2 * Videotoolbox hardware acceleration
4 * copyright (c) 2012 Sebastien Zwickert
6 * This file is part of FFmpeg.
8 * FFmpeg is free software; you can redistribute it and/or
9 * modify it under the terms of the GNU Lesser General Public
10 * License as published by the Free Software Foundation; either
11 * version 2.1 of the License, or (at your option) any later version.
13 * FFmpeg is distributed in the hope that it will be useful,
14 * but WITHOUT ANY WARRANTY; without even the implied warranty of
15 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
16 * Lesser General Public License for more details.
18 * You should have received a copy of the GNU Lesser General Public
19 * License along with FFmpeg; if not, write to the Free Software
20 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
24 #if CONFIG_VIDEOTOOLBOX
25 # include "videotoolbox.h"
29 #include "vda_vt_internal.h"
30 #include "libavutil/avutil.h"
31 #include "bytestream.h"
33 #include "mpegvideo.h"
35 #ifndef kVTVideoDecoderSpecification_RequireHardwareAcceleratedVideoDecoder
36 # define kVTVideoDecoderSpecification_RequireHardwareAcceleratedVideoDecoder CFSTR("RequireHardwareAcceleratedVideoDecoder")
39 #define VIDEOTOOLBOX_ESDS_EXTRADATA_PADDING 12
41 static void videotoolbox_buffer_release(void *opaque, uint8_t *data)
43 CVPixelBufferRef cv_buffer = (CVImageBufferRef)data;
44 CVPixelBufferRelease(cv_buffer);
47 static int videotoolbox_buffer_copy(VTContext *vtctx,
48 const uint8_t *buffer,
53 tmp = av_fast_realloc(vtctx->bitstream,
54 &vtctx->allocated_size,
58 return AVERROR(ENOMEM);
60 vtctx->bitstream = tmp;
61 memcpy(vtctx->bitstream, buffer, size);
62 vtctx->bitstream_size = size;
67 int ff_videotoolbox_alloc_frame(AVCodecContext *avctx, AVFrame *frame)
69 frame->width = avctx->width;
70 frame->height = avctx->height;
71 frame->format = avctx->pix_fmt;
72 frame->buf[0] = av_buffer_alloc(1);
75 return AVERROR(ENOMEM);
80 #define AV_W8(p, v) *(p) = (v)
82 CFDataRef ff_videotoolbox_avcc_extradata_create(AVCodecContext *avctx)
84 H264Context *h = avctx->priv_data;
85 CFDataRef data = NULL;
87 int vt_extradata_size = 6 + 3 + h->sps.data_size + 4 + h->pps.data_size;
88 uint8_t *vt_extradata = av_malloc(vt_extradata_size);
94 AV_W8(p + 0, 1); /* version */
95 AV_W8(p + 1, h->sps.data[0]); /* profile */
96 AV_W8(p + 2, h->sps.data[1]); /* profile compat */
97 AV_W8(p + 3, h->sps.data[2]); /* level */
98 AV_W8(p + 4, 0xff); /* 6 bits reserved (111111) + 2 bits nal size length - 3 (11) */
99 AV_W8(p + 5, 0xe1); /* 3 bits reserved (111) + 5 bits number of sps (00001) */
100 AV_WB16(p + 6, h->sps.data_size + 1);
101 AV_W8(p + 8, NAL_SPS | (3 << 5)); // NAL unit header
102 memcpy(p + 9, h->sps.data, h->sps.data_size);
103 p += 9 + h->sps.data_size;
104 AV_W8(p + 0, 1); /* number of pps */
105 AV_WB16(p + 1, h->pps.data_size + 1);
106 AV_W8(p + 3, NAL_PPS | (3 << 5)); // NAL unit header
107 memcpy(p + 4, h->pps.data, h->pps.data_size);
109 p += 4 + h->pps.data_size;
110 av_assert0(p - vt_extradata == vt_extradata_size);
112 data = CFDataCreate(kCFAllocatorDefault, vt_extradata, vt_extradata_size);
113 av_free(vt_extradata);
117 int ff_videotoolbox_buffer_create(VTContext *vtctx, AVFrame *frame)
119 av_buffer_unref(&frame->buf[0]);
121 frame->buf[0] = av_buffer_create((uint8_t*)vtctx->frame,
122 sizeof(vtctx->frame),
123 videotoolbox_buffer_release,
125 AV_BUFFER_FLAG_READONLY);
126 if (!frame->buf[0]) {
127 return AVERROR(ENOMEM);
130 frame->data[3] = (uint8_t*)vtctx->frame;
136 int ff_videotoolbox_h264_start_frame(AVCodecContext *avctx,
137 const uint8_t *buffer,
140 VTContext *vtctx = avctx->internal->hwaccel_priv_data;
141 H264Context *h = avctx->priv_data;
143 vtctx->bitstream_size = 0;
145 if (h->is_avc == 1) {
146 return videotoolbox_buffer_copy(vtctx, buffer, size);
152 int ff_videotoolbox_h264_decode_slice(AVCodecContext *avctx,
153 const uint8_t *buffer,
156 VTContext *vtctx = avctx->internal->hwaccel_priv_data;
157 H264Context *h = avctx->priv_data;
163 tmp = av_fast_realloc(vtctx->bitstream,
164 &vtctx->allocated_size,
165 vtctx->bitstream_size+size+4);
167 return AVERROR(ENOMEM);
169 vtctx->bitstream = tmp;
171 AV_WB32(vtctx->bitstream + vtctx->bitstream_size, size);
172 memcpy(vtctx->bitstream + vtctx->bitstream_size + 4, buffer, size);
174 vtctx->bitstream_size += size + 4;
179 int ff_videotoolbox_uninit(AVCodecContext *avctx)
181 VTContext *vtctx = avctx->internal->hwaccel_priv_data;
183 av_freep(&vtctx->bitstream);
185 CVPixelBufferRelease(vtctx->frame);
191 #if CONFIG_VIDEOTOOLBOX
192 static void videotoolbox_write_mp4_descr_length(PutByteContext *pb, int length)
197 for (i = 3; i >= 0; i--) {
198 b = (length >> (i * 7)) & 0x7F;
202 bytestream2_put_byteu(pb, b);
206 static CFDataRef videotoolbox_esds_extradata_create(AVCodecContext *avctx)
209 uint8_t *rw_extradata;
211 int full_size = 3 + 5 + 13 + 5 + avctx->extradata_size + 3;
212 // ES_DescrTag data + DecoderConfigDescrTag + data + DecSpecificInfoTag + size + SLConfigDescriptor
213 int config_size = 13 + 5 + avctx->extradata_size;
216 if (!(rw_extradata = av_mallocz(full_size + VIDEOTOOLBOX_ESDS_EXTRADATA_PADDING)))
219 bytestream2_init_writer(&pb, rw_extradata, full_size + VIDEOTOOLBOX_ESDS_EXTRADATA_PADDING);
220 bytestream2_put_byteu(&pb, 0); // version
221 bytestream2_put_ne24(&pb, 0); // flags
223 // elementary stream descriptor
224 bytestream2_put_byteu(&pb, 0x03); // ES_DescrTag
225 videotoolbox_write_mp4_descr_length(&pb, full_size);
226 bytestream2_put_ne16(&pb, 0); // esid
227 bytestream2_put_byteu(&pb, 0); // stream priority (0-32)
229 // decoder configuration descriptor
230 bytestream2_put_byteu(&pb, 0x04); // DecoderConfigDescrTag
231 videotoolbox_write_mp4_descr_length(&pb, config_size);
232 bytestream2_put_byteu(&pb, 32); // object type indication. 32 = AV_CODEC_ID_MPEG4
233 bytestream2_put_byteu(&pb, 0x11); // stream type
234 bytestream2_put_ne24(&pb, 0); // buffer size
235 bytestream2_put_ne32(&pb, 0); // max bitrate
236 bytestream2_put_ne32(&pb, 0); // avg bitrate
238 // decoder specific descriptor
239 bytestream2_put_byteu(&pb, 0x05); ///< DecSpecificInfoTag
240 videotoolbox_write_mp4_descr_length(&pb, avctx->extradata_size);
242 bytestream2_put_buffer(&pb, avctx->extradata, avctx->extradata_size);
244 // SLConfigDescriptor
245 bytestream2_put_byteu(&pb, 0x06); // SLConfigDescrTag
246 bytestream2_put_byteu(&pb, 0x01); // length
247 bytestream2_put_byteu(&pb, 0x02); //
249 s = bytestream2_size_p(&pb);
251 data = CFDataCreate(kCFAllocatorDefault, rw_extradata, s);
253 av_freep(&rw_extradata);
257 static CMSampleBufferRef videotoolbox_sample_buffer_create(CMFormatDescriptionRef fmt_desc,
262 CMBlockBufferRef block_buf;
263 CMSampleBufferRef sample_buf;
268 status = CMBlockBufferCreateWithMemoryBlock(kCFAllocatorDefault,// structureAllocator
269 buffer, // memoryBlock
271 kCFAllocatorNull, // blockAllocator
272 NULL, // customBlockSource
279 status = CMSampleBufferCreate(kCFAllocatorDefault, // allocator
280 block_buf, // dataBuffer
282 0, // makeDataReadyCallback
283 0, // makeDataReadyRefcon
284 fmt_desc, // formatDescription
286 0, // numSampleTimingEntries
287 NULL, // sampleTimingArray
288 0, // numSampleSizeEntries
289 NULL, // sampleSizeArray
294 CFRelease(block_buf);
299 static void videotoolbox_decoder_callback(void *opaque,
300 void *sourceFrameRefCon,
302 VTDecodeInfoFlags flags,
303 CVImageBufferRef image_buffer,
307 AVCodecContext *avctx = opaque;
308 VTContext *vtctx = avctx->internal->hwaccel_priv_data;
311 CVPixelBufferRelease(vtctx->frame);
316 av_log(NULL, AV_LOG_DEBUG, "vt decoder cb: output image buffer is null\n");
320 vtctx->frame = CVPixelBufferRetain(image_buffer);
323 static OSStatus videotoolbox_session_decode_frame(AVCodecContext *avctx)
326 CMSampleBufferRef sample_buf;
327 AVVideotoolboxContext *videotoolbox = avctx->hwaccel_context;
328 VTContext *vtctx = avctx->internal->hwaccel_priv_data;
330 sample_buf = videotoolbox_sample_buffer_create(videotoolbox->cm_fmt_desc,
332 vtctx->bitstream_size);
337 status = VTDecompressionSessionDecodeFrame(videotoolbox->session,
340 NULL, // sourceFrameRefCon
343 status = VTDecompressionSessionWaitForAsynchronousFrames(videotoolbox->session);
345 CFRelease(sample_buf);
350 static int videotoolbox_common_end_frame(AVCodecContext *avctx, AVFrame *frame)
353 AVVideotoolboxContext *videotoolbox = avctx->hwaccel_context;
354 VTContext *vtctx = avctx->internal->hwaccel_priv_data;
356 if (!videotoolbox->session || !vtctx->bitstream)
357 return AVERROR_INVALIDDATA;
359 status = videotoolbox_session_decode_frame(avctx);
362 av_log(avctx, AV_LOG_ERROR, "Failed to decode frame (%d)\n", status);
363 return AVERROR_UNKNOWN;
367 return AVERROR_UNKNOWN;
369 return ff_videotoolbox_buffer_create(vtctx, frame);
372 static int videotoolbox_h264_end_frame(AVCodecContext *avctx)
374 H264Context *h = avctx->priv_data;
375 AVFrame *frame = h->cur_pic_ptr->f;
377 return videotoolbox_common_end_frame(avctx, frame);
380 static int videotoolbox_mpeg_start_frame(AVCodecContext *avctx,
381 const uint8_t *buffer,
384 VTContext *vtctx = avctx->internal->hwaccel_priv_data;
386 return videotoolbox_buffer_copy(vtctx, buffer, size);
389 static int videotoolbox_mpeg_decode_slice(AVCodecContext *avctx,
390 const uint8_t *buffer,
396 static int videotoolbox_mpeg_end_frame(AVCodecContext *avctx)
398 MpegEncContext *s = avctx->priv_data;
399 AVFrame *frame = s->current_picture_ptr->f;
401 return videotoolbox_common_end_frame(avctx, frame);
404 static CFDictionaryRef videotoolbox_decoder_config_create(CMVideoCodecType codec_type,
405 AVCodecContext *avctx)
407 CFMutableDictionaryRef config_info = CFDictionaryCreateMutable(kCFAllocatorDefault,
409 &kCFTypeDictionaryKeyCallBacks,
410 &kCFTypeDictionaryValueCallBacks);
412 CFDictionarySetValue(config_info,
413 kVTVideoDecoderSpecification_RequireHardwareAcceleratedVideoDecoder,
416 if (avctx->extradata_size) {
417 CFMutableDictionaryRef avc_info;
418 CFDataRef data = NULL;
420 avc_info = CFDictionaryCreateMutable(kCFAllocatorDefault,
422 &kCFTypeDictionaryKeyCallBacks,
423 &kCFTypeDictionaryValueCallBacks);
425 switch (codec_type) {
426 case kCMVideoCodecType_MPEG4Video :
427 data = videotoolbox_esds_extradata_create(avctx);
429 CFDictionarySetValue(avc_info, CFSTR("esds"), data);
431 case kCMVideoCodecType_H264 :
432 data = ff_videotoolbox_avcc_extradata_create(avctx);
434 CFDictionarySetValue(avc_info, CFSTR("avcC"), data);
440 CFDictionarySetValue(config_info,
441 kCMFormatDescriptionExtension_SampleDescriptionExtensionAtoms,
452 static CFDictionaryRef videotoolbox_buffer_attributes_create(int width,
456 CFMutableDictionaryRef buffer_attributes;
457 CFMutableDictionaryRef io_surface_properties;
458 CFNumberRef cv_pix_fmt;
462 w = CFNumberCreate(kCFAllocatorDefault, kCFNumberSInt32Type, &width);
463 h = CFNumberCreate(kCFAllocatorDefault, kCFNumberSInt32Type, &height);
464 cv_pix_fmt = CFNumberCreate(kCFAllocatorDefault, kCFNumberSInt32Type, &pix_fmt);
466 buffer_attributes = CFDictionaryCreateMutable(kCFAllocatorDefault,
468 &kCFTypeDictionaryKeyCallBacks,
469 &kCFTypeDictionaryValueCallBacks);
470 io_surface_properties = CFDictionaryCreateMutable(kCFAllocatorDefault,
472 &kCFTypeDictionaryKeyCallBacks,
473 &kCFTypeDictionaryValueCallBacks);
475 CFDictionarySetValue(buffer_attributes, kCVPixelBufferPixelFormatTypeKey, cv_pix_fmt);
476 CFDictionarySetValue(buffer_attributes, kCVPixelBufferIOSurfacePropertiesKey, io_surface_properties);
477 CFDictionarySetValue(buffer_attributes, kCVPixelBufferWidthKey, w);
478 CFDictionarySetValue(buffer_attributes, kCVPixelBufferHeightKey, h);
480 CFRelease(io_surface_properties);
481 CFRelease(cv_pix_fmt);
485 return buffer_attributes;
488 static CMVideoFormatDescriptionRef videotoolbox_format_desc_create(CMVideoCodecType codec_type,
489 CFDictionaryRef decoder_spec,
493 CMFormatDescriptionRef cm_fmt_desc;
496 status = CMVideoFormatDescriptionCreate(kCFAllocatorDefault,
500 decoder_spec, // Dictionary of extension
509 static int videotoolbox_default_init(AVCodecContext *avctx)
511 AVVideotoolboxContext *videotoolbox = avctx->hwaccel_context;
513 VTDecompressionOutputCallbackRecord decoder_cb;
514 CFDictionaryRef decoder_spec;
515 CFDictionaryRef buf_attr;
518 av_log(avctx, AV_LOG_ERROR, "hwaccel context is not set\n");
522 switch( avctx->codec_id ) {
523 case AV_CODEC_ID_H263 :
524 videotoolbox->cm_codec_type = kCMVideoCodecType_H263;
526 case AV_CODEC_ID_H264 :
527 videotoolbox->cm_codec_type = kCMVideoCodecType_H264;
529 case AV_CODEC_ID_MPEG1VIDEO :
530 videotoolbox->cm_codec_type = kCMVideoCodecType_MPEG1Video;
532 case AV_CODEC_ID_MPEG2VIDEO :
533 videotoolbox->cm_codec_type = kCMVideoCodecType_MPEG2Video;
535 case AV_CODEC_ID_MPEG4 :
536 videotoolbox->cm_codec_type = kCMVideoCodecType_MPEG4Video;
542 decoder_spec = videotoolbox_decoder_config_create(videotoolbox->cm_codec_type, avctx);
544 videotoolbox->cm_fmt_desc = videotoolbox_format_desc_create(videotoolbox->cm_codec_type,
548 if (!videotoolbox->cm_fmt_desc) {
550 CFRelease(decoder_spec);
552 av_log(avctx, AV_LOG_ERROR, "format description creation failed\n");
556 buf_attr = videotoolbox_buffer_attributes_create(avctx->width,
558 videotoolbox->cv_pix_fmt_type);
560 decoder_cb.decompressionOutputCallback = videotoolbox_decoder_callback;
561 decoder_cb.decompressionOutputRefCon = avctx;
563 status = VTDecompressionSessionCreate(NULL, // allocator
564 videotoolbox->cm_fmt_desc, // videoFormatDescription
565 decoder_spec, // videoDecoderSpecification
566 buf_attr, // destinationImageBufferAttributes
567 &decoder_cb, // outputCallback
568 &videotoolbox->session); // decompressionSessionOut
571 CFRelease(decoder_spec);
576 case kVTVideoDecoderNotAvailableNowErr:
577 case kVTVideoDecoderUnsupportedDataFormatErr:
578 return AVERROR(ENOSYS);
579 case kVTVideoDecoderMalfunctionErr:
580 return AVERROR(EINVAL);
581 case kVTVideoDecoderBadDataErr :
582 return AVERROR_INVALIDDATA;
586 return AVERROR_UNKNOWN;
590 static void videotoolbox_default_free(AVCodecContext *avctx)
592 AVVideotoolboxContext *videotoolbox = avctx->hwaccel_context;
595 if (videotoolbox->cm_fmt_desc)
596 CFRelease(videotoolbox->cm_fmt_desc);
598 if (videotoolbox->session)
599 VTDecompressionSessionInvalidate(videotoolbox->session);
603 AVHWAccel ff_h263_videotoolbox_hwaccel = {
604 .name = "h263_videotoolbox",
605 .type = AVMEDIA_TYPE_VIDEO,
606 .id = AV_CODEC_ID_H263,
607 .pix_fmt = AV_PIX_FMT_VIDEOTOOLBOX,
608 .alloc_frame = ff_videotoolbox_alloc_frame,
609 .start_frame = videotoolbox_mpeg_start_frame,
610 .decode_slice = videotoolbox_mpeg_decode_slice,
611 .end_frame = videotoolbox_mpeg_end_frame,
612 .uninit = ff_videotoolbox_uninit,
613 .priv_data_size = sizeof(VTContext),
616 AVHWAccel ff_h264_videotoolbox_hwaccel = {
617 .name = "h264_videotoolbox",
618 .type = AVMEDIA_TYPE_VIDEO,
619 .id = AV_CODEC_ID_H264,
620 .pix_fmt = AV_PIX_FMT_VIDEOTOOLBOX,
621 .alloc_frame = ff_videotoolbox_alloc_frame,
622 .start_frame = ff_videotoolbox_h264_start_frame,
623 .decode_slice = ff_videotoolbox_h264_decode_slice,
624 .end_frame = videotoolbox_h264_end_frame,
625 .uninit = ff_videotoolbox_uninit,
626 .priv_data_size = sizeof(VTContext),
629 AVHWAccel ff_mpeg1_videotoolbox_hwaccel = {
630 .name = "mpeg1_videotoolbox",
631 .type = AVMEDIA_TYPE_VIDEO,
632 .id = AV_CODEC_ID_MPEG1VIDEO,
633 .pix_fmt = AV_PIX_FMT_VIDEOTOOLBOX,
634 .alloc_frame = ff_videotoolbox_alloc_frame,
635 .start_frame = videotoolbox_mpeg_start_frame,
636 .decode_slice = videotoolbox_mpeg_decode_slice,
637 .end_frame = videotoolbox_mpeg_end_frame,
638 .uninit = ff_videotoolbox_uninit,
639 .priv_data_size = sizeof(VTContext),
642 AVHWAccel ff_mpeg2_videotoolbox_hwaccel = {
643 .name = "mpeg2_videotoolbox",
644 .type = AVMEDIA_TYPE_VIDEO,
645 .id = AV_CODEC_ID_MPEG2VIDEO,
646 .pix_fmt = AV_PIX_FMT_VIDEOTOOLBOX,
647 .alloc_frame = ff_videotoolbox_alloc_frame,
648 .start_frame = videotoolbox_mpeg_start_frame,
649 .decode_slice = videotoolbox_mpeg_decode_slice,
650 .end_frame = videotoolbox_mpeg_end_frame,
651 .uninit = ff_videotoolbox_uninit,
652 .priv_data_size = sizeof(VTContext),
655 AVHWAccel ff_mpeg4_videotoolbox_hwaccel = {
656 .name = "mpeg4_videotoolbox",
657 .type = AVMEDIA_TYPE_VIDEO,
658 .id = AV_CODEC_ID_MPEG4,
659 .pix_fmt = AV_PIX_FMT_VIDEOTOOLBOX,
660 .alloc_frame = ff_videotoolbox_alloc_frame,
661 .start_frame = videotoolbox_mpeg_start_frame,
662 .decode_slice = videotoolbox_mpeg_decode_slice,
663 .end_frame = videotoolbox_mpeg_end_frame,
664 .uninit = ff_videotoolbox_uninit,
665 .priv_data_size = sizeof(VTContext),
668 AVVideotoolboxContext *av_videotoolbox_alloc_context(void)
670 AVVideotoolboxContext *ret = av_mallocz(sizeof(*ret));
673 ret->output_callback = videotoolbox_decoder_callback;
674 ret->cv_pix_fmt_type = kCVPixelFormatType_420YpCbCr8BiPlanarVideoRange;
680 int av_videotoolbox_default_init(AVCodecContext *avctx)
682 return av_videotoolbox_default_init2(avctx, NULL);
685 int av_videotoolbox_default_init2(AVCodecContext *avctx, AVVideotoolboxContext *vtctx)
687 avctx->hwaccel_context = vtctx ?: av_videotoolbox_alloc_context();
688 if (!avctx->hwaccel_context)
689 return AVERROR(ENOMEM);
690 return videotoolbox_default_init(avctx);
693 void av_videotoolbox_default_free(AVCodecContext *avctx)
696 videotoolbox_default_free(avctx);
697 av_freep(&avctx->hwaccel_context);
699 #endif /* CONFIG_VIDEOTOOLBOX */