gstreamer/gst/gstpad.c

847 lines
20 KiB
C
Raw Normal View History

/* Gnome-Streamer
* Copyright (C) <1999> Erik Walthinsen <omega@cse.ogi.edu>
*
* This library is free software; you can redistribute it and/or
* modify it under the terms of the GNU Library General Public
* License as published by the Free Software Foundation; either
* version 2 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
* Library General Public License for more details.
*
* You should have received a copy of the GNU Library General Public
* License along with this library; if not, write to the
* Free Software Foundation, Inc., 59 Temple Place - Suite 330,
* Boston, MA 02111-1307, USA.
*/
//#define DEBUG_ENABLED
#include <gst/gst.h>
#include <gst/gstpad.h>
#include <gst/gstelement.h>
#include <gst/gsttype.h>
/* Pad signals and args */
enum {
SET_ACTIVE,
/* FILL ME */
LAST_SIGNAL
};
enum {
ARG_0,
ARG_ACTIVE,
/* FILL ME */
};
static void gst_pad_class_init(GstPadClass *klass);
static void gst_pad_init(GstPad *pad);
static void gst_pad_set_arg(GtkObject *object,GtkArg *arg,guint id);
static void gst_pad_get_arg(GtkObject *object,GtkArg *arg,guint id);
static void gst_pad_real_destroy(GtkObject *object);
static GstObject *parent_class = NULL;
static guint gst_pad_signals[LAST_SIGNAL] = { 0 };
GtkType
gst_pad_get_type(void) {
static GtkType pad_type = 0;
if (!pad_type) {
static const GtkTypeInfo pad_info = {
"GstPad",
sizeof(GstPad),
sizeof(GstPadClass),
(GtkClassInitFunc)gst_pad_class_init,
(GtkObjectInitFunc)gst_pad_init,
(GtkArgSetFunc)NULL,
(GtkArgGetFunc)NULL,
(GtkClassInitFunc)NULL,
};
pad_type = gtk_type_unique(GST_TYPE_OBJECT,&pad_info);
}
return pad_type;
}
static void
gst_pad_class_init (GstPadClass *klass)
{
GtkObjectClass *gtkobject_class;
gtkobject_class = (GtkObjectClass*)klass;
parent_class = gtk_type_class(GST_TYPE_OBJECT);
gst_pad_signals[SET_ACTIVE] =
gtk_signal_new ("set_active", GTK_RUN_LAST, gtkobject_class->type,
GTK_SIGNAL_OFFSET (GstPadClass, set_active),
gtk_marshal_NONE__BOOL, GTK_TYPE_NONE, 1,
GTK_TYPE_BOOL);
gtk_object_add_arg_type ("GstPad::active", GTK_TYPE_BOOL,
GTK_ARG_READWRITE, ARG_ACTIVE);
gtkobject_class->destroy = gst_pad_real_destroy;
gtkobject_class->set_arg = gst_pad_set_arg;
gtkobject_class->get_arg = gst_pad_get_arg;
}
static void
gst_pad_init (GstPad *pad)
{
pad->direction = GST_PAD_UNKNOWN;
pad->peer = NULL;
pad->chainfunc = NULL;
pad->pullfunc = NULL;
pad->pullregionfunc = NULL;
pad->pushfunc = NULL;
pad->qosfunc = NULL;
pad->parent = NULL;
pad->ghostparents = NULL;
pad->types = NULL;
}
static void
gst_pad_set_arg (GtkObject *object,GtkArg *arg,guint id) {
g_return_if_fail(GST_IS_PAD(object));
switch (id) {
case ARG_ACTIVE:
if (GTK_VALUE_BOOL(*arg)) {
gst_info("gstpad: activating pad\n");
GST_FLAG_UNSET(object,GST_PAD_DISABLED);
} else {
gst_info("gstpad: de-activating pad\n");
GST_FLAG_SET(object,GST_PAD_DISABLED);
}
gtk_signal_emit(GTK_OBJECT(object), gst_pad_signals[SET_ACTIVE],
! GST_FLAG_IS_SET(object,GST_PAD_DISABLED));
break;
default:
break;
}
}
static void
gst_pad_get_arg (GtkObject *object,
GtkArg *arg,
guint id)
{
/* it's not null if we got it, but it might not be ours */
g_return_if_fail (GST_IS_PAD (object));
switch (id) {
case ARG_ACTIVE:
GTK_VALUE_BOOL (*arg) = ! GST_FLAG_IS_SET (object, GST_PAD_DISABLED);
break;
default:
break;
}
}
/**
* gst_pad_new:
* @name: name of new pad
* @direction: either GST_PAD_SRC or GST_PAD_SINK
*
* Create a new pad with given name.
*
* Returns: new pad
*/
GstPad*
gst_pad_new (gchar *name,
GstPadDirection direction)
{
GstPad *pad;
g_return_val_if_fail (name != NULL, NULL);
g_return_val_if_fail (direction != GST_PAD_UNKNOWN, NULL);
pad = GST_PAD (gtk_type_new (gst_pad_get_type ()));
pad->name = g_strdup (name);
pad->direction = direction;
return pad;
}
/**
* gst_pad_get_direction:
* @pad: the Pad to get the direction from
*
* get the direction of the pad
*
* Returns: the direction of the pad
*/
GstPadDirection
gst_pad_get_direction (GstPad *pad)
{
g_return_val_if_fail (pad != NULL, GST_PAD_UNKNOWN);
g_return_val_if_fail (GST_IS_PAD (pad), GST_PAD_UNKNOWN);
return pad->direction;
}
/**
* gst_pad_set_name:
* @pad: the pad to set the name of
* @name: the name of the pad
*
* set the name of a pad
*/
void
gst_pad_set_name (GstPad *pad,
const gchar *name)
{
g_return_if_fail (pad != NULL);
g_return_if_fail (GST_IS_PAD (pad));
if (pad->name != NULL)
g_free (pad->name);
pad->name = g_strdup (name);
}
/**
* gst_pad_get_name:
* @pad: the pad to get the name of
*
* get the name of a pad
*
* Returns: the name of the pad, don't free.
*/
const gchar*
gst_pad_get_name (GstPad *pad)
{
g_return_val_if_fail (pad != NULL, NULL);
g_return_val_if_fail (GST_IS_PAD (pad), NULL);
return pad->name;
}
/**
* gst_pad_set_pull_function:
* @pad: the pad to set the pull function for
* @pull: the pull function
*
* Set the given pull function for the pad
*/
void
gst_pad_set_pull_function (GstPad *pad,
GstPadPullFunction pull)
{
g_return_if_fail (pad != NULL);
g_return_if_fail (GST_IS_PAD (pad));
// the if and such should optimize out when DEBUG is off
DEBUG("setting pull function for %s:%s\n",GST_DEBUG_PAD_NAME(pad));
pad->pullfunc = pull;
DEBUG("pullfunc for %s:%s(%p) at %p is set to %p\n",GST_DEBUG_PAD_NAME(pad),pad,&pad->pullfunc,pull);
}
/**
* gst_pad_set_pullregion_function:
* @pad: the pad to set the pull function for
* @pull: the pull function
*
* Set the given pull function for the pad
*/
void
gst_pad_set_pullregion_function (GstPad *pad,
GstPadPullRegionFunction pull)
{
g_return_if_fail (pad != NULL);
g_return_if_fail (GST_IS_PAD (pad));
g_print("gstpad: pad setting pullregion function\n");
pad->pullregionfunc = pull;
}
/**
* gst_pad_set_chain_function:
* @pad: the pad to set the chain function for
* @chain: the chain function
*
* Set the given chain function for the pad
*/
void gst_pad_set_chain_function (GstPad *pad,
GstPadChainFunction chain)
{
g_return_if_fail (pad != NULL);
g_return_if_fail (GST_IS_PAD (pad));
pad->chainfunc = chain;
}
/**
* gst_pad_set_qos_function:
* @pad: the pad to set the qos function for
* @qos: the qos function
*
* Set the given qos function for the pad
*/
void
gst_pad_set_qos_function (GstPad *pad,
GstPadQoSFunction qos)
{
g_return_if_fail (pad != NULL);
g_return_if_fail (GST_IS_PAD (pad));
pad->qosfunc = qos;
}
/**
* gst_pad_push:
* @pad: the pad to push
* @buffer: the buffer to push
*
* pushes a buffer along a src pad
*/
void
gst_pad_push (GstPad *pad,
GstBuffer *buffer)
{
Changed the way things are scheduled, especially sources. A Src used to have a push() function, and optionally a pus... Original commit message from CVS: Changed the way things are scheduled, especially sources. A Src used to have a push() function, and optionally a pushregion() to deal with async reads, etc. That whole thing has gone away, in favor of providing a pull() function for the output (Src) pad instead, ala chain functions. This makes constructing cothreaded schedules out of non-loop elements somewhat easier. Basically there was always a question as to which pad was being dealt with. In the pullregion case, cothread-specific data was used to try to pass the region struct to the right place, which is a slow hack. And in general, the push function severely limited the kind of tricks that could be played when there's more than one output pad, such as a multi-out file reader with async capabilities on each pad independently. This changes the way cothread scheduling occurs. Instead of the hack to deal with Src's by calling their push() function (or optionally the pushregion(), in certain cases), we now are working towards a general mechanism where pads are the only thing that are dealt with directly. An optimization was made in the process of doing this: the loopfunction actually run as the outer [stack] frame of the cothread is now set more intelligently in create_plan() based on what kind of element it is. We now have: loopfunc_wrapper: used for loop-based elements, it simply calls the loopfunc in a loop, paying attention to COTHREAD_STOPPING (see below). It currently does other, soon to be depracated, stuff. pullsrc_wrapper: wraps a Src that's not loop-based (since your options are now loop- or pull-based) There will be a couple more to deal with other cases, such as Connections and chain-based elements. The general idea is that it's a lot more efficient to make the decisions once in create_plan than to keep doing this huge if/else chain in the wrapper. Just choose the right wrapper up front. It'll be most apparent performance-wise in the case of whichever element context is switched to first for each iteration, since the whole wrapper setup is done for every iteration. The tricky part is that there is now a bit of overloading of the function pointers in a pad. The current meanings (possibly to change a bit more soon) are: chainfunc: as always, chainfunc pointer is mirrored between peer pads (this may change, and the chain func may end up in pushfunc) pushfunc: SrcPad: gst_pad_pushfunc_proxy, cothread_switch to peer SinkPad: none (may take over chainfunc, see below) pullfunc: SrcPad: Src or Connection's function to construct buffers SinkPad: gst_pad_pullfunc_proxy, cothread_switch to peer There are a number of issues remaining with the scheduling, not the least of which is the fact that Connections are still dealt with the old way, with _push() functions and such. I'm trying to figure out a way to unify the system so it makes sense. Following the scheduling system is hard enough, trying to change it is murder. Another useful scheduling addition, mentioned above, is COTHREAD_STOPPING. It's an element flag that's used to signal whatever code is running in cothread context that it should be finishing up and exiting soon. An example of this is in plugins/cobin/spindentity.c. All the loops should now be composed of do/while loops, rather than while(1) loops: do { buf = gst_pad_pull(spindentity->sinkpad); gst_pad_push(spindentity->srcpad,buf); } while (!GST_ELEMENT_IS_COTHREAD_STOPPING(element)); The reason for this is that COTHREAD_STOPPING may be set before the above loop ever gets started. It wouldn't do for the body of the loop to never once get called, that would simply stall the pipeline. Note that only the core library code is ever responsible for setting and unsetting this flag. All elements have to do is respond to it by cleanly exiting the loop and the function holding it. This is needed primarily to allow iterations to occur properly. Basically, there's a single entry point in the cothread scheduling loop, gst_bin_iterate_func() simply switches to this cothread. If the element in this context is allowed to loop infinitely, nothing would even switch back to the context from which the iterate() was originally called. This is a bit of a problem. The solution is for there to be an implicit switch back to the originating context. Now, even I'm not sure exactly how this works, but if the cothread that's switched to actually returns, execution returns back to the calling context, i.e. iterate_func(). COTHREAD_STOPPING is therefore set just before switching into this (currently randomly chosen) context, on the assumption that it will return promptly after finishing its duties. The burden of clearing the flag falls to the various wrapper functions provided by the Bin code, thus element writers don't have to worry about doing that at all (and simply shouldn't). Related changes: All the sources in elements/ have been changed to reflect the new system. FIXMEs: 1) gstpipeline.c calls gst_src_push at some point, dunno why, it's commented out now. 2) any other sources, including vcdsrc, dvdsrc, and v4lsrc will break badly and need to be modified to work as pull-based sources.
2000-12-04 10:52:30 +00:00
GstPad *peer;
DEBUG_ENTER("(pad:'%s',buffer:%p)",gst_pad_get_name(pad),buffer);
g_return_if_fail(pad != NULL);
g_return_if_fail(GST_IS_PAD(pad));
g_return_if_fail(GST_PAD_CONNECTED(pad));
g_return_if_fail(buffer != NULL);
/* if the pad has been disabled, unreference the pad and let it drop */
if (GST_FLAG_IS_SET(pad,GST_PAD_DISABLED)) {
g_print("gst_pad_push: pad disabled, dropping buffer\n");
gst_buffer_unref(buffer);
return;
}
gst_trace_add_entry(NULL,0,buffer,"push buffer");
Changed the way things are scheduled, especially sources. A Src used to have a push() function, and optionally a pus... Original commit message from CVS: Changed the way things are scheduled, especially sources. A Src used to have a push() function, and optionally a pushregion() to deal with async reads, etc. That whole thing has gone away, in favor of providing a pull() function for the output (Src) pad instead, ala chain functions. This makes constructing cothreaded schedules out of non-loop elements somewhat easier. Basically there was always a question as to which pad was being dealt with. In the pullregion case, cothread-specific data was used to try to pass the region struct to the right place, which is a slow hack. And in general, the push function severely limited the kind of tricks that could be played when there's more than one output pad, such as a multi-out file reader with async capabilities on each pad independently. This changes the way cothread scheduling occurs. Instead of the hack to deal with Src's by calling their push() function (or optionally the pushregion(), in certain cases), we now are working towards a general mechanism where pads are the only thing that are dealt with directly. An optimization was made in the process of doing this: the loopfunction actually run as the outer [stack] frame of the cothread is now set more intelligently in create_plan() based on what kind of element it is. We now have: loopfunc_wrapper: used for loop-based elements, it simply calls the loopfunc in a loop, paying attention to COTHREAD_STOPPING (see below). It currently does other, soon to be depracated, stuff. pullsrc_wrapper: wraps a Src that's not loop-based (since your options are now loop- or pull-based) There will be a couple more to deal with other cases, such as Connections and chain-based elements. The general idea is that it's a lot more efficient to make the decisions once in create_plan than to keep doing this huge if/else chain in the wrapper. Just choose the right wrapper up front. It'll be most apparent performance-wise in the case of whichever element context is switched to first for each iteration, since the whole wrapper setup is done for every iteration. The tricky part is that there is now a bit of overloading of the function pointers in a pad. The current meanings (possibly to change a bit more soon) are: chainfunc: as always, chainfunc pointer is mirrored between peer pads (this may change, and the chain func may end up in pushfunc) pushfunc: SrcPad: gst_pad_pushfunc_proxy, cothread_switch to peer SinkPad: none (may take over chainfunc, see below) pullfunc: SrcPad: Src or Connection's function to construct buffers SinkPad: gst_pad_pullfunc_proxy, cothread_switch to peer There are a number of issues remaining with the scheduling, not the least of which is the fact that Connections are still dealt with the old way, with _push() functions and such. I'm trying to figure out a way to unify the system so it makes sense. Following the scheduling system is hard enough, trying to change it is murder. Another useful scheduling addition, mentioned above, is COTHREAD_STOPPING. It's an element flag that's used to signal whatever code is running in cothread context that it should be finishing up and exiting soon. An example of this is in plugins/cobin/spindentity.c. All the loops should now be composed of do/while loops, rather than while(1) loops: do { buf = gst_pad_pull(spindentity->sinkpad); gst_pad_push(spindentity->srcpad,buf); } while (!GST_ELEMENT_IS_COTHREAD_STOPPING(element)); The reason for this is that COTHREAD_STOPPING may be set before the above loop ever gets started. It wouldn't do for the body of the loop to never once get called, that would simply stall the pipeline. Note that only the core library code is ever responsible for setting and unsetting this flag. All elements have to do is respond to it by cleanly exiting the loop and the function holding it. This is needed primarily to allow iterations to occur properly. Basically, there's a single entry point in the cothread scheduling loop, gst_bin_iterate_func() simply switches to this cothread. If the element in this context is allowed to loop infinitely, nothing would even switch back to the context from which the iterate() was originally called. This is a bit of a problem. The solution is for there to be an implicit switch back to the originating context. Now, even I'm not sure exactly how this works, but if the cothread that's switched to actually returns, execution returns back to the calling context, i.e. iterate_func(). COTHREAD_STOPPING is therefore set just before switching into this (currently randomly chosen) context, on the assumption that it will return promptly after finishing its duties. The burden of clearing the flag falls to the various wrapper functions provided by the Bin code, thus element writers don't have to worry about doing that at all (and simply shouldn't). Related changes: All the sources in elements/ have been changed to reflect the new system. FIXMEs: 1) gstpipeline.c calls gst_src_push at some point, dunno why, it's commented out now. 2) any other sources, including vcdsrc, dvdsrc, and v4lsrc will break badly and need to be modified to work as pull-based sources.
2000-12-04 10:52:30 +00:00
peer = pad->peer;
g_return_if_fail(peer != NULL);
// first check to see if there's a push handler
if (pad->pushfunc != NULL) {
// put the buffer in peer's holding pen
Changed the way things are scheduled, especially sources. A Src used to have a push() function, and optionally a pus... Original commit message from CVS: Changed the way things are scheduled, especially sources. A Src used to have a push() function, and optionally a pushregion() to deal with async reads, etc. That whole thing has gone away, in favor of providing a pull() function for the output (Src) pad instead, ala chain functions. This makes constructing cothreaded schedules out of non-loop elements somewhat easier. Basically there was always a question as to which pad was being dealt with. In the pullregion case, cothread-specific data was used to try to pass the region struct to the right place, which is a slow hack. And in general, the push function severely limited the kind of tricks that could be played when there's more than one output pad, such as a multi-out file reader with async capabilities on each pad independently. This changes the way cothread scheduling occurs. Instead of the hack to deal with Src's by calling their push() function (or optionally the pushregion(), in certain cases), we now are working towards a general mechanism where pads are the only thing that are dealt with directly. An optimization was made in the process of doing this: the loopfunction actually run as the outer [stack] frame of the cothread is now set more intelligently in create_plan() based on what kind of element it is. We now have: loopfunc_wrapper: used for loop-based elements, it simply calls the loopfunc in a loop, paying attention to COTHREAD_STOPPING (see below). It currently does other, soon to be depracated, stuff. pullsrc_wrapper: wraps a Src that's not loop-based (since your options are now loop- or pull-based) There will be a couple more to deal with other cases, such as Connections and chain-based elements. The general idea is that it's a lot more efficient to make the decisions once in create_plan than to keep doing this huge if/else chain in the wrapper. Just choose the right wrapper up front. It'll be most apparent performance-wise in the case of whichever element context is switched to first for each iteration, since the whole wrapper setup is done for every iteration. The tricky part is that there is now a bit of overloading of the function pointers in a pad. The current meanings (possibly to change a bit more soon) are: chainfunc: as always, chainfunc pointer is mirrored between peer pads (this may change, and the chain func may end up in pushfunc) pushfunc: SrcPad: gst_pad_pushfunc_proxy, cothread_switch to peer SinkPad: none (may take over chainfunc, see below) pullfunc: SrcPad: Src or Connection's function to construct buffers SinkPad: gst_pad_pullfunc_proxy, cothread_switch to peer There are a number of issues remaining with the scheduling, not the least of which is the fact that Connections are still dealt with the old way, with _push() functions and such. I'm trying to figure out a way to unify the system so it makes sense. Following the scheduling system is hard enough, trying to change it is murder. Another useful scheduling addition, mentioned above, is COTHREAD_STOPPING. It's an element flag that's used to signal whatever code is running in cothread context that it should be finishing up and exiting soon. An example of this is in plugins/cobin/spindentity.c. All the loops should now be composed of do/while loops, rather than while(1) loops: do { buf = gst_pad_pull(spindentity->sinkpad); gst_pad_push(spindentity->srcpad,buf); } while (!GST_ELEMENT_IS_COTHREAD_STOPPING(element)); The reason for this is that COTHREAD_STOPPING may be set before the above loop ever gets started. It wouldn't do for the body of the loop to never once get called, that would simply stall the pipeline. Note that only the core library code is ever responsible for setting and unsetting this flag. All elements have to do is respond to it by cleanly exiting the loop and the function holding it. This is needed primarily to allow iterations to occur properly. Basically, there's a single entry point in the cothread scheduling loop, gst_bin_iterate_func() simply switches to this cothread. If the element in this context is allowed to loop infinitely, nothing would even switch back to the context from which the iterate() was originally called. This is a bit of a problem. The solution is for there to be an implicit switch back to the originating context. Now, even I'm not sure exactly how this works, but if the cothread that's switched to actually returns, execution returns back to the calling context, i.e. iterate_func(). COTHREAD_STOPPING is therefore set just before switching into this (currently randomly chosen) context, on the assumption that it will return promptly after finishing its duties. The burden of clearing the flag falls to the various wrapper functions provided by the Bin code, thus element writers don't have to worry about doing that at all (and simply shouldn't). Related changes: All the sources in elements/ have been changed to reflect the new system. FIXMEs: 1) gstpipeline.c calls gst_src_push at some point, dunno why, it's commented out now. 2) any other sources, including vcdsrc, dvdsrc, and v4lsrc will break badly and need to be modified to work as pull-based sources.
2000-12-04 10:52:30 +00:00
peer->bufpen = buffer;
// now inform the handler that the peer pad has something
Changed the way things are scheduled, especially sources. A Src used to have a push() function, and optionally a pus... Original commit message from CVS: Changed the way things are scheduled, especially sources. A Src used to have a push() function, and optionally a pushregion() to deal with async reads, etc. That whole thing has gone away, in favor of providing a pull() function for the output (Src) pad instead, ala chain functions. This makes constructing cothreaded schedules out of non-loop elements somewhat easier. Basically there was always a question as to which pad was being dealt with. In the pullregion case, cothread-specific data was used to try to pass the region struct to the right place, which is a slow hack. And in general, the push function severely limited the kind of tricks that could be played when there's more than one output pad, such as a multi-out file reader with async capabilities on each pad independently. This changes the way cothread scheduling occurs. Instead of the hack to deal with Src's by calling their push() function (or optionally the pushregion(), in certain cases), we now are working towards a general mechanism where pads are the only thing that are dealt with directly. An optimization was made in the process of doing this: the loopfunction actually run as the outer [stack] frame of the cothread is now set more intelligently in create_plan() based on what kind of element it is. We now have: loopfunc_wrapper: used for loop-based elements, it simply calls the loopfunc in a loop, paying attention to COTHREAD_STOPPING (see below). It currently does other, soon to be depracated, stuff. pullsrc_wrapper: wraps a Src that's not loop-based (since your options are now loop- or pull-based) There will be a couple more to deal with other cases, such as Connections and chain-based elements. The general idea is that it's a lot more efficient to make the decisions once in create_plan than to keep doing this huge if/else chain in the wrapper. Just choose the right wrapper up front. It'll be most apparent performance-wise in the case of whichever element context is switched to first for each iteration, since the whole wrapper setup is done for every iteration. The tricky part is that there is now a bit of overloading of the function pointers in a pad. The current meanings (possibly to change a bit more soon) are: chainfunc: as always, chainfunc pointer is mirrored between peer pads (this may change, and the chain func may end up in pushfunc) pushfunc: SrcPad: gst_pad_pushfunc_proxy, cothread_switch to peer SinkPad: none (may take over chainfunc, see below) pullfunc: SrcPad: Src or Connection's function to construct buffers SinkPad: gst_pad_pullfunc_proxy, cothread_switch to peer There are a number of issues remaining with the scheduling, not the least of which is the fact that Connections are still dealt with the old way, with _push() functions and such. I'm trying to figure out a way to unify the system so it makes sense. Following the scheduling system is hard enough, trying to change it is murder. Another useful scheduling addition, mentioned above, is COTHREAD_STOPPING. It's an element flag that's used to signal whatever code is running in cothread context that it should be finishing up and exiting soon. An example of this is in plugins/cobin/spindentity.c. All the loops should now be composed of do/while loops, rather than while(1) loops: do { buf = gst_pad_pull(spindentity->sinkpad); gst_pad_push(spindentity->srcpad,buf); } while (!GST_ELEMENT_IS_COTHREAD_STOPPING(element)); The reason for this is that COTHREAD_STOPPING may be set before the above loop ever gets started. It wouldn't do for the body of the loop to never once get called, that would simply stall the pipeline. Note that only the core library code is ever responsible for setting and unsetting this flag. All elements have to do is respond to it by cleanly exiting the loop and the function holding it. This is needed primarily to allow iterations to occur properly. Basically, there's a single entry point in the cothread scheduling loop, gst_bin_iterate_func() simply switches to this cothread. If the element in this context is allowed to loop infinitely, nothing would even switch back to the context from which the iterate() was originally called. This is a bit of a problem. The solution is for there to be an implicit switch back to the originating context. Now, even I'm not sure exactly how this works, but if the cothread that's switched to actually returns, execution returns back to the calling context, i.e. iterate_func(). COTHREAD_STOPPING is therefore set just before switching into this (currently randomly chosen) context, on the assumption that it will return promptly after finishing its duties. The burden of clearing the flag falls to the various wrapper functions provided by the Bin code, thus element writers don't have to worry about doing that at all (and simply shouldn't). Related changes: All the sources in elements/ have been changed to reflect the new system. FIXMEs: 1) gstpipeline.c calls gst_src_push at some point, dunno why, it's commented out now. 2) any other sources, including vcdsrc, dvdsrc, and v4lsrc will break badly and need to be modified to work as pull-based sources.
2000-12-04 10:52:30 +00:00
(pad->pushfunc)(peer);
// otherwise we assume we're chaining directly
Changed the way things are scheduled, especially sources. A Src used to have a push() function, and optionally a pus... Original commit message from CVS: Changed the way things are scheduled, especially sources. A Src used to have a push() function, and optionally a pushregion() to deal with async reads, etc. That whole thing has gone away, in favor of providing a pull() function for the output (Src) pad instead, ala chain functions. This makes constructing cothreaded schedules out of non-loop elements somewhat easier. Basically there was always a question as to which pad was being dealt with. In the pullregion case, cothread-specific data was used to try to pass the region struct to the right place, which is a slow hack. And in general, the push function severely limited the kind of tricks that could be played when there's more than one output pad, such as a multi-out file reader with async capabilities on each pad independently. This changes the way cothread scheduling occurs. Instead of the hack to deal with Src's by calling their push() function (or optionally the pushregion(), in certain cases), we now are working towards a general mechanism where pads are the only thing that are dealt with directly. An optimization was made in the process of doing this: the loopfunction actually run as the outer [stack] frame of the cothread is now set more intelligently in create_plan() based on what kind of element it is. We now have: loopfunc_wrapper: used for loop-based elements, it simply calls the loopfunc in a loop, paying attention to COTHREAD_STOPPING (see below). It currently does other, soon to be depracated, stuff. pullsrc_wrapper: wraps a Src that's not loop-based (since your options are now loop- or pull-based) There will be a couple more to deal with other cases, such as Connections and chain-based elements. The general idea is that it's a lot more efficient to make the decisions once in create_plan than to keep doing this huge if/else chain in the wrapper. Just choose the right wrapper up front. It'll be most apparent performance-wise in the case of whichever element context is switched to first for each iteration, since the whole wrapper setup is done for every iteration. The tricky part is that there is now a bit of overloading of the function pointers in a pad. The current meanings (possibly to change a bit more soon) are: chainfunc: as always, chainfunc pointer is mirrored between peer pads (this may change, and the chain func may end up in pushfunc) pushfunc: SrcPad: gst_pad_pushfunc_proxy, cothread_switch to peer SinkPad: none (may take over chainfunc, see below) pullfunc: SrcPad: Src or Connection's function to construct buffers SinkPad: gst_pad_pullfunc_proxy, cothread_switch to peer There are a number of issues remaining with the scheduling, not the least of which is the fact that Connections are still dealt with the old way, with _push() functions and such. I'm trying to figure out a way to unify the system so it makes sense. Following the scheduling system is hard enough, trying to change it is murder. Another useful scheduling addition, mentioned above, is COTHREAD_STOPPING. It's an element flag that's used to signal whatever code is running in cothread context that it should be finishing up and exiting soon. An example of this is in plugins/cobin/spindentity.c. All the loops should now be composed of do/while loops, rather than while(1) loops: do { buf = gst_pad_pull(spindentity->sinkpad); gst_pad_push(spindentity->srcpad,buf); } while (!GST_ELEMENT_IS_COTHREAD_STOPPING(element)); The reason for this is that COTHREAD_STOPPING may be set before the above loop ever gets started. It wouldn't do for the body of the loop to never once get called, that would simply stall the pipeline. Note that only the core library code is ever responsible for setting and unsetting this flag. All elements have to do is respond to it by cleanly exiting the loop and the function holding it. This is needed primarily to allow iterations to occur properly. Basically, there's a single entry point in the cothread scheduling loop, gst_bin_iterate_func() simply switches to this cothread. If the element in this context is allowed to loop infinitely, nothing would even switch back to the context from which the iterate() was originally called. This is a bit of a problem. The solution is for there to be an implicit switch back to the originating context. Now, even I'm not sure exactly how this works, but if the cothread that's switched to actually returns, execution returns back to the calling context, i.e. iterate_func(). COTHREAD_STOPPING is therefore set just before switching into this (currently randomly chosen) context, on the assumption that it will return promptly after finishing its duties. The burden of clearing the flag falls to the various wrapper functions provided by the Bin code, thus element writers don't have to worry about doing that at all (and simply shouldn't). Related changes: All the sources in elements/ have been changed to reflect the new system. FIXMEs: 1) gstpipeline.c calls gst_src_push at some point, dunno why, it's commented out now. 2) any other sources, including vcdsrc, dvdsrc, and v4lsrc will break badly and need to be modified to work as pull-based sources.
2000-12-04 10:52:30 +00:00
} else if (peer->chainfunc != NULL) {
//g_print("-- gst_pad_push(): calling chain handler\n");
Changed the way things are scheduled, especially sources. A Src used to have a push() function, and optionally a pus... Original commit message from CVS: Changed the way things are scheduled, especially sources. A Src used to have a push() function, and optionally a pushregion() to deal with async reads, etc. That whole thing has gone away, in favor of providing a pull() function for the output (Src) pad instead, ala chain functions. This makes constructing cothreaded schedules out of non-loop elements somewhat easier. Basically there was always a question as to which pad was being dealt with. In the pullregion case, cothread-specific data was used to try to pass the region struct to the right place, which is a slow hack. And in general, the push function severely limited the kind of tricks that could be played when there's more than one output pad, such as a multi-out file reader with async capabilities on each pad independently. This changes the way cothread scheduling occurs. Instead of the hack to deal with Src's by calling their push() function (or optionally the pushregion(), in certain cases), we now are working towards a general mechanism where pads are the only thing that are dealt with directly. An optimization was made in the process of doing this: the loopfunction actually run as the outer [stack] frame of the cothread is now set more intelligently in create_plan() based on what kind of element it is. We now have: loopfunc_wrapper: used for loop-based elements, it simply calls the loopfunc in a loop, paying attention to COTHREAD_STOPPING (see below). It currently does other, soon to be depracated, stuff. pullsrc_wrapper: wraps a Src that's not loop-based (since your options are now loop- or pull-based) There will be a couple more to deal with other cases, such as Connections and chain-based elements. The general idea is that it's a lot more efficient to make the decisions once in create_plan than to keep doing this huge if/else chain in the wrapper. Just choose the right wrapper up front. It'll be most apparent performance-wise in the case of whichever element context is switched to first for each iteration, since the whole wrapper setup is done for every iteration. The tricky part is that there is now a bit of overloading of the function pointers in a pad. The current meanings (possibly to change a bit more soon) are: chainfunc: as always, chainfunc pointer is mirrored between peer pads (this may change, and the chain func may end up in pushfunc) pushfunc: SrcPad: gst_pad_pushfunc_proxy, cothread_switch to peer SinkPad: none (may take over chainfunc, see below) pullfunc: SrcPad: Src or Connection's function to construct buffers SinkPad: gst_pad_pullfunc_proxy, cothread_switch to peer There are a number of issues remaining with the scheduling, not the least of which is the fact that Connections are still dealt with the old way, with _push() functions and such. I'm trying to figure out a way to unify the system so it makes sense. Following the scheduling system is hard enough, trying to change it is murder. Another useful scheduling addition, mentioned above, is COTHREAD_STOPPING. It's an element flag that's used to signal whatever code is running in cothread context that it should be finishing up and exiting soon. An example of this is in plugins/cobin/spindentity.c. All the loops should now be composed of do/while loops, rather than while(1) loops: do { buf = gst_pad_pull(spindentity->sinkpad); gst_pad_push(spindentity->srcpad,buf); } while (!GST_ELEMENT_IS_COTHREAD_STOPPING(element)); The reason for this is that COTHREAD_STOPPING may be set before the above loop ever gets started. It wouldn't do for the body of the loop to never once get called, that would simply stall the pipeline. Note that only the core library code is ever responsible for setting and unsetting this flag. All elements have to do is respond to it by cleanly exiting the loop and the function holding it. This is needed primarily to allow iterations to occur properly. Basically, there's a single entry point in the cothread scheduling loop, gst_bin_iterate_func() simply switches to this cothread. If the element in this context is allowed to loop infinitely, nothing would even switch back to the context from which the iterate() was originally called. This is a bit of a problem. The solution is for there to be an implicit switch back to the originating context. Now, even I'm not sure exactly how this works, but if the cothread that's switched to actually returns, execution returns back to the calling context, i.e. iterate_func(). COTHREAD_STOPPING is therefore set just before switching into this (currently randomly chosen) context, on the assumption that it will return promptly after finishing its duties. The burden of clearing the flag falls to the various wrapper functions provided by the Bin code, thus element writers don't have to worry about doing that at all (and simply shouldn't). Related changes: All the sources in elements/ have been changed to reflect the new system. FIXMEs: 1) gstpipeline.c calls gst_src_push at some point, dunno why, it's commented out now. 2) any other sources, including vcdsrc, dvdsrc, and v4lsrc will break badly and need to be modified to work as pull-based sources.
2000-12-04 10:52:30 +00:00
(peer->chainfunc)(peer,buffer);
// else we squawk
} else {
g_print("-- gst_pad_push(): houston, we have a problem, no way of talking to peer\n");
}
}
/**
* gst_pad_pull:
* @pad: the pad to pull
*
* pulls a buffer along a sink pad
*
* Returns: the buffer that was pulled
*/
GstBuffer*
gst_pad_pull (GstPad *pad)
{
GstBuffer *buf;
g_return_val_if_fail(pad != NULL, NULL);
g_return_val_if_fail(GST_IS_PAD(pad), NULL);
/* check to see if the peer pad is disabled. return NULL if it is */
/* FIXME: this may be the wrong way to go about it */
if (GST_FLAG_IS_SET(pad->peer,GST_PAD_DISABLED)) {
g_print("gst_pad_pull: pad disabled, returning NULL\n");
return NULL;
}
// if no buffer in pen and there's a pull handler, fire it
if (pad->bufpen == NULL) {
if (pad->pullfunc != NULL) {
(pad->pullfunc)(pad->peer);
} else {
g_print("-- gst_pad_pull(%s:%s): no buffer in pen, and no handler to get one there!!!\n",
GST_ELEMENT(pad->parent)->name, pad->name);
}
}
// if there's a buffer in the holding pen, use it
if (pad->bufpen != NULL) {
buf = pad->bufpen;
pad->bufpen = NULL;
return buf;
// else we have a big problem...
} else {
g_print("-- gst_pad_pull(%s:%s): no buffer in pen, and no handler\n",
GST_ELEMENT(pad->parent)->name, pad->peer->name);
return NULL;
}
return NULL;
}
/**
* gst_pad_pull_region:
* @pad: the pad to pull
* @offset: the offset to pull
* @size: the size to pull
*
* pulls a buffer along a sink pad with a given offset and size
*
* Returns: the buffer that was pulled
*/
GstBuffer*
gst_pad_pull_region (GstPad *pad,
gulong offset,
gulong size)
{
GstBuffer *buf;
g_return_val_if_fail(pad != NULL, NULL);
g_return_val_if_fail(GST_IS_PAD(pad), NULL);
DEBUG("-- gst_pad_pull_region(%s:%s): region (%lu,%lu)\n",
GST_ELEMENT(pad->parent)->name, pad->peer->name,
offset, size);
// if no buffer in pen and there's a pull handler, fire it
if (pad->bufpen == NULL) {
if (pad->pullregionfunc != NULL) {
(pad->pullregionfunc)(pad->peer, offset, size);
} else {
g_print("-- gst_pad_pull_region(%s:%s): no buffer in pen, and no handler to get one there!!!\n",
GST_ELEMENT(pad->parent)->name, pad->name);
}
}
// if there's a buffer in the holding pen, use it
if (pad->bufpen != NULL) {
buf = pad->bufpen;
pad->bufpen = NULL;
return buf;
// else we have a big problem...
} else {
g_print("-- gst_pad_pull_region(%s:%s): no buffer in pen, and no handler\n",
GST_ELEMENT(pad->parent)->name, pad->peer->name);
return NULL;
}
return NULL;
}
/**
* gst_pad_chain:
* @pad: the pad to chain
*
* call the chain function of the given pad
*/
void
gst_pad_chain (GstPad *pad)
{
g_return_if_fail (pad != NULL);
g_return_if_fail (GST_IS_PAD (pad));
g_return_if_fail (pad->peer != NULL);
g_return_if_fail (pad->chainfunc != NULL);
if (pad->bufpen && pad->chainfunc)
(pad->chainfunc) (pad,pad->bufpen);
}
/**
* gst_pad_handle_qos:
* @pad: the pad to handle the QoS message
* @qos_message: the QoS message to handle
*
* pass the qos message downstream
*/
void
gst_pad_handle_qos(GstPad *pad,
glong qos_message)
{
GstElement *element;
GList *pads;
GstPad *target_pad;
DEBUG("gst_pad_handle_qos(\"%s\",%08ld)\n", GST_ELEMENT(pad->parent)->name,qos_message);
if (pad->qosfunc) {
(pad->qosfunc) (pad,qos_message);
}
else {
element = GST_ELEMENT (pad->peer->parent);
pads = element->pads;
DEBUG("gst_pad_handle_qos recurse(\"%s\",%08ld)\n", element->name,qos_message);
while (pads) {
target_pad = GST_PAD (pads->data);
if (target_pad->direction == GST_PAD_SINK) {
gst_pad_handle_qos (target_pad, qos_message);
}
pads = g_list_next (pads);
}
}
return;
}
/**
* gst_pad_disconnect:
* @srcpad: the source pad to disconnect
* @sinkpad: the sink pad to disconnect
*
* disconnects the source pad from the sink pad
*/
void
gst_pad_disconnect (GstPad *srcpad,
GstPad *sinkpad)
{
/* generic checks */
g_return_if_fail (srcpad != NULL);
g_return_if_fail (GST_IS_PAD (srcpad));
g_return_if_fail (srcpad->peer != NULL);
g_return_if_fail (sinkpad != NULL);
g_return_if_fail (GST_IS_PAD (sinkpad));
g_return_if_fail (sinkpad->peer != NULL);
g_return_if_fail ((srcpad->direction == GST_PAD_SRC) &&
(sinkpad->direction == GST_PAD_SINK));
/* first clear peers */
srcpad->peer = NULL;
sinkpad->peer = NULL;
}
/**
* gst_pad_connect:
* @srcpad: the source pad to connect
* @sinkpad: the sink pad to connect
*
* connects the source pad to the sink pad
*/
void
gst_pad_connect (GstPad *srcpad,
GstPad *sinkpad)
{
GstPad *temppad;
/* generic checks */
g_return_if_fail(srcpad != NULL);
g_return_if_fail(GST_IS_PAD(srcpad));
g_return_if_fail(srcpad->peer == NULL);
g_return_if_fail(sinkpad != NULL);
g_return_if_fail(GST_IS_PAD(sinkpad));
g_return_if_fail(sinkpad->peer == NULL);
// g_return_if_fail(sinkpad->chain != NULL);
/* check for reversed directions and swap if necessary */
if ((srcpad->direction == GST_PAD_SINK) &&
(sinkpad->direction == GST_PAD_SRC)) {
temppad = srcpad;
srcpad = sinkpad;
sinkpad = temppad;
}
g_return_if_fail((srcpad->direction == GST_PAD_SRC) &&
(sinkpad->direction == GST_PAD_SINK));
/* first set peers */
srcpad->peer = sinkpad;
sinkpad->peer = srcpad;
/* now copy the chain pointer from sink to src */
srcpad->chainfunc = sinkpad->chainfunc;
/* and the pull function */
//srcpad->pullfunc = sinkpad->pullfunc;
/* set the connected flag */
/* FIXME: set connected flag */
}
/**
* gst_pad_set_parent:
* @pad: the pad to set the parent
* @parent: the object to set the parent to
*
* sets the parent object of a pad.
*/
void
gst_pad_set_parent (GstPad *pad,
GstObject *parent)
{
g_return_if_fail (pad != NULL);
g_return_if_fail (GST_IS_PAD (pad));
g_return_if_fail (pad->parent == NULL);
g_return_if_fail (parent != NULL);
g_return_if_fail (GTK_IS_OBJECT (parent));
g_return_if_fail ((gpointer)pad != (gpointer)parent);
//g_print("set parent %s\n", gst_element_get_name(parent));
pad->parent = parent;
}
/**
* gst_pad_add_ghost_parent:
* @pad: the pad to set the ghost parent
* @parent: the object to set the ghost parent to
*
* add a ghost parent object to a pad.
*/
void
gst_pad_add_ghost_parent (GstPad *pad,
GstObject *parent)
{
g_return_if_fail (pad != NULL);
g_return_if_fail (GST_IS_PAD (pad));
g_return_if_fail (parent != NULL);
g_return_if_fail (GTK_IS_OBJECT (parent));
pad->ghostparents = g_list_prepend (pad->ghostparents, parent);
}
/**
* gst_pad_remove_ghost_parent:
* @pad: the pad to remove the ghost parent
* @parent: the object to remove the ghost parent from
*
* remove a ghost parent object from a pad.
*/
void
gst_pad_remove_ghost_parent (GstPad *pad,
GstObject *parent)
{
g_return_if_fail (pad != NULL);
g_return_if_fail (GST_IS_PAD (pad));
g_return_if_fail (parent != NULL);
g_return_if_fail (GTK_IS_OBJECT (parent));
pad->ghostparents = g_list_remove (pad->ghostparents, parent);
}
/**
* gst_pad_get_parent:
* @pad: the pad to get the parent from
*
* get the parent object of this pad
*
* Returns: the parent object
*/
GstObject*
gst_pad_get_parent (GstPad *pad)
{
g_return_val_if_fail (pad != NULL, NULL);
g_return_val_if_fail (GST_IS_PAD (pad), NULL);
return pad->parent;
}
/**
* gst_pad_get_ghost_parents:
* @pad: the pad to get the ghost parents from
*
* get the ghost parents of this pad
*
* Returns: a list of ghost parent objects
*/
GList*
gst_pad_get_ghost_parents (GstPad *pad)
{
g_return_val_if_fail (pad != NULL, NULL);
g_return_val_if_fail (GST_IS_PAD (pad), NULL);
return pad->ghostparents;
}
/**
* gst_pad_get_type_ids:
* @pad: the pad to get the type ids from
*
* get the list of types for this pad
*
* Returns: a GList of types of this pad
*/
GList*
gst_pad_get_type_ids (GstPad *pad)
{
g_return_val_if_fail (pad != NULL, 0);
g_return_val_if_fail (GST_IS_PAD (pad), 0);
return pad->types;
}
// FIXME remove...
void
gst_pad_set_type_id (GstPad *pad,
guint16 id)
{
gst_pad_add_type_id (pad, id);
}
/**
* gst_pad_set_type_id:
* @pad: the pad to set the type id to
* @id: the type id to set this pad to
*
* set the type of this pad
*/
void
gst_pad_add_type_id (GstPad *pad,
guint16 id)
{
g_return_if_fail (pad != NULL);
g_return_if_fail (GST_IS_PAD (pad));
g_return_if_fail (gst_type_find_by_id (id) != NULL);
pad->types = g_list_append(pad->types, GINT_TO_POINTER((gint)id));
}
/**
* gst_pad_get_peer:
* @pad: the pad to get the peer from
*
* Get the peer pad of this pad
*
* Returns: the peer pad
*/
GstPad*
gst_pad_get_peer (GstPad *pad)
{
g_return_val_if_fail (pad != NULL, NULL);
g_return_val_if_fail (GST_IS_PAD (pad), NULL);
return pad->peer;
}
static void
gst_pad_real_destroy (GtkObject *object)
{
GstPad *pad = GST_PAD (object);
// g_print("in gst_pad_real_destroy()\n");
if (pad->name)
g_free (pad->name);
g_list_free (pad->ghostparents);
}
/**
* gst_pad_load_and_connect:
* @parent: the parent XML node to read the description from
* @element: the element that has the source pad
* @elements: a hashtable with elements
*
* Read the pad definition from the XML node and connect the given pad
* in element to a pad of an element in the hashtable.
*/
void
gst_pad_load_and_connect (xmlNodePtr parent,
GstObject *element,
GHashTable *elements)
{
xmlNodePtr field = parent->childs;
GstPad *pad = NULL, *targetpad;
guchar *peer = NULL;
gchar **split;
GstElement *target;
while (field) {
if (!strcmp(field->name, "name")) {
pad = gst_element_get_pad(GST_ELEMENT(element), xmlNodeGetContent(field));
}
else if (!strcmp(field->name, "peer")) {
peer = g_strdup(xmlNodeGetContent(field));
}
field = field->next;
}
g_return_if_fail(pad != NULL);
g_return_if_fail(peer != NULL);
split = g_strsplit(peer, ".", 2);
g_return_if_fail(split[0] != NULL);
g_return_if_fail(split[1] != NULL);
target = (GstElement *)g_hash_table_lookup(elements, split[0]);
if (target == NULL) goto cleanup;
targetpad = gst_element_get_pad(target, split[1]);
g_return_if_fail(targetpad != NULL);
gst_pad_connect(pad, targetpad);
cleanup:
g_strfreev(split);
}
/**
* gst_pad_save_thyself:
* @pad: the pad to save
* @parent: the parent XML node to save the description in
*
* Saves the pad into an xml representation
*
* Returns: the xml representation of the pad
*/
xmlNodePtr
gst_pad_save_thyself (GstPad *pad,
xmlNodePtr parent)
{
GstPad *peer;
xmlNewChild(parent,NULL,"name",pad->name);
if (pad->peer != NULL) {
peer = pad->peer;
// first check to see if the peer's parent's parent is the same
//if (pad->parent->parent == peer->parent->parent)
// we just save it off
xmlNewChild(parent,NULL,"peer",g_strdup_printf("%s.%s",
GST_ELEMENT(peer->parent)->name,peer->name));
} else
xmlNewChild(parent,NULL,"peer","");
return parent;
}
/**
* gst_pad_ghost_save_thyself:
* @pad: the pad to save
* @bin: the bin
* @parent: the parent XML node to save the description in
*
* Saves the ghost pad into an xml representation
*
* Returns: the xml representation of the pad
*/
xmlNodePtr
gst_pad_ghost_save_thyself (GstPad *pad,
GstElement *bin,
xmlNodePtr parent)
{
xmlNodePtr self;
self = xmlNewChild(parent,NULL,"ghostpad",NULL);
xmlNewChild(self,NULL,"name",pad->name);
xmlNewChild(self,NULL,"parent",GST_ELEMENT(pad->parent)->name);
return self;
}