impeg2d_fast_dec_vld_symbol(stream_t *ps_stream, const WORD16 ai2_code_table[][2], const UWORD16 au2_indexTable[][2], UWORD16 u2_max_len) { UWORD16 u2_cur_code; UWORD16 u2_num_bits; UWORD16 u2_vld_offset; UWORD16 u2_start_len; WORD16 u2_value; UWORD16 u2_len; UWORD16 u2_huffCode; u2_start_len = au2_indexTable[0][0]; u2_vld_offset = 0; u2_huffCode = impeg2d_bit_stream_nxt(ps_stream,u2_max_len); do { u2_cur_code = u2_huffCode >> (u2_max_len - u2_start_len); u2_num_bits = ai2_code_table[u2_cur_code + u2_vld_offset][0]; if(u2_num_bits == 0) { u2_huffCode &= ((1 << (u2_max_len - u2_start_len)) - 1); u2_max_len -= u2_start_len; u2_start_len = au2_indexTable[ai2_code_table[u2_cur_code + u2_vld_offset][1]][0]; u2_vld_offset = au2_indexTable[ai2_code_table[u2_cur_code + u2_vld_offset][1]][1]; } else { u2_value = ai2_code_table[u2_cur_code + u2_vld_offset][1]; u2_len = u2_num_bits; } }while(u2_num_bits == 0); impeg2d_bit_stream_flush(ps_stream,u2_len); return(u2_value); } /****************************************************************************** * * Function Name : impeg2d_dec_ac_coeff_zero * * Description : Decodes using Table B.14 * * Arguments : Pointer to VideoObjectLayerStructure * * Values Returned : Decoded value * * Revision History: * * 28 02 2002 AR Creation *******************************************************************************/ UWORD16 impeg2d_dec_ac_coeff_zero(stream_t *ps_stream, UWORD16* pu2_sym_len, UWORD16* pu2_sym_val) { UWORD16 u2_offset,u2_decoded_value; UWORD8 u1_shift; UWORD32 u4_bits_read; u4_bits_read = (UWORD16)impeg2d_bit_stream_nxt(ps_stream,MPEG2_AC_COEFF_MAX_LEN); if ((UWORD16)u4_bits_read >= 0x0800) { u2_offset = (UWORD16)u4_bits_read >> 11; } else if ((UWORD16)u4_bits_read >= 0x40) { u2_offset = 31 + ((UWORD16)u4_bits_read >> 6); } else if ((UWORD16)u4_bits_read >= 0x20) { u2_offset = 64; } else { u2_offset = 63; u4_bits_read = (UWORD16)u4_bits_read - 0x10; } /*----------------------------------------------------------------------- * The table gOffset contains both the offset for the group to which the * Vld code belongs in the Ac Coeff Table and the no of bits with which * the BitsRead should be shifted *-----------------------------------------------------------------------*/ u2_offset = gau2_impeg2d_offset_zero[u2_offset]; u1_shift = u2_offset & 0xF; /*----------------------------------------------------------------------- * Depending upon the vld code, we index exactly to that particular * Vld codes value in the Ac Coeff Table. * (Offset >> 4) gives the offset for the group in the AcCoeffTable. * (BitsRead >> shift) gives the offset within its group *-----------------------------------------------------------------------*/ u2_offset = (u2_offset >> 4) + ((UWORD16)u4_bits_read >> u1_shift); /*----------------------------------------------------------------------- * DecodedValue has the Run, Level and the number of bits used by Vld code *-----------------------------------------------------------------------*/ u2_decoded_value = gau2_impeg2d_dct_coeff_zero[u2_offset]; if(u2_decoded_value == END_OF_BLOCK) { *pu2_sym_len = 2; *pu2_sym_val = EOB_CODE_VALUE; } else if(u2_decoded_value == ESCAPE_CODE) { *pu2_sym_len = u2_decoded_value & 0x1F; *pu2_sym_val = ESC_CODE_VALUE; } else { *pu2_sym_len = u2_decoded_value & 0x1F; *pu2_sym_val = u2_decoded_value >> 5; } return(u2_decoded_value); } /****************************************************************************** * * Function Name : impeg2d_dec_ac_coeff_one * * Description : Decodes using Table B.15 * * Arguments : Pointer to VideoObjectLayerStructure * * Values Returned : Decoded value * * Revision History: * * 28 02 2002 AR Creation *******************************************************************************/ UWORD16 impeg2d_dec_ac_coeff_one(stream_t *ps_stream, UWORD16* pu2_sym_len, UWORD16* pu2_sym_val) { UWORD16 u2_offset, u2_decoded_value; UWORD8 u1_shift; UWORD32 u4_bits_read; u4_bits_read = (UWORD16)impeg2d_bit_stream_nxt(ps_stream,MPEG2_AC_COEFF_MAX_LEN); if ((UWORD16)u4_bits_read >= 0x8000) { /* If the MSB of the vld code is 1 */ if (((UWORD16)u4_bits_read >> 12) == 0xF) u2_offset = ((UWORD16)u4_bits_read >> 8) & 0xF; else u2_offset = (UWORD16)u4_bits_read >> 11; u2_offset += gau2_impeg2d_offset_one[0]; } else if ((UWORD16)u4_bits_read >= 0x400) { u2_offset =(UWORD16) u4_bits_read >> 10; u2_offset = gau2_impeg2d_offset_one[u2_offset]; u1_shift = u2_offset & 0xF; u2_offset = (u2_offset >> 4) + ((UWORD16)u4_bits_read >> u1_shift); } else if ((UWORD16)u4_bits_read >= 0x20) { u2_offset = ((UWORD16)u4_bits_read >> 5) + 31; u2_offset = gau2_impeg2d_offset_one[u2_offset]; u1_shift = u2_offset & 0xF; u2_offset = (u2_offset >> 4) + ((UWORD16)u4_bits_read >> u1_shift); } else { u2_offset = gau2_impeg2d_offset_one[63] + ((UWORD16)u4_bits_read & 0xF); } /*----------------------------------------------------------------------- * DecodedValue has the Run, Level and the number of bits used by Vld code *-----------------------------------------------------------------------*/ u2_decoded_value = gau2_impeg2d_dct_coeff_one[u2_offset]; if(u2_decoded_value == END_OF_BLOCK) { *pu2_sym_len = 4; *pu2_sym_val = EOB_CODE_VALUE; } else if(u2_decoded_value == ESCAPE_CODE) { *pu2_sym_len = u2_decoded_value & 0x1F; *pu2_sym_val = ESC_CODE_VALUE; } else { *pu2_sym_len = u2_decoded_value & 0x1F; *pu2_sym_val = u2_decoded_value >> 5; } return(u2_decoded_value); } /****************************************************************************** * * Function Name : impeg2d_vld_inv_quant_mpeg1 * * Description : Performs VLD operation for MPEG1/2 * * Arguments : * state : VLCD state parameter * regs : Registers of VLCD * * Values Returned : None ******************************************************************************/ IMPEG2D_ERROR_CODES_T impeg2d_vld_inv_quant_mpeg1( void *pv_dec, /* Decoder State */ WORD16 *pi2_out_addr, /*!< Address where decoded symbols will be stored */ const UWORD8 *pu1_scan, /*!< Scan table to be used */ UWORD16 u2_intra_flag, /*!< Intra Macroblock or not */ UWORD16 u2_colr_comp, /*!< 0 - Luma,1 - U comp, 2 - V comp */ UWORD16 u2_d_picture /*!< D Picture or not */ ) { UWORD8 *pu1_weighting_matrix; dec_state_t *ps_dec = (dec_state_t *) pv_dec; IMPEG2D_ERROR_CODES_T e_error = (IMPEG2D_ERROR_CODES_T)IVD_ERROR_NONE; WORD16 pi2_coeffs[NUM_COEFFS]; UWORD8 pu1_pos[NUM_COEFFS]; WORD32 i4_num_coeffs; /* Perform VLD on the stream to get the coefficients and their positions */ e_error = impeg2d_vld_decode(ps_dec, pi2_coeffs, pu1_scan, pu1_pos, u2_intra_flag, u2_colr_comp, u2_d_picture, ps_dec->u2_intra_vlc_format, ps_dec->u2_is_mpeg2, &i4_num_coeffs); if ((IMPEG2D_ERROR_CODES_T)IVD_ERROR_NONE != e_error) { return e_error; } /* For YUV420 format,Select the weighting matrix according to Table 7.5 */ pu1_weighting_matrix = (u2_intra_flag == 1) ? ps_dec->au1_intra_quant_matrix: ps_dec->au1_inter_quant_matrix; IMPEG2D_IQNT_INP_STATISTICS(pi2_out_addr, ps_dec->u4_non_zero_cols, ps_dec->u4_non_zero_rows); /* Inverse Quantize the Output of VLD */ PROFILE_DISABLE_INVQUANT_IF0 { /* Clear output matrix */ PROFILE_DISABLE_MEMSET_RESBUF_IF0 if (1 != (ps_dec->u4_non_zero_cols | ps_dec->u4_non_zero_rows)) { ps_dec->pf_memset_16bit_8x8_linear_block (pi2_out_addr); } impeg2d_inv_quant_mpeg1(pi2_out_addr, pu1_weighting_matrix, ps_dec->u1_quant_scale, u2_intra_flag, i4_num_coeffs, pi2_coeffs, pu1_pos, pu1_scan, &ps_dec->u2_def_dc_pred[u2_colr_comp], ps_dec->u2_intra_dc_precision); if (0 != pi2_out_addr[0]) { /* The first coeff might've become non-zero due to intra_dc_decision * value. So, check here after inverse quantization. */ ps_dec->u4_non_zero_cols |= 0x1; ps_dec->u4_non_zero_rows |= 0x1; } } return e_error; } /****************************************************************************** * * Function Name : impeg2d_vld_inv_quant_mpeg2 * * Description : Performs VLD operation for MPEG1/2 * * Arguments : * state : VLCD state parameter * regs : Registers of VLCD * * Values Returned : None ******************************************************************************/ IMPEG2D_ERROR_CODES_T impeg2d_vld_inv_quant_mpeg2( void *pv_dec, /* Decoder State */ WORD16 *pi2_out_addr, /*!< Address where decoded symbols will be stored */ const UWORD8 *pu1_scan, /*!< Scan table to be used */ UWORD16 u2_intra_flag, /*!< Intra Macroblock or not */ UWORD16 u2_colr_comp, /*!< 0 - Luma,1 - U comp, 2 - V comp */ UWORD16 u2_d_picture /*!< D Picture or not */ ) { UWORD8 *pu1_weighting_matrix; WORD32 i4_sum; dec_state_t *ps_dec = (dec_state_t *)pv_dec; IMPEG2D_ERROR_CODES_T e_error = (IMPEG2D_ERROR_CODES_T)IVD_ERROR_NONE; WORD16 pi2_coeffs[NUM_COEFFS]; UWORD8 pi4_pos[NUM_COEFFS]; WORD32 i4_num_coeffs; /* Perform VLD on the stream to get the coefficients and their positions */ e_error = impeg2d_vld_decode(ps_dec, pi2_coeffs, pu1_scan, pi4_pos, u2_intra_flag, u2_colr_comp, u2_d_picture, ps_dec->u2_intra_vlc_format, ps_dec->u2_is_mpeg2, &i4_num_coeffs); if ((IMPEG2D_ERROR_CODES_T)IVD_ERROR_NONE != e_error) { return e_error; } /* For YUV420 format,Select the weighting matrix according to Table 7.5 */ pu1_weighting_matrix = (u2_intra_flag == 1) ? ps_dec->au1_intra_quant_matrix: ps_dec->au1_inter_quant_matrix; /*mismatch control for mpeg2*/ /* Check if the block has only one non-zero coeff which is DC */ ps_dec->i4_last_value_one = 0; IMPEG2D_IQNT_INP_STATISTICS(pi2_out_addr, ps_dec->u4_non_zero_cols, ps_dec->u4_non_zero_rows); /* Inverse Quantize the Output of VLD */ PROFILE_DISABLE_INVQUANT_IF0 { /* Clear output matrix */ PROFILE_DISABLE_MEMSET_RESBUF_IF0 if (1 != (ps_dec->u4_non_zero_cols | ps_dec->u4_non_zero_rows)) { ps_dec->pf_memset_16bit_8x8_linear_block (pi2_out_addr); } i4_sum = impeg2d_inv_quant_mpeg2(pi2_out_addr, pu1_weighting_matrix, ps_dec->u1_quant_scale, u2_intra_flag, i4_num_coeffs, pi2_coeffs, pi4_pos, pu1_scan, &ps_dec->u2_def_dc_pred[u2_colr_comp], ps_dec->u2_intra_dc_precision); if (0 != pi2_out_addr[0]) { /* The first coeff might've become non-zero due to intra_dc_decision * value. So, check here after inverse quantization. */ ps_dec->u4_non_zero_cols |= 0x1; ps_dec->u4_non_zero_rows |= 0x1; } if (1 == (ps_dec->u4_non_zero_cols | ps_dec->u4_non_zero_rows)) { ps_dec->i4_last_value_one = 1 - (pi2_out_addr[0] & 1); } else { /*toggle last bit if sum is even ,else retain it as it is*/ pi2_out_addr[63] ^= (i4_sum & 1); if (0 != pi2_out_addr[63]) { ps_dec->u4_non_zero_cols |= 0x80; ps_dec->u4_non_zero_rows |= 0x80; } } } return e_error; } /****************************************************************************** * * Function Name : impeg2d_vld_decode * * Description : Performs VLD operation for MPEG1/2 * * Arguments : * state : VLCD state parameter * regs : Registers of VLCD * * Values Returned : None ******************************************************************************/ IMPEG2D_ERROR_CODES_T impeg2d_vld_decode( dec_state_t *ps_dec, WORD16 *pi2_outAddr, /*!< Address where decoded symbols will be stored */ const UWORD8 *pu1_scan, /*!< Scan table to be used */ UWORD8 *pu1_pos, /*!< Scan table to be used */ UWORD16 u2_intra_flag, /*!< Intra Macroblock or not */ UWORD16 u2_chroma_flag, /*!< Chroma Block or not */ UWORD16 u2_d_picture, /*!< D Picture or not */ UWORD16 u2_intra_vlc_format, /*!< Intra VLC format */ UWORD16 u2_mpeg2, /*!< MPEG-2 or not */ WORD32 *pi4_num_coeffs /*!< Returns the number of coeffs in block */ ) { UWORD32 u4_sym_len; UWORD32 u4_decoded_value; WORD32 i4_level_first_byte; WORD32 i4_level; UWORD32 u4_run, u4_numCoeffs; UWORD32 u4_buf; UWORD32 u4_buf_nxt; UWORD32 u4_offset; UWORD32 *pu4_buf_aligned; UWORD32 u4_bits; stream_t *ps_stream = &ps_dec->s_bit_stream; WORD32 u4_pos; UWORD32 u4_nz_cols; UWORD32 u4_nz_rows; *pi4_num_coeffs = 0; ps_dec->u4_non_zero_cols = 0; ps_dec->u4_non_zero_rows = 0; u4_nz_cols = ps_dec->u4_non_zero_cols; u4_nz_rows = ps_dec->u4_non_zero_rows; GET_TEMP_STREAM_DATA(u4_buf,u4_buf_nxt,u4_offset,pu4_buf_aligned,ps_stream) /**************************************************************************/ /* Decode the DC coefficient in case of Intra block */ /**************************************************************************/ if(u2_intra_flag) { WORD32 dc_size; WORD32 dc_diff; WORD32 maxLen; WORD32 idx; maxLen = MPEG2_DCT_DC_SIZE_LEN; idx = 0; if(u2_chroma_flag != 0) { maxLen += 1; idx++; } { WORD16 end = 0; UWORD32 maxLen_tmp = maxLen; UWORD16 m_iBit; /* Get the maximum number of bits needed to decode a symbol */ IBITS_NXT(u4_buf,u4_buf_nxt,u4_offset,u4_bits,maxLen) do { maxLen_tmp--; /* Read one bit at a time from the variable to decode the huffman code */ m_iBit = (UWORD8)((u4_bits >> maxLen_tmp) & 0x1); /* Get the next node pointer or the symbol from the tree */ end = gai2_impeg2d_dct_dc_size[idx][end][m_iBit]; }while(end > 0); dc_size = end + MPEG2_DCT_DC_SIZE_OFFSET; /* Flush the appropriate number of bits from the stream */ FLUSH_BITS(u4_offset,u4_buf,u4_buf_nxt,(maxLen - maxLen_tmp),pu4_buf_aligned) } if (dc_size != 0) { UWORD32 u4_bits; IBITS_GET(u4_buf,u4_buf_nxt,u4_offset,u4_bits,pu4_buf_aligned, dc_size) dc_diff = u4_bits; if ((dc_diff & (1 << (dc_size - 1))) == 0) //v Probably the prediction algo? dc_diff -= (1 << dc_size) - 1; } else { dc_diff = 0; } pi2_outAddr[*pi4_num_coeffs] = dc_diff; /* This indicates the position of the coefficient. Since this is the DC * coefficient, we put the position as 0. */ pu1_pos[*pi4_num_coeffs] = pu1_scan[0]; (*pi4_num_coeffs)++; if (0 != dc_diff) { u4_nz_cols |= 0x01; u4_nz_rows |= 0x01; } u4_numCoeffs = 1; } /**************************************************************************/ /* Decoding of first AC coefficient in case of non Intra block */ /**************************************************************************/ else { /* First symbol can be 1s */ UWORD32 u4_bits; IBITS_NXT(u4_buf,u4_buf_nxt,u4_offset,u4_bits,1) if(u4_bits == 1) { FLUSH_BITS(u4_offset,u4_buf,u4_buf_nxt,1, pu4_buf_aligned) IBITS_GET(u4_buf,u4_buf_nxt,u4_offset,u4_bits,pu4_buf_aligned, 1) if(u4_bits == 1) { pi2_outAddr[*pi4_num_coeffs] = -1; } else { pi2_outAddr[*pi4_num_coeffs] = 1; } /* This indicates the position of the coefficient. Since this is the DC * coefficient, we put the position as 0. */ pu1_pos[*pi4_num_coeffs] = pu1_scan[0]; (*pi4_num_coeffs)++; u4_numCoeffs = 1; u4_nz_cols |= 0x01; u4_nz_rows |= 0x01; } else { u4_numCoeffs = 0; } } if (1 == u2_d_picture) { PUT_TEMP_STREAM_DATA(u4_buf, u4_buf_nxt, u4_offset, pu4_buf_aligned, ps_stream) ps_dec->u4_non_zero_cols = u4_nz_cols; ps_dec->u4_non_zero_rows = u4_nz_rows; return ((IMPEG2D_ERROR_CODES_T)IVD_ERROR_NONE); } if (1 == u2_intra_vlc_format && u2_intra_flag) { while(1) { //Putting the impeg2d_dec_ac_coeff_one function inline. UWORD32 lead_zeros; WORD16 DecodedValue; u4_sym_len = 17; IBITS_NXT(u4_buf,u4_buf_nxt,u4_offset,u4_bits,u4_sym_len) /* There cannot be more than 11 leading zeros in the decoded * symbol. The symbol is only 17 bits long, so we subtract 15. */ lead_zeros = CLZ(u4_bits) - 15; if (lead_zeros > 11) { return IMPEG2D_MB_DATA_DECODE_ERR; } DecodedValue = gau2_impeg2d_tab_one_1_9[u4_bits >> 8]; u4_sym_len = (DecodedValue & 0xf); i4_level = DecodedValue >> 9; /* One table lookup */ if(0 != i4_level) { u4_run = ((DecodedValue >> 4) & 0x1f); u4_numCoeffs += u4_run; if (u4_numCoeffs >= NUM_COEFFS) { return IMPEG2D_MB_TEX_DECODE_ERR; } u4_pos = pu1_scan[u4_numCoeffs++]; pu1_pos[*pi4_num_coeffs] = u4_pos; FLUSH_BITS(u4_offset,u4_buf,u4_buf_nxt,u4_sym_len,pu4_buf_aligned) pi2_outAddr[*pi4_num_coeffs] = i4_level; (*pi4_num_coeffs)++; } else { if (DecodedValue == END_OF_BLOCK_ONE) { u4_sym_len = 4; break; } else { /*Second table lookup*/ lead_zeros = CLZ(u4_bits) - 20;/* -16 since we are dealing with WORD32 */ if (0 != lead_zeros) { u4_bits = (u4_bits >> (6 - lead_zeros)) & 0x001F; /* Flush the number of bits */ if (1 == lead_zeros) { u4_sym_len = ((u4_bits & 0x18) >> 3) == 2 ? 11:10; } else { u4_sym_len = 11 + lead_zeros; } /* flushing */ FLUSH_BITS(u4_offset,u4_buf,u4_buf_nxt,u4_sym_len,pu4_buf_aligned) /* Calculate the address */ u4_bits = ((lead_zeros - 1) << 5) + u4_bits; DecodedValue = gau2_impeg2d_tab_one_10_16[u4_bits]; u4_run = BITS(DecodedValue, 8,4); i4_level = ((WORD16) DecodedValue) >> 9; u4_numCoeffs += u4_run; if (u4_numCoeffs >= NUM_COEFFS) { return IMPEG2D_MB_TEX_DECODE_ERR; } u4_pos = pu1_scan[u4_numCoeffs++]; pu1_pos[*pi4_num_coeffs] = u4_pos; pi2_outAddr[*pi4_num_coeffs] = i4_level; (*pi4_num_coeffs)++; } /*********************************************************************/ /* MPEG2 Escape Code */ /*********************************************************************/ else if(u2_mpeg2 == 1) { u4_sym_len = 6; FLUSH_BITS(u4_offset,u4_buf,u4_buf_nxt,u4_sym_len,pu4_buf_aligned) IBITS_GET(u4_buf,u4_buf_nxt,u4_offset,u4_bits,pu4_buf_aligned,18) u4_decoded_value = u4_bits; u4_run = (u4_decoded_value >> 12); i4_level = (u4_decoded_value & 0x0FFF); if (i4_level) i4_level = (i4_level - ((i4_level & 0x0800) << 1)); u4_numCoeffs += u4_run; if (u4_numCoeffs >= NUM_COEFFS) { return IMPEG2D_MB_TEX_DECODE_ERR; } u4_pos = pu1_scan[u4_numCoeffs++]; pu1_pos[*pi4_num_coeffs] = u4_pos; pi2_outAddr[*pi4_num_coeffs] = i4_level; (*pi4_num_coeffs)++; } /*********************************************************************/ /* MPEG1 Escape Code */ /*********************************************************************/ else { /*----------------------------------------------------------- * MPEG-1 Stream * * <See D.9.3 of MPEG-2> Run-level escape syntax * Run-level values that cannot be coded with a VLC are coded * by the escape code '0000 01' followed by * either a 14-bit FLC (127 <= level <= 127), * or a 22-bit FLC (255 <= level <= 255). * This is described in Annex B,B.5f of MPEG-1.standard *-----------------------------------------------------------*/ /*----------------------------------------------------------- * First 6 bits are the value of the Run. Next is First 8 bits * of Level. These bits decide whether it is 14 bit FLC or * 22-bit FLC. * * If( first 8 bits of Level == '1000000' or '00000000') * then its is 22-bit FLC. * else * it is 14-bit FLC. *-----------------------------------------------------------*/ u4_sym_len = 6; FLUSH_BITS(u4_offset,u4_buf,u4_buf_nxt,u4_sym_len,pu4_buf_aligned) IBITS_GET(u4_buf,u4_buf_nxt,u4_offset,u4_bits,pu4_buf_aligned,14) u4_decoded_value = u4_bits; u4_run = (u4_decoded_value >> 8); i4_level_first_byte = (u4_decoded_value & 0x0FF); if(i4_level_first_byte & 0x7F) { /*------------------------------------------------------- * First 8 bits of level are neither 1000000 nor 00000000 * Hence 14-bit FLC (Last 8 bits are used to get level) * * Level = (msb of Level_First_Byte is 1)? * Level_First_Byte - 256 : Level_First_Byte *-------------------------------------------------------*/ i4_level = (i4_level_first_byte - ((i4_level_first_byte & 0x80) << 1)); } else { /*------------------------------------------------------- * Next 8 bits are either 1000000 or 00000000 * Hence 22-bit FLC (Last 16 bits are used to get level) * * Level = (msb of Level_First_Byte is 1)? * Level_Second_Byte - 256 : Level_Second_Byte *-------------------------------------------------------*/ IBITS_GET(u4_buf,u4_buf_nxt,u4_offset,u4_bits,pu4_buf_aligned,8) i4_level = u4_bits; i4_level = (i4_level - (i4_level_first_byte << 1)); } u4_numCoeffs += u4_run; if (u4_numCoeffs >= NUM_COEFFS) { return IMPEG2D_MB_TEX_DECODE_ERR; } u4_pos = pu1_scan[u4_numCoeffs++]; pu1_pos[*pi4_num_coeffs] = u4_pos; pi2_outAddr[*pi4_num_coeffs] = i4_level; (*pi4_num_coeffs)++; } } } u4_nz_cols |= 1 << (u4_pos & 0x7); u4_nz_rows |= 1 << (u4_pos >> 0x3); } IBITS_GET(u4_buf,u4_buf_nxt,u4_offset,u4_bits,pu4_buf_aligned,u4_sym_len) } else { // Inline while(1) { UWORD32 lead_zeros; UWORD16 DecodedValue; u4_sym_len = 17; IBITS_NXT(u4_buf, u4_buf_nxt, u4_offset, u4_bits, u4_sym_len) /* There cannot be more than 11 leading zeros in the decoded * symbol. The symbol is only 17 bits long, so we subtract 15. */ lead_zeros = CLZ(u4_bits) - 15; if (lead_zeros > 11) { return IMPEG2D_MB_DATA_DECODE_ERR; } DecodedValue = gau2_impeg2d_tab_zero_1_9[u4_bits >> 8]; u4_sym_len = BITS(DecodedValue, 3, 0); i4_level = ((WORD16) DecodedValue) >> 9; if (0 != i4_level) { u4_run = BITS(DecodedValue, 8,4); u4_numCoeffs += u4_run; if (u4_numCoeffs >= NUM_COEFFS) { return IMPEG2D_MB_TEX_DECODE_ERR; } u4_pos = pu1_scan[u4_numCoeffs++]; pu1_pos[*pi4_num_coeffs] = u4_pos; FLUSH_BITS(u4_offset,u4_buf,u4_buf_nxt,u4_sym_len,pu4_buf_aligned) pi2_outAddr[*pi4_num_coeffs] = i4_level; (*pi4_num_coeffs)++; } else { if(DecodedValue == END_OF_BLOCK_ZERO) { u4_sym_len = 2; break; } else { lead_zeros = CLZ(u4_bits) - 20;/* -15 since we are dealing with WORD32 */ /*Second table lookup*/ if (0 != lead_zeros) { u4_bits = (u4_bits >> (6 - lead_zeros)) & 0x001F; /* Flush the number of bits */ u4_sym_len = 11 + lead_zeros; /* Calculate the address */ u4_bits = ((lead_zeros - 1) << 5) + u4_bits; DecodedValue = gau2_impeg2d_tab_zero_10_16[u4_bits]; u4_run = BITS(DecodedValue, 8,4); i4_level = ((WORD16) DecodedValue) >> 9; u4_numCoeffs += u4_run; if (u4_numCoeffs >= NUM_COEFFS) { return IMPEG2D_MB_TEX_DECODE_ERR; } u4_pos = pu1_scan[u4_numCoeffs++]; pu1_pos[*pi4_num_coeffs] = u4_pos; if (1 == lead_zeros) u4_sym_len--; /* flushing */ FLUSH_BITS(u4_offset,u4_buf,u4_buf_nxt,u4_sym_len,pu4_buf_aligned) pi2_outAddr[*pi4_num_coeffs] = i4_level; (*pi4_num_coeffs)++; } /*Escape Sequence*/ else if(u2_mpeg2 == 1) { u4_sym_len = 6; FLUSH_BITS(u4_offset,u4_buf,u4_buf_nxt,u4_sym_len,pu4_buf_aligned) IBITS_GET(u4_buf,u4_buf_nxt,u4_offset,u4_bits,pu4_buf_aligned,18) u4_decoded_value = u4_bits; u4_run = (u4_decoded_value >> 12); i4_level = (u4_decoded_value & 0x0FFF); if (i4_level) i4_level = (i4_level - ((i4_level & 0x0800) << 1)); u4_numCoeffs += u4_run; if (u4_numCoeffs >= NUM_COEFFS) { return IMPEG2D_MB_TEX_DECODE_ERR; } u4_pos = pu1_scan[u4_numCoeffs++]; pu1_pos[*pi4_num_coeffs] = u4_pos; pi2_outAddr[*pi4_num_coeffs] = i4_level; (*pi4_num_coeffs)++; } /*********************************************************************/ /* MPEG1 Escape Code */ /*********************************************************************/ else { /*----------------------------------------------------------- * MPEG-1 Stream * * <See D.9.3 of MPEG-2> Run-level escape syntax * Run-level values that cannot be coded with a VLC are coded * by the escape code '0000 01' followed by * either a 14-bit FLC (127 <= level <= 127), * or a 22-bit FLC (255 <= level <= 255). * This is described in Annex B,B.5f of MPEG-1.standard *-----------------------------------------------------------*/ /*----------------------------------------------------------- * First 6 bits are the value of the Run. Next is First 8 bits * of Level. These bits decide whether it is 14 bit FLC or * 22-bit FLC. * * If( first 8 bits of Level == '1000000' or '00000000') * then its is 22-bit FLC. * else * it is 14-bit FLC. *-----------------------------------------------------------*/ u4_sym_len = 6; FLUSH_BITS(u4_offset,u4_buf,u4_buf_nxt,u4_sym_len,pu4_buf_aligned) IBITS_GET(u4_buf,u4_buf_nxt,u4_offset,u4_bits,pu4_buf_aligned,14) u4_decoded_value = u4_bits; u4_run = (u4_decoded_value >> 8); i4_level_first_byte = (u4_decoded_value & 0x0FF); if(i4_level_first_byte & 0x7F) { /*------------------------------------------------------- * First 8 bits of level are neither 1000000 nor 00000000 * Hence 14-bit FLC (Last 8 bits are used to get level) * * Level = (msb of Level_First_Byte is 1)? * Level_First_Byte - 256 : Level_First_Byte *-------------------------------------------------------*/ i4_level = (i4_level_first_byte - ((i4_level_first_byte & 0x80) << 1)); } else { /*------------------------------------------------------- * Next 8 bits are either 1000000 or 00000000 * Hence 22-bit FLC (Last 16 bits are used to get level) * * Level = (msb of Level_First_Byte is 1)? * Level_Second_Byte - 256 : Level_Second_Byte *-------------------------------------------------------*/ IBITS_GET(u4_buf,u4_buf_nxt,u4_offset,u4_bits,pu4_buf_aligned,8) i4_level = u4_bits; i4_level = (i4_level - (i4_level_first_byte << 1)); } u4_numCoeffs += u4_run; if (u4_numCoeffs >= NUM_COEFFS) { return IMPEG2D_MB_TEX_DECODE_ERR; } u4_pos = pu1_scan[u4_numCoeffs++]; pu1_pos[*pi4_num_coeffs] = u4_pos; pi2_outAddr[*pi4_num_coeffs] = i4_level; (*pi4_num_coeffs)++; } } } u4_nz_cols |= 1 << (u4_pos & 0x7); u4_nz_rows |= 1 << (u4_pos >> 0x3); } IBITS_GET(u4_buf,u4_buf_nxt,u4_offset,u4_bits,pu4_buf_aligned,u4_sym_len) } PUT_TEMP_STREAM_DATA(u4_buf, u4_buf_nxt, u4_offset, pu4_buf_aligned, ps_stream) ps_dec->u4_non_zero_cols = u4_nz_cols; ps_dec->u4_non_zero_rows = u4_nz_rows; return (IMPEG2D_ERROR_CODES_T)IVD_ERROR_NONE; } /*****************************************************************************/ /* */ /* Function Name : impeg2d_inv_quant_mpeg1 */ /* */ /* Description : Inverse quantizes the output of VLD */ /* */ /* Inputs : */ /* blk, - Block to be inverse quantized */ /* weighting_matrix - Matrix to be used in inverse quant */ /* intra_dc_precision- Precision reqd to scale intra DC value */ /* quant_scale - Quanization scale for inverse quant */ /* intra_flag - Intra or Not */ /* */ /* Globals : None */ /* */ /* Processing : Implements the inverse quantize equation */ /* */ /* Outputs : Inverse quantized values in the block */ /* */ /* Returns : None */ /* */ /* Issues : None */ /* */ /* Revision History: */ /* */ /* DD MM YYYY Author(s) Changes */ /* 05 09 2005 Harish M First Version */ /* */ /*****************************************************************************/ WORD32 impeg2d_inv_quant_mpeg1(WORD16 *pi2_blk, UWORD8 *pu1_weighting_matrix, UWORD8 u1_quant_scale, WORD32 u4_intra_flag, WORD32 i4_num_coeffs, WORD16 *pi2_coeffs, UWORD8 *pu1_pos, const UWORD8 *pu1_scan, UWORD16 *pu2_def_dc_pred, UWORD16 u2_intra_dc_precision) { UWORD16 i4_pos; WORD32 i4_iter; /* Inverse Quantize the predicted DC value for intra MB*/ if(u4_intra_flag == 1) { /**************************************************************************/ /* Decode the DC coefficient in case of Intra block and also update */ /* DC predictor value of the corresponding color component */ /**************************************************************************/ { pi2_coeffs[0] += *pu2_def_dc_pred; *pu2_def_dc_pred = pi2_coeffs[0]; pi2_coeffs[0] <<= (3 - u2_intra_dc_precision); pi2_coeffs[0] = CLIP_S12(pi2_coeffs[0]); } pi2_blk[pu1_scan[0]] = pi2_coeffs[0]; } /************************************************************************/ /* Inverse quantization of other DCT coefficients */ /************************************************************************/ for(i4_iter = u4_intra_flag; i4_iter < i4_num_coeffs; i4_iter++) { WORD16 sign; WORD32 temp, temp1; /* Position is the inverse scan of the index stored */ i4_pos = pu1_pos[i4_iter]; pi2_blk[i4_pos] = pi2_coeffs[i4_iter]; sign = SIGN(pi2_blk[i4_pos]); temp = ABS(pi2_blk[i4_pos] << 1); /* pi2_coeffs has only non-zero elements. So no need to check * if the coeff is non-zero. */ temp = temp + (1 * !u4_intra_flag); temp = temp * pu1_weighting_matrix[i4_pos] * u1_quant_scale; temp = temp >> 5; temp1 = temp | 1; temp1 = (temp1 > temp) ? (temp1 - temp) : (temp - temp1); temp = temp - temp1; if(temp < 0) { temp = 0; } temp = temp * sign; temp = CLIP_S12(temp); pi2_blk[i4_pos] = temp; } /*return value is used in the case of mpeg2 for mismatch control*/ return (0); } /* End of inv_quant() */ /*****************************************************************************/ /* */ /* Function Name : impeg2d_inv_quant_mpeg2 */ /* */ /* Description : Inverse quantizes the output of VLD */ /* */ /* Inputs : */ /* blk, - Block to be inverse quantized */ /* weighting_matrix - Matrix to be used in inverse quant */ /* intra_dc_precision- Precision reqd to scale intra DC value */ /* quant_scale - Quanization scale for inverse quant */ /* intra_flag - Intra or Not */ /* */ /* Globals : None */ /* */ /* Processing : Implements the inverse quantize equation */ /* */ /* Outputs : Inverse quantized values in the block */ /* */ /* Returns : None */ /* */ /* Issues : None */ /* */ /* Revision History: */ /* */ /* DD MM YYYY Author(s) Changes */ /* 05 09 2005 Harish M First Version */ /* */ /*****************************************************************************/ WORD32 impeg2d_inv_quant_mpeg2(WORD16 *pi2_blk, UWORD8 *pu1_weighting_matrix, UWORD8 u1_quant_scale, WORD32 u4_intra_flag, WORD32 i4_num_coeffs, WORD16 *pi2_coeffs, UWORD8 *pu1_pos, const UWORD8 *pu1_scan, UWORD16 *pu2_def_dc_pred, UWORD16 u2_intra_dc_precision) { WORD32 i4_pos; /* Used for Mismatch control */ WORD32 sum; WORD32 i4_iter; sum = 0; /* Inverse Quantize the predicted DC value for intra MB*/ if(u4_intra_flag == 1) { /**************************************************************************/ /* Decode the DC coefficient in case of Intra block and also update */ /* DC predictor value of the corresponding color component */ /**************************************************************************/ { pi2_coeffs[0] += *pu2_def_dc_pred; *pu2_def_dc_pred = pi2_coeffs[0]; pi2_coeffs[0] <<= (3 - u2_intra_dc_precision); pi2_coeffs[0] = CLIP_S12(pi2_coeffs[0]); } pi2_blk[pu1_scan[0]] = pi2_coeffs[0]; sum = pi2_blk[0]; } /************************************************************************/ /* Inverse quantization of other DCT coefficients */ /************************************************************************/ for(i4_iter = u4_intra_flag; i4_iter < i4_num_coeffs; i4_iter++) { WORD16 sign; WORD32 temp; /* Position is the inverse scan of the index stored */ i4_pos = pu1_pos[i4_iter]; pi2_blk[i4_pos] = pi2_coeffs[i4_iter]; sign = SIGN(pi2_blk[i4_pos]); temp = ABS(pi2_blk[i4_pos] << 1); temp = temp + (1 * !u4_intra_flag); temp = temp * pu1_weighting_matrix[i4_pos] * u1_quant_scale; temp = temp >> 5; temp = temp * sign; temp = CLIP_S12(temp); pi2_blk[i4_pos] = temp; sum += temp; } return (sum ^ 1); } /* End of inv_quant() */