mirrors/gstreamer
mirrors/gstreamer
1
1
Fork 0
mirror of https://gitlab.freedesktop.org/gstreamer/gstreamer.git synced 2024-09-20 02:50:16 +00:00
Code Issues 1 Projects Releases Wiki Activity
gstreamer/gst-libs/gst/vaapi/gstvaapidecoder_objects.c

474 lines
14 KiB
C
Raw Normal View History

decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
/*
* gstvaapidecoder_objects.c - VA decoder objects helpers
*
* Copyright (C) 2010-2011 Splitted-Desktop Systems
legal: fix year for some copyright notices (2013).
2013-01-29 13:14:45 +00:00
* Copyright (C) 2011-2013 Intel Corporation
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
*
* This library 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.
*
* This library 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 this library; if not, write to the Free
* Software Foundation, Inc., 51 Franklin Street, Fifth Floor,
* Boston, MA 02110-1301 USA
*/
Add header for system-dependent definitions.
2012-01-30 17:12:59 +00:00
#include "sysdeps.h"
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
#include <string.h>
#include <gst/vaapi/gstvaapicontext.h>
#include "gstvaapidecoder_objects.h"
#include "gstvaapidecoder_priv.h"
decoder: use new GstVaapiSurfaceProxy utility functions. Use new GstVaapiSurfaceProxy internal helper functions to propagate the necessary GstVideoCodecFrame flags to vaapidecode (GStreamer 0.10). Also make GstVaapiDecoder push_frame() operate similarly to drop_frame(). i.e. increase the GstVideoCodecFrame reference count in push_frame rather than gst_vaapi_picture_output().
2013-04-16 16:56:24 +00:00
#include "gstvaapisurfaceproxy_priv.h"
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
#include "gstvaapicompat.h"
#include "gstvaapiutils.h"
#define DEBUG 1
#include "gstvaapidebug.h"
#define GET_DECODER(obj) GST_VAAPI_DECODER_CAST((obj)->parent_instance.codec)
libs: use GstVaapiMiniObject for video decoders. Port GstVaapiDecoder and GstVaapiDecoder{MPEG2,MPEG4,JPEG,H264,VC1} to GstVaapiMiniObject. Add gst_vaapi_decoder_set_codec_state_changed_func() helper function to let the user add a callback to a function triggered whenever the codec state (e.g. caps) changes.
2013-05-06 12:07:17 +00:00
#define GET_CONTEXT(obj) GET_DECODER(obj)->context
#define GET_VA_DISPLAY(obj) GET_DECODER(obj)->va_display
#define GET_VA_CONTEXT(obj) GET_DECODER(obj)->va_context
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
decoder: dispose GstVideoCodecFrame earlier. Once the picture was output, it is no longer necessary to keep an extra reference to the underlying GstVideoCodecFrame. So, we can release it earlier, and maybe subsequently release the associate surface proxy earlier.
2013-07-15 15:43:34 +00:00
static inline void
gst_video_codec_frame_clear(GstVideoCodecFrame **frame_ptr)
{
if (!*frame_ptr)
return;
gst_video_codec_frame_unref(*frame_ptr);
*frame_ptr = NULL;
}
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
/* ------------------------------------------------------------------------- */
/* --- Pictures --- */
/* ------------------------------------------------------------------------- */
Port codec objects to GstVaapiMiniObject.
2012-12-03 10:19:08 +00:00
GST_VAAPI_CODEC_DEFINE_TYPE(GstVaapiPicture, gst_vaapi_picture);
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
decoder: allow pictures to be cloned for field decoding. Add gst_vaapi_picture_new_field() function that clones a picture, while preserving the parent picture surface. i.e. the surface proxy reference count is increased and other fields copied as is. Besides, the picture is reset into a "non-output" mode.
2012-03-28 14:05:58 +00:00
enum {
GST_VAAPI_CREATE_PICTURE_FLAG_CLONE = 1 << 0,
GST_VAAPI_CREATE_PICTURE_FLAG_FIELD = 1 << 1,
};
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
static void
destroy_slice_cb(gpointer data, gpointer user_data)
{
Port codec objects to GstVaapiMiniObject.
2012-12-03 10:19:08 +00:00
GstVaapiMiniObject * const object = data;
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
Port codec objects to GstVaapiMiniObject.
2012-12-03 10:19:08 +00:00
gst_vaapi_mini_object_unref(object);
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
}
Port codec objects to GstVaapiMiniObject.
2012-12-03 10:19:08 +00:00
void
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
gst_vaapi_picture_destroy(GstVaapiPicture *picture)
{
if (picture->slices) {
g_ptr_array_foreach(picture->slices, destroy_slice_cb, NULL);
g_ptr_array_free(picture->slices, TRUE);
picture->slices = NULL;
}
Port codec objects to GstVaapiMiniObject.
2012-12-03 10:19:08 +00:00
gst_vaapi_mini_object_replace((GstVaapiMiniObject **)&picture->iq_matrix,
NULL);
gst_vaapi_mini_object_replace((GstVaapiMiniObject **)&picture->huf_table,
NULL);
gst_vaapi_mini_object_replace((GstVaapiMiniObject **)&picture->bitplane,
NULL);
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
decoder: retain proxy surface until the GstVaapiPicture is destroyed. Keep a valid reference to the proxy in GstVaapiPicture so that frames marked as "used for reference" could be kept during the lifetime of the picture. i.e. don't release them too soon as they could be re-used right away.
2012-01-26 14:28:42 +00:00
if (picture->proxy) {
surfaceproxy: port to GstVaapiMiniObject. GstVaapiSurfaceProxy does not use any particular functionality from GObject. Actually, it only needs a basic object type with reference counting. This is an API and ABI change.
2012-12-03 12:46:28 +00:00
gst_vaapi_surface_proxy_unref(picture->proxy);
decoder: retain proxy surface until the GstVaapiPicture is destroyed. Keep a valid reference to the proxy in GstVaapiPicture so that frames marked as "used for reference" could be kept during the lifetime of the picture. i.e. don't release them too soon as they could be re-used right away.
2012-01-26 14:28:42 +00:00
picture->proxy = NULL;
}
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
picture->surface_id = VA_INVALID_ID;
picture->surface = NULL;
vaapi_destroy_buffer(GET_VA_DISPLAY(picture), &picture->param_id);
picture->param = NULL;
decoder: maintain decoded frames as GstVideoCodecFrame objects. Maintain decoded surfaces as GstVideoCodecFrame objects instead of GstVaapiSurfaceProxy objects. The latter will tend to be reduced to the strict minimum: a context and a surface.
2012-12-13 14:34:10 +00:00
decoder: dispose GstVideoCodecFrame earlier. Once the picture was output, it is no longer necessary to keep an extra reference to the underlying GstVideoCodecFrame. So, we can release it earlier, and maybe subsequently release the associate surface proxy earlier.
2013-07-15 15:43:34 +00:00
gst_video_codec_frame_clear(&picture->frame);
decoder: fix memory leak when processing interlaced pictures. Fix memory leak when processing interlaced pictures and that occurs because the first field, represented as a GstVideoCodecFrame, never gets released. i.e. when the picture is completed, this is generally the case when the second field is successfully decoded, we need to propagate the GstVideoCodecFrame of the first field to the original GstVideoDecoder so that it could reclaim memory. Otherwise, we keep accumulating the first fields into GstVideoDecoder private frames list until the end-of-stream is reached. The frames are eventually released there, but too late, i.e. too much memory may have been consumed. https://bugzilla.gnome.org/show_bug.cgi?id=701257
2013-06-11 13:11:34 +00:00
gst_vaapi_picture_replace(&picture->parent_picture, NULL);
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
}
Port codec objects to GstVaapiMiniObject.
2012-12-03 10:19:08 +00:00
gboolean
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
gst_vaapi_picture_create(
GstVaapiPicture *picture,
const GstVaapiCodecObjectConstructorArgs *args
)
{
decoder: optimize slice data buffers initialization. VA drivers may have a faster means to transfer user buffers to GPU buffers than using memcpy(). In particular, on Intel Gen graphics, we can use pwrite(). This provides for faster upload of bitstream and can help higher bitrates. vaapi_create_buffer() helper function was also updated to allow for un-mapped buffers and pre-initialized data for buffers.
2012-01-30 09:15:32 +00:00
gboolean success;
decoder: allow pictures to be cloned for field decoding. Add gst_vaapi_picture_new_field() function that clones a picture, while preserving the parent picture surface. i.e. the surface proxy reference count is increased and other fields copied as is. Besides, the picture is reset into a "non-output" mode.
2012-03-28 14:05:58 +00:00
if (args->flags & GST_VAAPI_CREATE_PICTURE_FLAG_CLONE) {
GstVaapiPicture * const parent_picture = GST_VAAPI_PICTURE(args->data);
decoder: fix memory leak when processing interlaced pictures. Fix memory leak when processing interlaced pictures and that occurs because the first field, represented as a GstVideoCodecFrame, never gets released. i.e. when the picture is completed, this is generally the case when the second field is successfully decoded, we need to propagate the GstVideoCodecFrame of the first field to the original GstVideoDecoder so that it could reclaim memory. Otherwise, we keep accumulating the first fields into GstVideoDecoder private frames list until the end-of-stream is reached. The frames are eventually released there, but too late, i.e. too much memory may have been consumed. https://bugzilla.gnome.org/show_bug.cgi?id=701257
2013-06-11 13:11:34 +00:00
picture->parent_picture = gst_vaapi_picture_ref(parent_picture);
surfaceproxy: port to GstVaapiMiniObject. GstVaapiSurfaceProxy does not use any particular functionality from GObject. Actually, it only needs a basic object type with reference counting. This is an API and ABI change.
2012-12-03 12:46:28 +00:00
picture->proxy = gst_vaapi_surface_proxy_ref(parent_picture->proxy);
decoder: allow pictures to be cloned for field decoding. Add gst_vaapi_picture_new_field() function that clones a picture, while preserving the parent picture surface. i.e. the surface proxy reference count is increased and other fields copied as is. Besides, the picture is reset into a "non-output" mode.
2012-03-28 14:05:58 +00:00
picture->type = parent_picture->type;
picture->pts = parent_picture->pts;
decoder: use POC to maintain the DPB. Introduce a POC field in GstVaapiPicture so that to store simpler sequential numbers. A signed 32-bit integer should be enough for 1 year of continuous video streaming at 60 Hz. Use this new POC value to maintain the DPB, instead of 64-bit timestamps. This also aligns with H.264 that will be migrated to GstVaapiDpb infrastructure.
2012-03-30 15:03:28 +00:00
picture->poc = parent_picture->poc;
decoder: allow pictures to be cloned for field decoding. Add gst_vaapi_picture_new_field() function that clones a picture, while preserving the parent picture surface. i.e. the surface proxy reference count is increased and other fields copied as is. Besides, the picture is reset into a "non-output" mode.
2012-03-28 14:05:58 +00:00
// Copy all picture flags but "output"
GST_VAAPI_PICTURE_FLAG_SET(
picture,
GST_VAAPI_PICTURE_FLAGS(parent_picture) &
(GST_VAAPI_PICTURE_FLAG_SKIPPED |
GST_VAAPI_PICTURE_FLAG_REFERENCE |
GST_VAAPI_PICTURE_FLAG_INTERLACED |
decoder: add first-field flag to picture. Add first-field (FF) flag to GstVaapiPicture, thus not requiring is_first_field member in each decoder. Rather, when a GstVaapiPicture is created, it is considered as the first field. Any subsequent allocated field will become the second field.
2012-03-28 15:50:28 +00:00
GST_VAAPI_PICTURE_FLAG_FF |
decoder: allow pictures to be cloned for field decoding. Add gst_vaapi_picture_new_field() function that clones a picture, while preserving the parent picture surface. i.e. the surface proxy reference count is increased and other fields copied as is. Besides, the picture is reset into a "non-output" mode.
2012-03-28 14:05:58 +00:00
GST_VAAPI_PICTURE_FLAG_TFF)
);
picture->structure = parent_picture->structure;
if ((args->flags & GST_VAAPI_CREATE_PICTURE_FLAG_FIELD) &&
GST_VAAPI_PICTURE_IS_INTERLACED(picture)) {
switch (picture->structure) {
case GST_VAAPI_PICTURE_STRUCTURE_TOP_FIELD:
picture->structure = GST_VAAPI_PICTURE_STRUCTURE_BOTTOM_FIELD;
break;
case GST_VAAPI_PICTURE_STRUCTURE_BOTTOM_FIELD:
picture->structure = GST_VAAPI_PICTURE_STRUCTURE_TOP_FIELD;
break;
}
decoder: add first-field flag to picture. Add first-field (FF) flag to GstVaapiPicture, thus not requiring is_first_field member in each decoder. Rather, when a GstVaapiPicture is created, it is considered as the first field. Any subsequent allocated field will become the second field.
2012-03-28 15:50:28 +00:00
GST_VAAPI_PICTURE_FLAG_UNSET(picture, GST_VAAPI_PICTURE_FLAG_FF);
decoder: allow pictures to be cloned for field decoding. Add gst_vaapi_picture_new_field() function that clones a picture, while preserving the parent picture surface. i.e. the surface proxy reference count is increased and other fields copied as is. Besides, the picture is reset into a "non-output" mode.
2012-03-28 14:05:58 +00:00
}
decoder: add support for video cropping. Add gst_vaapi_picture_set_crop_rect() helper function to copy the video cropping information from raw bitstreams to each picture being decoded. Also add helper function to surface proxy to propagate that information outside of libgstvaapi. e.g. plug-in elements or standalone applications. Signed-off-by: Sreerenj Balachandran <sreerenj.balachandran@intel.com> Signed-off-by: Gwenole Beauchesne <gwenole.beauchesne@intel.com>
2013-02-15 16:42:12 +00:00
if (parent_picture->has_crop_rect) {
picture->has_crop_rect = TRUE;
picture->crop_rect = parent_picture->crop_rect;
}
decoder: allow pictures to be cloned for field decoding. Add gst_vaapi_picture_new_field() function that clones a picture, while preserving the parent picture surface. i.e. the surface proxy reference count is increased and other fields copied as is. Besides, the picture is reset into a "non-output" mode.
2012-03-28 14:05:58 +00:00
}
else {
Port codec objects to GstVaapiMiniObject.
2012-12-03 10:19:08 +00:00
picture->type = GST_VAAPI_PICTURE_TYPE_NONE;
picture->pts = GST_CLOCK_TIME_NONE;
decoder: allow pictures to be cloned for field decoding. Add gst_vaapi_picture_new_field() function that clones a picture, while preserving the parent picture surface. i.e. the surface proxy reference count is increased and other fields copied as is. Besides, the picture is reset into a "non-output" mode.
2012-03-28 14:05:58 +00:00
picture->proxy =
decoder: simplify acquisition/release of spare surface. Introduce gst_vaapi_surface_proxy_new_from_pool() to allocate a new surface proxy from the context surface pool. This change also makes sure to retain the parent surface pool in the proxy. Besides, it was also totally useless to attach/detach parent context to VA surface each time we acquire/release it. Since the whole context owns all associated VA surfaces, we can mark this as such only once and for all.
2013-04-03 09:37:44 +00:00
gst_vaapi_context_get_surface_proxy(GET_CONTEXT(picture));
decoder: allow pictures to be cloned for field decoding. Add gst_vaapi_picture_new_field() function that clones a picture, while preserving the parent picture surface. i.e. the surface proxy reference count is increased and other fields copied as is. Besides, the picture is reset into a "non-output" mode.
2012-03-28 14:05:58 +00:00
if (!picture->proxy)
return FALSE;
picture->structure = GST_VAAPI_PICTURE_STRUCTURE_FRAME;
decoder: add first-field flag to picture. Add first-field (FF) flag to GstVaapiPicture, thus not requiring is_first_field member in each decoder. Rather, when a GstVaapiPicture is created, it is considered as the first field. Any subsequent allocated field will become the second field.
2012-03-28 15:50:28 +00:00
GST_VAAPI_PICTURE_FLAG_SET(picture, GST_VAAPI_PICTURE_FLAG_FF);
decoder: allow pictures to be cloned for field decoding. Add gst_vaapi_picture_new_field() function that clones a picture, while preserving the parent picture surface. i.e. the surface proxy reference count is increased and other fields copied as is. Besides, the picture is reset into a "non-output" mode.
2012-03-28 14:05:58 +00:00
}
decoder: use new GstVaapiSurfaceProxy utility functions. Use new GstVaapiSurfaceProxy internal helper functions to propagate the necessary GstVideoCodecFrame flags to vaapidecode (GStreamer 0.10). Also make GstVaapiDecoder push_frame() operate similarly to drop_frame(). i.e. increase the GstVideoCodecFrame reference count in push_frame rather than gst_vaapi_picture_output().
2013-04-16 16:56:24 +00:00
picture->surface = GST_VAAPI_SURFACE_PROXY_SURFACE(picture->proxy);
picture->surface_id = GST_VAAPI_SURFACE_PROXY_SURFACE_ID(picture->proxy);
decoder: allocate proxy surface earlier. This simplifies gst_vaapi_picture_output() to only update the presentation timestamp and submit the proxy to the decoder for output.
2012-01-31 10:34:17 +00:00
Port codec objects to GstVaapiMiniObject.
2012-12-03 10:19:08 +00:00
picture->param_id = VA_INVALID_ID;
decoder: optimize slice data buffers initialization. VA drivers may have a faster means to transfer user buffers to GPU buffers than using memcpy(). In particular, on Intel Gen graphics, we can use pwrite(). This provides for faster upload of bitstream and can help higher bitrates. vaapi_create_buffer() helper function was also updated to allow for un-mapped buffers and pre-initialized data for buffers.
2012-01-30 09:15:32 +00:00
success = vaapi_create_buffer(
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
GET_VA_DISPLAY(picture),
GET_VA_CONTEXT(picture),
VAPictureParameterBufferType,
args->param_size,
decoder: optimize slice data buffers initialization. VA drivers may have a faster means to transfer user buffers to GPU buffers than using memcpy(). In particular, on Intel Gen graphics, we can use pwrite(). This provides for faster upload of bitstream and can help higher bitrates. vaapi_create_buffer() helper function was also updated to allow for un-mapped buffers and pre-initialized data for buffers.
2012-01-30 09:15:32 +00:00
args->param,
&picture->param_id,
&picture->param
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
);
decoder: optimize slice data buffers initialization. VA drivers may have a faster means to transfer user buffers to GPU buffers than using memcpy(). In particular, on Intel Gen graphics, we can use pwrite(). This provides for faster upload of bitstream and can help higher bitrates. vaapi_create_buffer() helper function was also updated to allow for un-mapped buffers and pre-initialized data for buffers.
2012-01-30 09:15:32 +00:00
if (!success)
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
return FALSE;
decoder: allow pictures to be cloned for field decoding. Add gst_vaapi_picture_new_field() function that clones a picture, while preserving the parent picture surface. i.e. the surface proxy reference count is increased and other fields copied as is. Besides, the picture is reset into a "non-output" mode.
2012-03-28 14:05:58 +00:00
picture->param_size = args->param_size;
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
picture->slices = g_ptr_array_new();
if (!picture->slices)
return FALSE;
decoder: expose new parse/decode API. Introduce new decoding process whereby a GstVideoCodecFrame is created first. Next, input stream buffers are accumulated into a GstAdapter, that is then passed to the _parse() function. The GstVaapiDecoder object accumulates all parsed units and when a complete frame or field is detected, that GstVideoCodecFrame is passed to the _decode() function. Ultimately, the caller receives a GstVaapiSurfaceProxy if decoding process was successful.
2012-11-29 14:06:00 +00:00
decoder: maintain decoded frames as GstVideoCodecFrame objects. Maintain decoded surfaces as GstVideoCodecFrame objects instead of GstVaapiSurfaceProxy objects. The latter will tend to be reduced to the strict minimum: a context and a surface.
2012-12-13 14:34:10 +00:00
picture->frame = gst_video_codec_frame_ref(
GST_VAAPI_DECODER_CODEC_FRAME(GET_DECODER(picture)));
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
return TRUE;
}
GstVaapiPicture *
gst_vaapi_picture_new(
GstVaapiDecoder *decoder,
gconstpointer param,
guint param_size
)
{
GstVaapiCodecObject *object;
object = gst_vaapi_codec_object_new(
Port codec objects to GstVaapiMiniObject.
2012-12-03 10:19:08 +00:00
&GstVaapiPictureClass,
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
GST_VAAPI_CODEC_BASE(decoder),
param, param_size,
Port codec objects to GstVaapiMiniObject.
2012-12-03 10:19:08 +00:00
NULL, 0,
0
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
);
if (!object)
return NULL;
return GST_VAAPI_PICTURE_CAST(object);
}
decoder: allow pictures to be cloned for field decoding. Add gst_vaapi_picture_new_field() function that clones a picture, while preserving the parent picture surface. i.e. the surface proxy reference count is increased and other fields copied as is. Besides, the picture is reset into a "non-output" mode.
2012-03-28 14:05:58 +00:00
GstVaapiPicture *
gst_vaapi_picture_new_field(GstVaapiPicture *picture)
{
Port codec objects to GstVaapiMiniObject.
2012-12-03 10:19:08 +00:00
GstVaapiDecoder * const decoder = GET_DECODER(picture);
GstVaapiCodecObject *object;
decoder: allow pictures to be cloned for field decoding. Add gst_vaapi_picture_new_field() function that clones a picture, while preserving the parent picture surface. i.e. the surface proxy reference count is increased and other fields copied as is. Besides, the picture is reset into a "non-output" mode.
2012-03-28 14:05:58 +00:00
Port codec objects to GstVaapiMiniObject.
2012-12-03 10:19:08 +00:00
object = gst_vaapi_codec_object_new(
gst_vaapi_codec_object_get_class(&picture->parent_instance),
GST_VAAPI_CODEC_BASE(decoder),
NULL, picture->param_size,
picture, 0,
(GST_VAAPI_CREATE_PICTURE_FLAG_CLONE|
GST_VAAPI_CREATE_PICTURE_FLAG_FIELD)
);
if (!object)
decoder: allow pictures to be cloned for field decoding. Add gst_vaapi_picture_new_field() function that clones a picture, while preserving the parent picture surface. i.e. the surface proxy reference count is increased and other fields copied as is. Besides, the picture is reset into a "non-output" mode.
2012-03-28 14:05:58 +00:00
return NULL;
Port codec objects to GstVaapiMiniObject.
2012-12-03 10:19:08 +00:00
return GST_VAAPI_PICTURE_CAST(object);
decoder: allow pictures to be cloned for field decoding. Add gst_vaapi_picture_new_field() function that clones a picture, while preserving the parent picture surface. i.e. the surface proxy reference count is increased and other fields copied as is. Besides, the picture is reset into a "non-output" mode.
2012-03-28 14:05:58 +00:00
}
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
void
gst_vaapi_picture_add_slice(GstVaapiPicture *picture, GstVaapiSlice *slice)
{
g_return_if_fail(GST_VAAPI_IS_PICTURE(picture));
g_return_if_fail(GST_VAAPI_IS_SLICE(slice));
g_ptr_array_add(picture->slices, slice);
}
decoder: fix double buffer free with some VA drivers. vaRenderPicture() implicitly disposes VA buffers. Some VA drivers would push the VA buffer object into a list of free buffers to be re-used. However, reference pictures (and data) that was kept would explicitly release the VA buffer object later on, thus possibly destroying a valid (re-used) object. Besides, some other VA drivers don't support correctly the vaRenderPicture() semantics for VA buffers disposal and would leak memory if there is no explicit vaDestroyBuffer(). The temporary workaround is to explcitily destroy VA buffers right after vaRenderPicture(). All VA drivers need to be aligned.
2012-02-13 09:10:35 +00:00
static gboolean
do_decode(VADisplay dpy, VAContextID ctx, VABufferID *buf_id, void **buf_ptr)
{
VAStatus status;
vaapi_unmap_buffer(dpy, *buf_id, buf_ptr);
status = vaRenderPicture(dpy, ctx, buf_id, 1);
if (!vaapi_check_status(status, "vaRenderPicture()"))
return FALSE;
/* XXX: vaRenderPicture() is meant to destroy the VA buffer implicitly */
vaapi_destroy_buffer(dpy, buf_id);
return TRUE;
}
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
gboolean
gst_vaapi_picture_decode(GstVaapiPicture *picture)
{
GstVaapiIqMatrix *iq_matrix;
GstVaapiBitPlane *bitplane;
Add initial JPEG decoder. Signed-off-by: Gwenole Beauchesne <gwenole.beauchesne@intel.com>
2012-02-09 16:21:04 +00:00
GstVaapiHuffmanTable *huf_table;
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
VADisplay va_display;
VAContextID va_context;
VAStatus status;
decoder: fix double buffer free with some VA drivers. vaRenderPicture() implicitly disposes VA buffers. Some VA drivers would push the VA buffer object into a list of free buffers to be re-used. However, reference pictures (and data) that was kept would explicitly release the VA buffer object later on, thus possibly destroying a valid (re-used) object. Besides, some other VA drivers don't support correctly the vaRenderPicture() semantics for VA buffers disposal and would leak memory if there is no explicit vaDestroyBuffer(). The temporary workaround is to explcitily destroy VA buffers right after vaRenderPicture(). All VA drivers need to be aligned.
2012-02-13 09:10:35 +00:00
guint i;
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
g_return_val_if_fail(GST_VAAPI_IS_PICTURE(picture), FALSE);
va_display = GET_VA_DISPLAY(picture);
va_context = GET_VA_CONTEXT(picture);
GST_DEBUG("decode picture 0x%08x", picture->surface_id);
status = vaBeginPicture(va_display, va_context, picture->surface_id);
if (!vaapi_check_status(status, "vaBeginPicture()"))
return FALSE;
decoder: fix double buffer free with some VA drivers. vaRenderPicture() implicitly disposes VA buffers. Some VA drivers would push the VA buffer object into a list of free buffers to be re-used. However, reference pictures (and data) that was kept would explicitly release the VA buffer object later on, thus possibly destroying a valid (re-used) object. Besides, some other VA drivers don't support correctly the vaRenderPicture() semantics for VA buffers disposal and would leak memory if there is no explicit vaDestroyBuffer(). The temporary workaround is to explcitily destroy VA buffers right after vaRenderPicture(). All VA drivers need to be aligned.
2012-02-13 09:10:35 +00:00
if (!do_decode(va_display, va_context, &picture->param_id, &picture->param))
return FALSE;
iq_matrix = picture->iq_matrix;
if (iq_matrix && !do_decode(va_display, va_context,
&iq_matrix->param_id, &iq_matrix->param))
return FALSE;
bitplane = picture->bitplane;
if (bitplane && !do_decode(va_display, va_context,
&bitplane->data_id, (void **)&bitplane->data))
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
return FALSE;
Add initial JPEG decoder. Signed-off-by: Gwenole Beauchesne <gwenole.beauchesne@intel.com>
2012-02-09 16:21:04 +00:00
huf_table = picture->huf_table;
if (huf_table && !do_decode(va_display, va_context,
&huf_table->param_id,
(void **)&huf_table->param))
return FALSE;
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
for (i = 0; i < picture->slices->len; i++) {
GstVaapiSlice * const slice = g_ptr_array_index(picture->slices, i);
decoder: fix double buffer free with some VA drivers. vaRenderPicture() implicitly disposes VA buffers. Some VA drivers would push the VA buffer object into a list of free buffers to be re-used. However, reference pictures (and data) that was kept would explicitly release the VA buffer object later on, thus possibly destroying a valid (re-used) object. Besides, some other VA drivers don't support correctly the vaRenderPicture() semantics for VA buffers disposal and would leak memory if there is no explicit vaDestroyBuffer(). The temporary workaround is to explcitily destroy VA buffers right after vaRenderPicture(). All VA drivers need to be aligned.
2012-02-13 09:10:35 +00:00
VABufferID va_buffers[2];
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
jpeg: add support for multiscan images. Add support for images with multiple scans per frame. The Huffman table can be updated before SOS, and thus possibly requiring multiple uploads of Huffman tables to the VA driver. So, the latter must be able to cope with multiple VA buffers of type 'huffman-table' and with the correct sequential order.
2013-09-23 14:49:41 +00:00
huf_table = slice->huf_table;
if (huf_table && !do_decode(va_display, va_context,
&huf_table->param_id, (void **)&huf_table->param))
return FALSE;
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
vaapi_unmap_buffer(va_display, slice->param_id, NULL);
va_buffers[0] = slice->param_id;
va_buffers[1] = slice->data_id;
decoder: fix double buffer free with some VA drivers. vaRenderPicture() implicitly disposes VA buffers. Some VA drivers would push the VA buffer object into a list of free buffers to be re-used. However, reference pictures (and data) that was kept would explicitly release the VA buffer object later on, thus possibly destroying a valid (re-used) object. Besides, some other VA drivers don't support correctly the vaRenderPicture() semantics for VA buffers disposal and would leak memory if there is no explicit vaDestroyBuffer(). The temporary workaround is to explcitily destroy VA buffers right after vaRenderPicture(). All VA drivers need to be aligned.
2012-02-13 09:10:35 +00:00
status = vaRenderPicture(va_display, va_context, va_buffers, 2);
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
if (!vaapi_check_status(status, "vaRenderPicture()"))
return FALSE;
decoder: fix double buffer free with some VA drivers. vaRenderPicture() implicitly disposes VA buffers. Some VA drivers would push the VA buffer object into a list of free buffers to be re-used. However, reference pictures (and data) that was kept would explicitly release the VA buffer object later on, thus possibly destroying a valid (re-used) object. Besides, some other VA drivers don't support correctly the vaRenderPicture() semantics for VA buffers disposal and would leak memory if there is no explicit vaDestroyBuffer(). The temporary workaround is to explcitily destroy VA buffers right after vaRenderPicture(). All VA drivers need to be aligned.
2012-02-13 09:10:35 +00:00
vaapi_destroy_buffer(va_display, &slice->param_id);
vaapi_destroy_buffer(va_display, &slice->data_id);
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
}
status = vaEndPicture(va_display, va_context);
if (!vaapi_check_status(status, "vaEndPicture()"))
return FALSE;
return TRUE;
}
decoder: fix memory leak when processing interlaced pictures. Fix memory leak when processing interlaced pictures and that occurs because the first field, represented as a GstVideoCodecFrame, never gets released. i.e. when the picture is completed, this is generally the case when the second field is successfully decoded, we need to propagate the GstVideoCodecFrame of the first field to the original GstVideoDecoder so that it could reclaim memory. Otherwise, we keep accumulating the first fields into GstVideoDecoder private frames list until the end-of-stream is reached. The frames are eventually released there, but too late, i.e. too much memory may have been consumed. https://bugzilla.gnome.org/show_bug.cgi?id=701257
2013-06-11 13:11:34 +00:00
static gboolean
do_output(GstVaapiPicture *picture)
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
{
decoder: use new GstVaapiSurfaceProxy utility functions. Use new GstVaapiSurfaceProxy internal helper functions to propagate the necessary GstVideoCodecFrame flags to vaapidecode (GStreamer 0.10). Also make GstVaapiDecoder push_frame() operate similarly to drop_frame(). i.e. increase the GstVideoCodecFrame reference count in push_frame rather than gst_vaapi_picture_output().
2013-04-16 16:56:24 +00:00
GstVideoCodecFrame * const out_frame = picture->frame;
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
GstVaapiSurfaceProxy *proxy;
decoder: use new GstVaapiSurfaceProxy utility functions. Use new GstVaapiSurfaceProxy internal helper functions to propagate the necessary GstVideoCodecFrame flags to vaapidecode (GStreamer 0.10). Also make GstVaapiDecoder push_frame() operate similarly to drop_frame(). i.e. increase the GstVideoCodecFrame reference count in push_frame rather than gst_vaapi_picture_output().
2013-04-16 16:56:24 +00:00
guint flags = 0;
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
decoder: fix memory leak when processing interlaced pictures. Fix memory leak when processing interlaced pictures and that occurs because the first field, represented as a GstVideoCodecFrame, never gets released. i.e. when the picture is completed, this is generally the case when the second field is successfully decoded, we need to propagate the GstVideoCodecFrame of the first field to the original GstVideoDecoder so that it could reclaim memory. Otherwise, we keep accumulating the first fields into GstVideoDecoder private frames list until the end-of-stream is reached. The frames are eventually released there, but too late, i.e. too much memory may have been consumed. https://bugzilla.gnome.org/show_bug.cgi?id=701257
2013-06-11 13:11:34 +00:00
if (GST_VAAPI_PICTURE_IS_OUTPUT(picture))
return TRUE;
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
decoder: allocate proxy surface earlier. This simplifies gst_vaapi_picture_output() to only update the presentation timestamp and submit the proxy to the decoder for output.
2012-01-31 10:34:17 +00:00
if (!picture->proxy)
return FALSE;
decoder: simplify output of decoded frames. Drop obsolete gst_vaapi_decoder_push_surface() that was no longer used. Change gst_vaapi_decoder_push_surface_proxy() semantics to assume PTS is already set correctly and reference count increased, if necessary.
2012-01-26 13:54:31 +00:00
decoder: maintain decoded frames as GstVideoCodecFrame objects. Maintain decoded surfaces as GstVideoCodecFrame objects instead of GstVaapiSurfaceProxy objects. The latter will tend to be reduced to the strict minimum: a context and a surface.
2012-12-13 14:34:10 +00:00
proxy = gst_vaapi_surface_proxy_ref(picture->proxy);
decoder: add support for video cropping. Add gst_vaapi_picture_set_crop_rect() helper function to copy the video cropping information from raw bitstreams to each picture being decoded. Also add helper function to surface proxy to propagate that information outside of libgstvaapi. e.g. plug-in elements or standalone applications. Signed-off-by: Sreerenj Balachandran <sreerenj.balachandran@intel.com> Signed-off-by: Gwenole Beauchesne <gwenole.beauchesne@intel.com>
2013-02-15 16:42:12 +00:00
if (picture->has_crop_rect)
gst_vaapi_surface_proxy_set_crop_rect(proxy, &picture->crop_rect);
decoder: maintain decoded frames as GstVideoCodecFrame objects. Maintain decoded surfaces as GstVideoCodecFrame objects instead of GstVaapiSurfaceProxy objects. The latter will tend to be reduced to the strict minimum: a context and a surface.
2012-12-13 14:34:10 +00:00
gst_video_codec_frame_set_user_data(out_frame,
proxy, (GDestroyNotify)gst_vaapi_mini_object_unref);
decoder: always use the calculated presentation timestamp. Use PTS value computed by the decoder, which could also be derived from the GstVideoCodecFrame PTS. This makes it possible to fix up the PTS if the original one was miscomputed or only represented a DTS instead.
2013-01-03 12:05:47 +00:00
out_frame->pts = picture->pts;
decoder: use new GstVaapiSurfaceProxy utility functions. Use new GstVaapiSurfaceProxy internal helper functions to propagate the necessary GstVideoCodecFrame flags to vaapidecode (GStreamer 0.10). Also make GstVaapiDecoder push_frame() operate similarly to drop_frame(). i.e. increase the GstVideoCodecFrame reference count in push_frame rather than gst_vaapi_picture_output().
2013-04-16 16:56:24 +00:00
decoder: maintain decoded frames as GstVideoCodecFrame objects. Maintain decoded surfaces as GstVideoCodecFrame objects instead of GstVaapiSurfaceProxy objects. The latter will tend to be reduced to the strict minimum: a context and a surface.
2012-12-13 14:34:10 +00:00
if (GST_VAAPI_PICTURE_IS_SKIPPED(picture))
GST_VIDEO_CODEC_FRAME_FLAG_SET(out_frame,
GST_VIDEO_CODEC_FRAME_FLAG_DECODE_ONLY);
decoder: use new GstVaapiSurfaceProxy utility functions. Use new GstVaapiSurfaceProxy internal helper functions to propagate the necessary GstVideoCodecFrame flags to vaapidecode (GStreamer 0.10). Also make GstVaapiDecoder push_frame() operate similarly to drop_frame(). i.e. increase the GstVideoCodecFrame reference count in push_frame rather than gst_vaapi_picture_output().
2013-04-16 16:56:24 +00:00
if (GST_VAAPI_PICTURE_IS_INTERLACED(picture)) {
flags |= GST_VAAPI_SURFACE_PROXY_FLAG_INTERLACED;
if (GST_VAAPI_PICTURE_IS_TFF(picture))
flags |= GST_VAAPI_SURFACE_PROXY_FLAG_TFF;
}
GST_VAAPI_SURFACE_PROXY_FLAG_SET(proxy, flags);
decoder: maintain decoded frames as GstVideoCodecFrame objects. Maintain decoded surfaces as GstVideoCodecFrame objects instead of GstVaapiSurfaceProxy objects. The latter will tend to be reduced to the strict minimum: a context and a surface.
2012-12-13 14:34:10 +00:00
gst_vaapi_decoder_push_frame(GET_DECODER(picture), out_frame);
decoder: dispose GstVideoCodecFrame earlier. Once the picture was output, it is no longer necessary to keep an extra reference to the underlying GstVideoCodecFrame. So, we can release it earlier, and maybe subsequently release the associate surface proxy earlier.
2013-07-15 15:43:34 +00:00
gst_video_codec_frame_clear(&picture->frame);
decoder: maintain decoded frames as GstVideoCodecFrame objects. Maintain decoded surfaces as GstVideoCodecFrame objects instead of GstVaapiSurfaceProxy objects. The latter will tend to be reduced to the strict minimum: a context and a surface.
2012-12-13 14:34:10 +00:00
decoder: add OUTPUT flag to pictures. Allow pictures to be marked as output gst_vaapi_picture_output().
2012-03-28 12:28:26 +00:00
GST_VAAPI_PICTURE_FLAG_SET(picture, GST_VAAPI_PICTURE_FLAG_OUTPUT);
decoder: simplify output of decoded frames. Drop obsolete gst_vaapi_decoder_push_surface() that was no longer used. Change gst_vaapi_decoder_push_surface_proxy() semantics to assume PTS is already set correctly and reference count increased, if necessary.
2012-01-26 13:54:31 +00:00
return TRUE;
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
}
decoder: fix memory leak when processing interlaced pictures. Fix memory leak when processing interlaced pictures and that occurs because the first field, represented as a GstVideoCodecFrame, never gets released. i.e. when the picture is completed, this is generally the case when the second field is successfully decoded, we need to propagate the GstVideoCodecFrame of the first field to the original GstVideoDecoder so that it could reclaim memory. Otherwise, we keep accumulating the first fields into GstVideoDecoder private frames list until the end-of-stream is reached. The frames are eventually released there, but too late, i.e. too much memory may have been consumed. https://bugzilla.gnome.org/show_bug.cgi?id=701257
2013-06-11 13:11:34 +00:00
gboolean
gst_vaapi_picture_output(GstVaapiPicture *picture)
{
g_return_val_if_fail(GST_VAAPI_IS_PICTURE(picture), FALSE);
if (G_UNLIKELY(picture->parent_picture)) {
/* Emit the first field to GstVideoDecoder so that to release
the underlying GstVideoCodecFrame. However, mark this
picture as skipped so that to not display it */
GstVaapiPicture * const parent_picture = picture->parent_picture;
do {
if (!GST_VAAPI_PICTURE_IS_INTERLACED(parent_picture))
break;
if (!GST_VAAPI_PICTURE_IS_FIRST_FIELD(parent_picture))
break;
GST_VAAPI_PICTURE_FLAG_SET(parent_picture,
GST_VAAPI_PICTURE_FLAG_SKIPPED);
if (!do_output(parent_picture))
return FALSE;
} while (0);
}
return do_output(picture);
}
decoder: add support for video cropping. Add gst_vaapi_picture_set_crop_rect() helper function to copy the video cropping information from raw bitstreams to each picture being decoded. Also add helper function to surface proxy to propagate that information outside of libgstvaapi. e.g. plug-in elements or standalone applications. Signed-off-by: Sreerenj Balachandran <sreerenj.balachandran@intel.com> Signed-off-by: Gwenole Beauchesne <gwenole.beauchesne@intel.com>
2013-02-15 16:42:12 +00:00
void
gst_vaapi_picture_set_crop_rect(GstVaapiPicture *picture,
const GstVaapiRectangle *crop_rect)
{
g_return_if_fail(GST_VAAPI_IS_PICTURE(picture));
picture->has_crop_rect = crop_rect != NULL;
if (picture->has_crop_rect)
picture->crop_rect = *crop_rect;
}
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
/* ------------------------------------------------------------------------- */
/* --- Slices --- */
/* ------------------------------------------------------------------------- */
Port codec objects to GstVaapiMiniObject.
2012-12-03 10:19:08 +00:00
GST_VAAPI_CODEC_DEFINE_TYPE(GstVaapiSlice, gst_vaapi_slice);
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
Port codec objects to GstVaapiMiniObject.
2012-12-03 10:19:08 +00:00
void
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
gst_vaapi_slice_destroy(GstVaapiSlice *slice)
{
VADisplay const va_display = GET_VA_DISPLAY(slice);
jpeg: add support for multiscan images. Add support for images with multiple scans per frame. The Huffman table can be updated before SOS, and thus possibly requiring multiple uploads of Huffman tables to the VA driver. So, the latter must be able to cope with multiple VA buffers of type 'huffman-table' and with the correct sequential order.
2013-09-23 14:49:41 +00:00
gst_vaapi_mini_object_replace((GstVaapiMiniObject **)&slice->huf_table,
NULL);
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
vaapi_destroy_buffer(va_display, &slice->data_id);
vaapi_destroy_buffer(va_display, &slice->param_id);
slice->param = NULL;
}
Port codec objects to GstVaapiMiniObject.
2012-12-03 10:19:08 +00:00
gboolean
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
gst_vaapi_slice_create(
GstVaapiSlice *slice,
const GstVaapiCodecObjectConstructorArgs *args
)
{
VASliceParameterBufferBase *slice_param;
decoder: optimize slice data buffers initialization. VA drivers may have a faster means to transfer user buffers to GPU buffers than using memcpy(). In particular, on Intel Gen graphics, we can use pwrite(). This provides for faster upload of bitstream and can help higher bitrates. vaapi_create_buffer() helper function was also updated to allow for un-mapped buffers and pre-initialized data for buffers.
2012-01-30 09:15:32 +00:00
gboolean success;
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
Port codec objects to GstVaapiMiniObject.
2012-12-03 10:19:08 +00:00
slice->data_id = VA_INVALID_ID;
decoder: optimize slice data buffers initialization. VA drivers may have a faster means to transfer user buffers to GPU buffers than using memcpy(). In particular, on Intel Gen graphics, we can use pwrite(). This provides for faster upload of bitstream and can help higher bitrates. vaapi_create_buffer() helper function was also updated to allow for un-mapped buffers and pre-initialized data for buffers.
2012-01-30 09:15:32 +00:00
success = vaapi_create_buffer(
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
GET_VA_DISPLAY(slice),
GET_VA_CONTEXT(slice),
VASliceDataBufferType,
args->data_size,
decoder: optimize slice data buffers initialization. VA drivers may have a faster means to transfer user buffers to GPU buffers than using memcpy(). In particular, on Intel Gen graphics, we can use pwrite(). This provides for faster upload of bitstream and can help higher bitrates. vaapi_create_buffer() helper function was also updated to allow for un-mapped buffers and pre-initialized data for buffers.
2012-01-30 09:15:32 +00:00
args->data,
&slice->data_id,
NULL
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
);
decoder: optimize slice data buffers initialization. VA drivers may have a faster means to transfer user buffers to GPU buffers than using memcpy(). In particular, on Intel Gen graphics, we can use pwrite(). This provides for faster upload of bitstream and can help higher bitrates. vaapi_create_buffer() helper function was also updated to allow for un-mapped buffers and pre-initialized data for buffers.
2012-01-30 09:15:32 +00:00
if (!success)
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
return FALSE;
Port codec objects to GstVaapiMiniObject.
2012-12-03 10:19:08 +00:00
slice->param_id = VA_INVALID_ID;
decoder: optimize slice data buffers initialization. VA drivers may have a faster means to transfer user buffers to GPU buffers than using memcpy(). In particular, on Intel Gen graphics, we can use pwrite(). This provides for faster upload of bitstream and can help higher bitrates. vaapi_create_buffer() helper function was also updated to allow for un-mapped buffers and pre-initialized data for buffers.
2012-01-30 09:15:32 +00:00
success = vaapi_create_buffer(
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
GET_VA_DISPLAY(slice),
GET_VA_CONTEXT(slice),
VASliceParameterBufferType,
args->param_size,
decoder: optimize slice data buffers initialization. VA drivers may have a faster means to transfer user buffers to GPU buffers than using memcpy(). In particular, on Intel Gen graphics, we can use pwrite(). This provides for faster upload of bitstream and can help higher bitrates. vaapi_create_buffer() helper function was also updated to allow for un-mapped buffers and pre-initialized data for buffers.
2012-01-30 09:15:32 +00:00
args->param,
&slice->param_id,
&slice->param
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
);
decoder: optimize slice data buffers initialization. VA drivers may have a faster means to transfer user buffers to GPU buffers than using memcpy(). In particular, on Intel Gen graphics, we can use pwrite(). This provides for faster upload of bitstream and can help higher bitrates. vaapi_create_buffer() helper function was also updated to allow for un-mapped buffers and pre-initialized data for buffers.
2012-01-30 09:15:32 +00:00
if (!success)
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
return FALSE;
slice_param = slice->param;
slice_param->slice_data_size = args->data_size;
slice_param->slice_data_offset = 0;
slice_param->slice_data_flag = VA_SLICE_DATA_FLAG_ALL;
return TRUE;
}
GstVaapiSlice *
gst_vaapi_slice_new(
GstVaapiDecoder *decoder,
gconstpointer param,
guint param_size,
const guchar *data,
guint data_size
)
{
GstVaapiCodecObject *object;
object = gst_vaapi_codec_object_new(
Port codec objects to GstVaapiMiniObject.
2012-12-03 10:19:08 +00:00
&GstVaapiSliceClass,
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
GST_VAAPI_CODEC_BASE(decoder),
param, param_size,
Port codec objects to GstVaapiMiniObject.
2012-12-03 10:19:08 +00:00
data, data_size,
0
decoder: rework the internal VA objects API. The new API simplifies a lot reference counting and makes it more flexible for future additions/changes. The GstVaapiCodecInfo is also gone. Rather, new helper macros are provided to allocate picture, slice and quantization matrix parameter buffers.
2012-01-26 08:48:11 +00:00
);
return GST_VAAPI_SLICE_CAST(object);
}
Reference in a new issue Copy permalink