ffmpeg/libavcodec/codec_internal.h

311 lines
11 KiB
C
Raw Normal View History

/*
* This file is part of FFmpeg.
*
* FFmpeg is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2.1 of the License, or (at your option) any later version.
*
* FFmpeg is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with FFmpeg; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
#ifndef AVCODEC_CODEC_INTERNAL_H
#define AVCODEC_CODEC_INTERNAL_H
#include <stdint.h>
#include "libavutil/attributes.h"
#include "codec.h"
#include "config.h"
/**
* The codec is not known to be init-threadsafe (i.e. it might be unsafe
* to initialize this codec and another codec concurrently, typically because
* the codec calls external APIs that are not known to be thread-safe).
* Therefore calling the codec's init function needs to be guarded with a lock.
*/
#define FF_CODEC_CAP_NOT_INIT_THREADSAFE (1 << 0)
/**
* The codec allows calling the close function for deallocation even if
* the init function returned a failure. Without this capability flag, a
* codec does such cleanup internally when returning failures from the
* init function and does not expect the close function to be called at
* all.
*/
#define FF_CODEC_CAP_INIT_CLEANUP (1 << 1)
/**
* Decoders marked with FF_CODEC_CAP_SETS_PKT_DTS want to set
* AVFrame.pkt_dts manually. If the flag is set, decode.c won't overwrite
* this field. If it's unset, decode.c tries to guess the pkt_dts field
* from the input AVPacket.
*/
#define FF_CODEC_CAP_SETS_PKT_DTS (1 << 2)
/**
* The decoder extracts and fills its parameters even if the frame is
* skipped due to the skip_frame setting.
*/
#define FF_CODEC_CAP_SKIP_FRAME_FILL_PARAM (1 << 3)
/**
* The decoder sets the cropping fields in the output frames manually.
* If this cap is set, the generic code will initialize output frame
* dimensions to coded rather than display values.
*/
#define FF_CODEC_CAP_EXPORTS_CROPPING (1 << 4)
/**
* Codec initializes slice-based threading with a main function
*/
#define FF_CODEC_CAP_SLICE_THREAD_HAS_MF (1 << 5)
avcodec/decode: Add new ProgressFrame API Frame-threaded decoders with inter-frame dependencies use the ThreadFrame API for syncing. It works as follows: During init each thread allocates an AVFrame for every ThreadFrame. Thread A reads the header of its packet and allocates a buffer for an AVFrame with ff_thread_get_ext_buffer() (which also allocates a small structure that is shared with other references to this frame) and sets its fields, including side data. Then said thread calls ff_thread_finish_setup(). From that moment onward it is not allowed to change any of the AVFrame fields at all any more, but it may change fields which are an indirection away, like the content of AVFrame.data or already existing side data. After thread A has called ff_thread_finish_setup(), another thread (the user one) calls the codec's update_thread_context callback which in turn calls ff_thread_ref_frame() which calls av_frame_ref() which reads every field of A's AVFrame; hence the above restriction on modifications of the AVFrame (as any modification of the AVFrame by A after ff_thread_finish_setup() would be a data race). Of course, this av_frame_ref() also incurs allocations and therefore needs to be checked. ff_thread_ref_frame() also references the small structure used for communicating progress. This av_frame_ref() makes it awkward to propagate values that only become known during decoding to later threads (in case of frame reordering or other mechanisms of delayed output (like show-existing-frames) it's not the decoding thread, but a later thread that returns the AVFrame). E.g. for VP9 when exporting video encoding parameters as side data the number of blocks only becomes known during decoding, so one can't allocate the side data before ff_thread_finish_setup(). It is currently being done afterwards and this leads to a data race in the vp9-encparams test when using frame-threading. Returning decode_error_flags is also complicated by this. To perform this exchange a buffer shared between the references is needed (notice that simply giving the later threads a pointer to the original AVFrame does not work, because said AVFrame will be reused lateron when thread A decodes the next packet given to it). One could extend the buffer already used for progress for this or use a new one (requiring yet another allocation), yet both of these approaches have the drawback of being unnatural, ugly and requiring quite a lot of ad-hoc code. E.g. in case of the VP9 side data mentioned above one could not simply use the helper that allocates and adds the side data to an AVFrame in one go. The ProgressFrame API meanwhile offers a different solution to all of this. It is based around the idea that the most natural shared object for sharing information about an AVFrame between decoding threads is the AVFrame itself. To actually implement this the AVFrame needs to be reference counted. This is achieved by putting a (ownership) pointer into a shared (and opaque) structure that is managed by the RefStruct API and which also contains the stuff necessary for progress reporting. The users get a pointer to this AVFrame with the understanding that the owner may set all the fields until it has indicated that it has finished decoding this AVFrame; then the users are allowed to read everything. Every decoder may of course employ a different contract than the one outlined above. Given that there is no underlying av_frame_ref(), creating references to a ProgressFrame can't fail. Only ff_thread_progress_get_buffer() can fail, but given that it will replace calls to ff_thread_get_ext_buffer() it is at places where errors are already expected and properly taken care of. The ProgressFrames are empty (i.e. the AVFrame pointer is NULL and the AVFrames are not allocated during init at all) while not being in use; ff_thread_progress_get_buffer() both sets up the actual ProgressFrame and already calls ff_thread_get_buffer(). So instead of checking for ThreadFrame.f->data[0] or ThreadFrame.f->buf[0] being NULL for "this reference frame is non-existing" one should check for ProgressFrame.f. This also implies that one can only set AVFrame properties after having allocated the buffer. This restriction is not deep: if it becomes onerous for any codec, ff_thread_progress_get_buffer() can be broken up. The user would then have to get a buffer himself. In order to avoid unnecessary allocations, the shared structure is pooled, so that both the structure as well as the AVFrame itself are reused. This means that there won't be lots of unnecessary allocations in case of non-frame-threaded decoding. It might even turn out to have fewer than the current code (the current code allocates AVFrames for every DPB slot, but these are often excessively large and not completely used; the new code allocates them on demand). Pooling relies on the reset function of the RefStruct pool API, it would be impossible to implement with the AVBufferPool API. Finally, ProgressFrames have no notion of owner; they are built on top of the ThreadProgress API which also lacks such a concept. Instead every ThreadProgress and every ProgressFrame contains its own mutex and condition variable, making it completely independent of pthread_frame.c. Just like the ThreadFrame API it is simply presumed that only the actual owner/producer of a frame reports progress on said frame. Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
2022-08-12 00:17:39 +00:00
/**
* The decoder might make use of the ProgressFrame API.
*/
avcodec/codec_internal: Remove FF_CODEC_CAP_ALLOCATE_PROGRESS Before commit f025b8e110b36c1cdb4fb56c4cd57aeca1767b5b, every frame-threaded decoder used ThreadFrames, even when they did not have any inter-frame dependencies at all. In order to distinguish those decoders that need the AVBuffer for progress communication from those that do not (to avoid the allocation for the latter), the former decoders were marked with the FF_CODEC_CAP_ALLOCATE_PROGRESS internal codec cap. Yet distinguishing these two can be done in a more natural way: Don't use ThreadFrames when not needed and split ff_thread_get_buffer() into a core function that calls the user's get_buffer2 callback and a wrapper around it that also allocates the progress AVBuffer. This has been done in 02220b88fc38ef9dd4f2d519f5d3e4151258b60c and since that commit the ALLOCATE_PROGRESS cap was nearly redundant. The only exception was WebP and VP8. WebP can contain VP8 and uses the VP8 decoder directly (i.e. they share the same AVCodecContext). Both decoders are frame-threaded and VP8 has inter-frame dependencies (in general, not in valid WebP) and therefore the ALLOCATE_PROGRESS cap. In order to avoid allocating progress in case of a frame-threaded WebP decoder the cap and the check for the cap has been kept in place. Yet now the VP8 decoder has been switched to use ProgressFrames and therefore there is just no reason any more for this check and the cap. This commit therefore removes both. Also change the value of FF_CODEC_CAP_USES_PROGRESSFRAMES to leave no gaps. Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
2023-09-18 23:43:29 +00:00
#define FF_CODEC_CAP_USES_PROGRESSFRAMES (1 << 6)
/**
* Codec handles avctx->thread_count == 0 (auto) internally.
*/
#define FF_CODEC_CAP_AUTO_THREADS (1 << 7)
/**
* Codec handles output frame properties internally instead of letting the
* internal logic derive them from AVCodecInternal.last_pkt_props.
*/
#define FF_CODEC_CAP_SETS_FRAME_PROPS (1 << 8)
/**
* Codec supports embedded ICC profiles (AV_FRAME_DATA_ICC_PROFILE).
*/
#define FF_CODEC_CAP_ICC_PROFILES (1 << 9)
/**
* The encoder has AV_CODEC_CAP_DELAY set, but does not actually have delay - it
* only wants to be flushed at the end to update some context variables (e.g.
* 2pass stats) or produce a trailing packet. Besides that it immediately
* produces exactly one output packet per each input frame, just as no-delay
* encoders do.
*/
#define FF_CODEC_CAP_EOF_FLUSH (1 << 10)
/**
* FFCodec.codec_tags termination value
*/
#define FF_CODEC_TAGS_END -1
typedef struct FFCodecDefault {
const char *key;
const char *value;
} FFCodecDefault;
struct AVCodecContext;
struct AVSubtitle;
struct AVPacket;
enum FFCodecType {
/* The codec is a decoder using the decode callback;
* audio and video codecs only. */
FF_CODEC_CB_TYPE_DECODE,
/* The codec is a decoder using the decode_sub callback;
* subtitle codecs only. */
FF_CODEC_CB_TYPE_DECODE_SUB,
/* The codec is a decoder using the receive_frame callback;
* audio and video codecs only. */
FF_CODEC_CB_TYPE_RECEIVE_FRAME,
/* The codec is an encoder using the encode callback;
* audio and video codecs only. */
FF_CODEC_CB_TYPE_ENCODE,
/* The codec is an encoder using the encode_sub callback;
* subtitle codecs only. */
FF_CODEC_CB_TYPE_ENCODE_SUB,
/* The codec is an encoder using the receive_packet callback;
* audio and video codecs only. */
FF_CODEC_CB_TYPE_RECEIVE_PACKET,
};
typedef struct FFCodec {
/**
* The public AVCodec. See codec.h for it.
*/
AVCodec p;
/**
* Internal codec capabilities FF_CODEC_CAP_*.
*/
unsigned caps_internal:29;
/**
* This field determines the type of the codec (decoder/encoder)
* and also the exact callback cb implemented by the codec.
* cb_type uses enum FFCodecType values.
*/
unsigned cb_type:3;
int priv_data_size;
/**
* @name Frame-level threading support functions
* @{
*/
/**
* Copy necessary context variables from a previous thread context to the current one.
* If not defined, the next thread will start automatically; otherwise, the codec
* must call ff_thread_finish_setup().
*
* dst and src will (rarely) point to the same context, in which case memcpy should be skipped.
*/
int (*update_thread_context)(struct AVCodecContext *dst, const struct AVCodecContext *src);
/**
* Copy variables back to the user-facing context
*/
int (*update_thread_context_for_user)(struct AVCodecContext *dst, const struct AVCodecContext *src);
/** @} */
/**
* Private codec-specific defaults.
*/
const FFCodecDefault *defaults;
/**
* Initialize codec static data, called from av_codec_iterate().
*
* This is not intended for time consuming operations as it is
* run for every codec regardless of that codec being used.
*/
void (*init_static_data)(struct FFCodec *codec);
int (*init)(struct AVCodecContext *);
union {
/**
* Decode to an AVFrame.
* cb is in this state if cb_type is FF_CODEC_CB_TYPE_DECODE.
*
* @param avctx codec context
* @param[out] frame AVFrame for output
* @param[out] got_frame_ptr decoder sets to 0 or 1 to indicate that
* a non-empty frame was returned in frame.
* @param[in] avpkt AVPacket containing the data to be decoded
* @return amount of bytes read from the packet on success,
* negative error code on failure
*/
int (*decode)(struct AVCodecContext *avctx, struct AVFrame *frame,
int *got_frame_ptr, struct AVPacket *avpkt);
/**
* Decode subtitle data to an AVSubtitle.
* cb is in this state if cb_type is FF_CODEC_CB_TYPE_DECODE_SUB.
*
* Apart from that this is like the decode callback.
*/
int (*decode_sub)(struct AVCodecContext *avctx, struct AVSubtitle *sub,
int *got_frame_ptr, const struct AVPacket *avpkt);
/**
* Decode API with decoupled packet/frame dataflow.
* cb is in this state if cb_type is FF_CODEC_CB_TYPE_RECEIVE_FRAME.
*
* This function is called to get one output frame. It should call
* ff_decode_get_packet() to obtain input data.
*/
int (*receive_frame)(struct AVCodecContext *avctx, struct AVFrame *frame);
/**
* Encode data to an AVPacket.
* cb is in this state if cb_type is FF_CODEC_CB_TYPE_ENCODE
*
* @param avctx codec context
* @param[out] avpkt output AVPacket
* @param[in] frame AVFrame containing the input to be encoded
* @param[out] got_packet_ptr encoder sets to 0 or 1 to indicate that a
* non-empty packet was returned in avpkt.
* @return 0 on success, negative error code on failure
*/
int (*encode)(struct AVCodecContext *avctx, struct AVPacket *avpkt,
const struct AVFrame *frame, int *got_packet_ptr);
/**
* Encode subtitles to a raw buffer.
* cb is in this state if cb_type is FF_CODEC_CB_TYPE_ENCODE_SUB.
*/
int (*encode_sub)(struct AVCodecContext *avctx, uint8_t *buf,
int buf_size, const struct AVSubtitle *sub);
/**
* Encode API with decoupled frame/packet dataflow.
* cb is in this state if cb_type is FF_CODEC_CB_TYPE_RECEIVE_PACKET.
*
* This function is called to get one output packet.
* It should call ff_encode_get_frame() to obtain input data.
*/
int (*receive_packet)(struct AVCodecContext *avctx, struct AVPacket *avpkt);
} cb;
int (*close)(struct AVCodecContext *);
/**
* Flush buffers.
* Will be called when seeking
*/
void (*flush)(struct AVCodecContext *);
/**
* Decoding only, a comma-separated list of bitstream filters to apply to
* packets before decoding.
*/
const char *bsfs;
/**
* Array of pointers to hardware configurations supported by the codec,
* or NULL if no hardware supported. The array is terminated by a NULL
* pointer.
*
* The user can only access this field via avcodec_get_hw_config().
*/
const struct AVCodecHWConfigInternal *const *hw_configs;
/**
* List of supported codec_tags, terminated by FF_CODEC_TAGS_END.
*/
const uint32_t *codec_tags;
} FFCodec;
#if CONFIG_SMALL
#define CODEC_LONG_NAME(str) .p.long_name = NULL
#else
#define CODEC_LONG_NAME(str) .p.long_name = str
#endif
#if HAVE_THREADS
#define UPDATE_THREAD_CONTEXT(func) \
.update_thread_context = (func)
#define UPDATE_THREAD_CONTEXT_FOR_USER(func) \
.update_thread_context_for_user = (func)
#else
#define UPDATE_THREAD_CONTEXT(func) \
.update_thread_context = NULL
#define UPDATE_THREAD_CONTEXT_FOR_USER(func) \
.update_thread_context_for_user = NULL
#endif
#define FF_CODEC_DECODE_CB(func) \
.cb_type = FF_CODEC_CB_TYPE_DECODE, \
.cb.decode = (func)
#define FF_CODEC_DECODE_SUB_CB(func) \
.cb_type = FF_CODEC_CB_TYPE_DECODE_SUB, \
.cb.decode_sub = (func)
#define FF_CODEC_RECEIVE_FRAME_CB(func) \
.cb_type = FF_CODEC_CB_TYPE_RECEIVE_FRAME, \
.cb.receive_frame = (func)
#define FF_CODEC_ENCODE_CB(func) \
.cb_type = FF_CODEC_CB_TYPE_ENCODE, \
.cb.encode = (func)
#define FF_CODEC_ENCODE_SUB_CB(func) \
.cb_type = FF_CODEC_CB_TYPE_ENCODE_SUB, \
.cb.encode_sub = (func)
#define FF_CODEC_RECEIVE_PACKET_CB(func) \
.cb_type = FF_CODEC_CB_TYPE_RECEIVE_PACKET, \
.cb.receive_packet = (func)
static av_always_inline const FFCodec *ffcodec(const AVCodec *codec)
{
return (const FFCodec*)codec;
}
#endif /* AVCODEC_CODEC_INTERNAL_H */