libavcodec/qsvenc_hevc.c

   1 /*
   2  * Intel MediaSDK QSV based HEVC encoder
   3  *
   4  * This file is part of Libav.
   5  *
   6  * Libav is free software; you can redistribute it and/or
   7  * modify it under the terms of the GNU Lesser General Public
   8  * License as published by the Free Software Foundation; either
   9  * version 2.1 of the License, or (at your option) any later version.
  10  *
  11  * Libav is distributed in the hope that it will be useful,
  12  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  13  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  14  * Lesser General Public License for more details.
  15  *
  16  * You should have received a copy of the GNU Lesser General Public
  17  * License along with Libav; if not, write to the Free Software
  18  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  19  */
  20
  21
  22 #include <stdint.h>
  23 #include <sys/types.h>
  24
  25 #include <mfx/mfxvideo.h>
  26
  27 #include "libavutil/common.h"
  28 #include "libavutil/opt.h"
  29
  30 #include "avcodec.h"
  31 #include "bytestream.h"
  32 #include "get_bits.h"
  33 #include "hevc.h"
  34 #include "hevcdec.h"
  35 #include "h2645_parse.h"
  36 #include "internal.h"
  37 #include "qsv.h"
  38 #include "qsv_internal.h"
  39 #include "qsvenc.h"
  40
  41 enum LoadPlugin {
  42     LOAD_PLUGIN_NONE,
  43     LOAD_PLUGIN_HEVC_SW,
  44     LOAD_PLUGIN_HEVC_HW,
  45 };
  46
  47 typedef struct QSVHEVCEncContext {
  48     AVClass *class;
  49     QSVEncContext qsv;
  50     int load_plugin;
  51 } QSVHEVCEncContext;
  52
  53 static int generate_fake_vps(QSVEncContext *q, AVCodecContext *avctx)
  54 {
  55     GetByteContext gbc;
  56     PutByteContext pbc;
  57
  58     GetBitContext gb;
  59     H2645NAL sps_nal = { NULL };
  60     HEVCSPS sps = { 0 };
  61     HEVCVPS vps = { 0 };
  62     uint8_t vps_buf[128], vps_rbsp_buf[128];
  63     uint8_t *new_extradata;
  64     unsigned int sps_id;
  65     int ret, i, type, vps_size;
  66
  67     if (!avctx->extradata_size) {
  68         av_log(avctx, AV_LOG_ERROR, "No extradata returned from libmfx\n");
  69         return AVERROR_UNKNOWN;
  70     }
  71
  72     /* parse the SPS */
  73     ret = ff_h2645_extract_rbsp(avctx->extradata + 4, avctx->extradata_size - 4, &sps_nal);
  74     if (ret < 0) {
  75         av_log(avctx, AV_LOG_ERROR, "Error unescaping the SPS buffer\n");
  76         return ret;
  77     }
  78
  79     ret = init_get_bits8(&gb, sps_nal.data, sps_nal.size);
  80     if (ret < 0) {
  81         av_freep(&sps_nal.rbsp_buffer);
  82         return ret;
  83     }
  84
  85     get_bits(&gb, 1);
  86     type = get_bits(&gb, 6);
  87     if (type != HEVC_NAL_SPS) {
  88         av_log(avctx, AV_LOG_ERROR, "Unexpected NAL type in the extradata: %d\n",
  89                type);
  90         av_freep(&sps_nal.rbsp_buffer);
  91         return AVERROR_INVALIDDATA;
  92     }
  93     get_bits(&gb, 9);
  94
  95     ret = ff_hevc_parse_sps(&sps, &gb, &sps_id, 0, NULL, avctx);
  96     av_freep(&sps_nal.rbsp_buffer);
  97     if (ret < 0) {
  98         av_log(avctx, AV_LOG_ERROR, "Error parsing the SPS\n");
  99         return ret;
 100     }
 101
 102     /* generate the VPS */
 103     vps.vps_max_layers     = 1;
 104     vps.vps_max_sub_layers = sps.max_sub_layers;
 105     memcpy(&vps.ptl, &sps.ptl, sizeof(vps.ptl));
 106     vps.vps_sub_layer_ordering_info_present_flag = 1;
 107     for (i = 0; i < HEVC_MAX_SUB_LAYERS; i++) {
 108         vps.vps_max_dec_pic_buffering[i] = sps.temporal_layer[i].max_dec_pic_buffering;
 109         vps.vps_num_reorder_pics[i]      = sps.temporal_layer[i].num_reorder_pics;
 110         vps.vps_max_latency_increase[i]  = sps.temporal_layer[i].max_latency_increase;
 111     }
 112
 113     vps.vps_num_layer_sets                  = 1;
 114     vps.vps_timing_info_present_flag        = sps.vui.vui_timing_info_present_flag;
 115     vps.vps_num_units_in_tick               = sps.vui.vui_num_units_in_tick;
 116     vps.vps_time_scale                      = sps.vui.vui_time_scale;
 117     vps.vps_poc_proportional_to_timing_flag = sps.vui.vui_poc_proportional_to_timing_flag;
 118     vps.vps_num_ticks_poc_diff_one          = sps.vui.vui_num_ticks_poc_diff_one_minus1 + 1;
 119
 120     /* generate the encoded RBSP form of the VPS */
 121     ret = ff_hevc_encode_nal_vps(&vps, sps.vps_id, vps_rbsp_buf, sizeof(vps_rbsp_buf));
 122     if (ret < 0) {
 123         av_log(avctx, AV_LOG_ERROR, "Error writing the VPS\n");
 124         return ret;
 125     }
 126
 127     /* escape and add the startcode */
 128     bytestream2_init(&gbc, vps_rbsp_buf, ret);
 129     bytestream2_init_writer(&pbc, vps_buf, sizeof(vps_buf));
 130
 131     bytestream2_put_be32(&pbc, 1);                 // startcode
 132     bytestream2_put_byte(&pbc, HEVC_NAL_VPS << 1); // NAL
 133     bytestream2_put_byte(&pbc, 1);                 // header
 134
 135     while (bytestream2_get_bytes_left(&gbc)) {
 136         uint32_t b = bytestream2_peek_be24(&gbc);
 137         if (b <= 3) {
 138             bytestream2_put_be24(&pbc, 3);
 139             bytestream2_skip(&gbc, 2);
 140         } else
 141             bytestream2_put_byte(&pbc, bytestream2_get_byte(&gbc));
 142     }
 143
 144     vps_size = bytestream2_tell_p(&pbc);
 145     new_extradata = av_mallocz(vps_size + avctx->extradata_size + AV_INPUT_BUFFER_PADDING_SIZE);
 146     if (!new_extradata)
 147         return AVERROR(ENOMEM);
 148     memcpy(new_extradata, vps_buf, vps_size);
 149     memcpy(new_extradata + vps_size, avctx->extradata, avctx->extradata_size);
 150
 151     av_freep(&avctx->extradata);
 152     avctx->extradata       = new_extradata;
 153     avctx->extradata_size += vps_size;
 154
 155     return 0;
 156 }
 157
 158 static av_cold int qsv_enc_init(AVCodecContext *avctx)
 159 {
 160     QSVHEVCEncContext *q = avctx->priv_data;
 161     int ret;
 162
 163     if (q->load_plugin != LOAD_PLUGIN_NONE) {
 164         static const char * const uid_hevcenc_sw = "2fca99749fdb49aeb121a5b63ef568f7";
 165         static const char * const uid_hevcenc_hw = "6fadc791a0c2eb479ab6dcd5ea9da347";
 166
 167         if (q->qsv.load_plugins[0]) {
 168             av_log(avctx, AV_LOG_WARNING,
 169                    "load_plugins is not empty, but load_plugin is not set to 'none'."
 170                    "The load_plugin value will be ignored.\n");
 171         } else {
 172             av_freep(&q->qsv.load_plugins);
 173
 174             if (q->load_plugin == LOAD_PLUGIN_HEVC_SW)
 175                 q->qsv.load_plugins = av_strdup(uid_hevcenc_sw);
 176             else
 177                 q->qsv.load_plugins = av_strdup(uid_hevcenc_hw);
 178
 179             if (!q->qsv.load_plugins)
 180                 return AVERROR(ENOMEM);
 181         }
 182     }
 183
 184     // HEVC and H264 meaning of the value is shifted by 1, make it consistent
 185     q->qsv.idr_interval++;
 186
 187     ret = ff_qsv_enc_init(avctx, &q->qsv);
 188     if (ret < 0)
 189         return ret;
 190
 191     ret = generate_fake_vps(&q->qsv, avctx);
 192     if (ret < 0) {
 193         ff_qsv_enc_close(avctx, &q->qsv);
 194         return ret;
 195     }
 196
 197     return 0;
 198 }
 199
 200 static int qsv_enc_frame(AVCodecContext *avctx, AVPacket *pkt,
 201                          const AVFrame *frame, int *got_packet)
 202 {
 203     QSVHEVCEncContext *q = avctx->priv_data;
 204
 205     return ff_qsv_encode(avctx, &q->qsv, pkt, frame, got_packet);
 206 }
 207
 208 static av_cold int qsv_enc_close(AVCodecContext *avctx)
 209 {
 210     QSVHEVCEncContext *q = avctx->priv_data;
 211
 212     return ff_qsv_enc_close(avctx, &q->qsv);
 213 }
 214
 215 #if defined(_WIN32)
 216 #define LOAD_PLUGIN_DEFAULT LOAD_PLUGIN_HEVC_SW
 217 #else
 218 #define LOAD_PLUGIN_DEFAULT LOAD_PLUGIN_HEVC_HW
 219 #endif
 220
 221 #define OFFSET(x) offsetof(QSVHEVCEncContext, x)
 222 #define VE AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_ENCODING_PARAM
 223 static const AVOption options[] = {
 224     QSV_COMMON_OPTS
 225
 226     { "idr_interval", "Distance (in I-frames) between IDR frames", OFFSET(qsv.idr_interval), AV_OPT_TYPE_INT, { .i64 = 0 }, -1, INT_MAX, VE, "idr_interval" },
 227     { "begin_only", "Output an IDR-frame only at the beginning of the stream", 0, AV_OPT_TYPE_CONST, { .i64 = -1 }, 0, 0, VE, "idr_interval" },
 228     { "load_plugin", "A user plugin to load in an internal session", OFFSET(load_plugin), AV_OPT_TYPE_INT, { .i64 = LOAD_PLUGIN_DEFAULT }, LOAD_PLUGIN_NONE, LOAD_PLUGIN_HEVC_HW, VE, "load_plugin" },
 229     { "none",     NULL, 0, AV_OPT_TYPE_CONST, { .i64 = LOAD_PLUGIN_NONE },    0, 0, VE, "load_plugin" },
 230     { "hevc_sw",  NULL, 0, AV_OPT_TYPE_CONST, { .i64 = LOAD_PLUGIN_HEVC_SW }, 0, 0, VE, "load_plugin" },
 231     { "hevc_hw",  NULL, 0, AV_OPT_TYPE_CONST, { .i64 = LOAD_PLUGIN_HEVC_HW }, 0, 0, VE, "load_plugin" },
 232
 233     { "load_plugins", "A :-separate list of hexadecimal plugin UIDs to load in an internal session",
 234         OFFSET(qsv.load_plugins), AV_OPT_TYPE_STRING, { .str = "" }, 0, 0, VE },
 235
 236     { "profile", NULL, OFFSET(qsv.profile), AV_OPT_TYPE_INT, { .i64 = MFX_PROFILE_UNKNOWN }, 0, INT_MAX, VE, "profile" },
 237     { "unknown", NULL, 0, AV_OPT_TYPE_CONST, { .i64 = MFX_PROFILE_UNKNOWN      }, INT_MIN, INT_MAX,     VE, "profile" },
 238     { "main",    NULL, 0, AV_OPT_TYPE_CONST, { .i64 = MFX_PROFILE_HEVC_MAIN    }, INT_MIN, INT_MAX,     VE, "profile" },
 239     { "main10",  NULL, 0, AV_OPT_TYPE_CONST, { .i64 = MFX_PROFILE_HEVC_MAIN10  }, INT_MIN, INT_MAX,     VE, "profile" },
 240     { "mainsp",  NULL, 0, AV_OPT_TYPE_CONST, { .i64 = MFX_PROFILE_HEVC_MAINSP  }, INT_MIN, INT_MAX,     VE, "profile" },
 241 #if QSV_HAVE_VDENC
 242     { "low_power", "enable low power mode (experimental, many limitations by mfx version, HW platform, BRC modes, etc.", OFFSET(qsv.low_power), AV_OPT_TYPE_INT, { .i64 =  0 },  0,  1, VE },
 243 #endif
 244
 245     { NULL },
 246 };
 247
 248 static const AVClass class = {
 249     .class_name = "hevc_qsv encoder",
 250     .item_name  = av_default_item_name,
 251     .option     = options,
 252     .version    = LIBAVUTIL_VERSION_INT,
 253 };
 254
 255 static const AVCodecDefault qsv_enc_defaults[] = {
 256     { "b",         "1M"    },
 257     { "refs",      "0"     },
 258     // same as the x264 default
 259     { "g",         "248"   },
 260     { "bf",        "8"     },
 261
 262     { "flags",     "+cgop" },
 263 #if FF_API_PRIVATE_OPT
 264     { "b_strategy", "-1"   },
 265 #endif
 266     { NULL },
 267 };
 268
 269 AVCodec ff_hevc_qsv_encoder = {
 270     .name           = "hevc_qsv",
 271     .long_name      = NULL_IF_CONFIG_SMALL("HEVC (Intel Quick Sync Video acceleration)"),
 272     .priv_data_size = sizeof(QSVHEVCEncContext),
 273     .type           = AVMEDIA_TYPE_VIDEO,
 274     .id             = AV_CODEC_ID_HEVC,
 275     .init           = qsv_enc_init,
 276     .encode2        = qsv_enc_frame,
 277     .close          = qsv_enc_close,
 278     .capabilities   = AV_CODEC_CAP_DELAY | AV_CODEC_CAP_HYBRID,
 279     .pix_fmts       = (const enum AVPixelFormat[]){ AV_PIX_FMT_NV12,
 280                                                     AV_PIX_FMT_P010,
 281                                                     AV_PIX_FMT_QSV,
 282                                                     AV_PIX_FMT_NONE },
 283     .priv_class     = &class,
 284     .defaults       = qsv_enc_defaults,
 285     .caps_internal  = FF_CODEC_CAP_INIT_CLEANUP,
 286     .wrapper_name   = "qsv",
 287 };