gstreamer/gst/schedulers/gstbasicscheduler.c

1433 lines
46 KiB
C
Raw Normal View History

- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
/* GStreamer
* Copyright (C) 1999,2000 Erik Walthinsen <omega@cse.ogi.edu>
* 2000 Wim Taymans <wtay@chello.be>
*
* gstscheduler.c: Default scheduling code for most cases
*
* 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 GST_DEBUG_ENABLED */
#include "../gst.h"
#include "cothreads_compat.h"
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
typedef struct _GstSchedulerChain GstSchedulerChain;
#define GST_ELEMENT_THREADSTATE(elem) (cothread*) (GST_ELEMENT_CAST (elem)->sched_private)
#define GST_RPAD_BUFPEN(pad) (GstBuffer*) (GST_REAL_PAD_CAST(pad)->sched_private)
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
#define GST_ELEMENT_COTHREAD_STOPPING GST_ELEMENT_SCHEDULER_PRIVATE1
#define GST_ELEMENT_IS_COTHREAD_STOPPING(element) GST_FLAG_IS_SET((element), GST_ELEMENT_COTHREAD_STOPPING)
typedef struct _GstBasicScheduler GstBasicScheduler;
typedef struct _GstBasicSchedulerClass GstBasicSchedulerClass;
#ifdef _COTHREADS_STANDARD
# define _SCHEDULER_NAME "standard"
#else
# define _SCHEDULER_NAME "basic"
#endif
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
struct _GstSchedulerChain {
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
GstBasicScheduler *sched;
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
GList *disabled;
GList *elements;
gint num_elements;
GstElement *entry;
gint cothreaded_elements;
gboolean schedule;
};
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
#define GST_TYPE_BASIC_SCHEDULER \
(gst_basic_scheduler_get_type())
#define GST_BASIC_SCHEDULER(obj) \
(G_TYPE_CHECK_INSTANCE_CAST((obj),GST_TYPE_BASIC_SCHEDULER,GstBasicScheduler))
#define GST_BASIC_SCHEDULER_CLASS(klass) \
(G_TYPE_CHECK_CLASS_CAST((klass),GST_TYPE_BASIC_SCHEDULER,GstBasicSchedulerClass))
#define GST_IS_BASIC_SCHEDULER(obj) \
(G_TYPE_CHECK_INSTANCE_TYPE((obj),GST_TYPE_BASIC_SCHEDULER))
#define GST_IS_BASIC_SCHEDULER_CLASS(obj) \
(G_TYPE_CHECK_CLASS_TYPE((klass),GST_TYPE_BASIC_SCHEDULER))
#define GST_BASIC_SCHEDULER_CAST(sched) ((GstBasicScheduler *)(sched))
#define SCHED(element) GST_BASIC_SCHEDULER_CAST (GST_ELEMENT_SCHED (element))
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
typedef enum {
GST_BASIC_SCHEDULER_STATE_NONE,
GST_BASIC_SCHEDULER_STATE_STOPPED,
GST_BASIC_SCHEDULER_STATE_ERROR,
GST_BASIC_SCHEDULER_STATE_RUNNING,
} GstBasicSchedulerState;
typedef enum {
/* something important has changed inside the scheduler */
GST_BASIC_SCHEDULER_CHANGE = GST_SCHEDULER_FLAG_LAST,
} GstBasicSchedulerFlags;
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
struct _GstBasicScheduler {
GstScheduler parent;
GList *elements;
gint num_elements;
GList *chains;
gint num_chains;
GstBasicSchedulerState state;
cothread_context *context;
GstElement *current;
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
};
struct _GstBasicSchedulerClass {
GstSchedulerClass parent_class;
};
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
static GType _gst_basic_scheduler_type = 0;
static void gst_basic_scheduler_class_init (GstBasicSchedulerClass * klass);
static void gst_basic_scheduler_init (GstBasicScheduler * scheduler);
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
static void gst_basic_scheduler_dispose (GObject *object);
static void gst_basic_scheduler_setup (GstScheduler *sched);
static void gst_basic_scheduler_reset (GstScheduler *sched);
static void gst_basic_scheduler_add_element (GstScheduler *sched, GstElement *element);
static void gst_basic_scheduler_remove_element (GstScheduler *sched, GstElement *element);
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
static GstElementStateReturn
gst_basic_scheduler_state_transition (GstScheduler *sched, GstElement *element, gint transition);
static void gst_basic_scheduler_lock_element (GstScheduler *sched, GstElement *element);
static void gst_basic_scheduler_unlock_element (GstScheduler *sched, GstElement *element);
static void gst_basic_scheduler_yield (GstScheduler *sched, GstElement *element);
static gboolean gst_basic_scheduler_interrupt (GstScheduler *sched, GstElement *element);
static void gst_basic_scheduler_error (GstScheduler *sched, GstElement *element);
static void gst_basic_scheduler_pad_connect (GstScheduler *sched, GstPad *srcpad, GstPad *sinkpad);
static void gst_basic_scheduler_pad_disconnect (GstScheduler *sched, GstPad *srcpad, GstPad *sinkpad);
static GstPad* gst_basic_scheduler_pad_select (GstScheduler *sched, GList *padlist);
static GstClockReturn gst_basic_scheduler_clock_wait (GstScheduler *sched, GstElement *element,
GstClock *clock, GstClockTime time, GstClockTimeDiff *jitter);
static GstSchedulerState
gst_basic_scheduler_iterate (GstScheduler *sched);
static void gst_basic_scheduler_show (GstScheduler *sched);
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
static GstSchedulerClass *parent_class = NULL;
/* for threaded bins, these pre- and post-run functions lock and unlock the
* elements. we have to avoid deadlocks, so we make these convenience macros
* that will avoid using do_cothread_switch from within the scheduler. */
#define do_element_switch(element) G_STMT_START{ \
GstElement *from = SCHED (element)->current; \
if (from && from->post_run_func) \
from->post_run_func (from); \
SCHED (element)->current = element; \
if (element->pre_run_func) \
element->pre_run_func (element); \
do_cothread_switch (GST_ELEMENT_THREADSTATE (element)); \
}G_STMT_END
#define do_switch_to_main(sched) G_STMT_START{ \
GstElement *current = ((GstBasicScheduler*)sched)->current; \
if (current && current->post_run_func) \
current->post_run_func (current); \
SCHED (current)->current = NULL; \
do_cothread_switch \
(do_cothread_get_main \
(((GstBasicScheduler*)sched)->context)); \
}G_STMT_END
#define do_switch_from_main(entry) G_STMT_START{ \
if (entry->pre_run_func) \
entry->pre_run_func (entry); \
SCHED (entry)->current = entry; \
do_cothread_switch (GST_ELEMENT_THREADSTATE (entry)); \
}G_STMT_END
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
static GType
gst_basic_scheduler_get_type (void)
{
if (!_gst_basic_scheduler_type) {
static const GTypeInfo scheduler_info = {
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
sizeof (GstBasicSchedulerClass),
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
NULL,
NULL,
(GClassInitFunc) gst_basic_scheduler_class_init,
NULL,
NULL,
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
sizeof (GstBasicScheduler),
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
0,
(GInstanceInitFunc) gst_basic_scheduler_init,
NULL
};
_gst_basic_scheduler_type = g_type_register_static (GST_TYPE_SCHEDULER, "Gst"COTHREADS_NAME_CAPITAL"Scheduler", &scheduler_info, 0);
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
}
return _gst_basic_scheduler_type;
}
static void
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
gst_basic_scheduler_class_init (GstBasicSchedulerClass * klass)
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
{
GObjectClass *gobject_class;
GstObjectClass *gstobject_class;
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
GstSchedulerClass *gstscheduler_class;
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
gobject_class = (GObjectClass*)klass;
gstobject_class = (GstObjectClass*)klass;
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
gstscheduler_class = (GstSchedulerClass*)klass;
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
parent_class = g_type_class_ref (GST_TYPE_SCHEDULER);
gobject_class->dispose = GST_DEBUG_FUNCPTR (gst_basic_scheduler_dispose);
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
gstscheduler_class->setup = GST_DEBUG_FUNCPTR (gst_basic_scheduler_setup);
gstscheduler_class->reset = GST_DEBUG_FUNCPTR (gst_basic_scheduler_reset);
gstscheduler_class->add_element = GST_DEBUG_FUNCPTR (gst_basic_scheduler_add_element);
gstscheduler_class->remove_element = GST_DEBUG_FUNCPTR (gst_basic_scheduler_remove_element);
gstscheduler_class->state_transition = GST_DEBUG_FUNCPTR (gst_basic_scheduler_state_transition);
gstscheduler_class->lock_element = GST_DEBUG_FUNCPTR (gst_basic_scheduler_lock_element);
gstscheduler_class->unlock_element = GST_DEBUG_FUNCPTR (gst_basic_scheduler_unlock_element);
gstscheduler_class->yield = GST_DEBUG_FUNCPTR (gst_basic_scheduler_yield);
gstscheduler_class->interrupt = GST_DEBUG_FUNCPTR (gst_basic_scheduler_interrupt);
gstscheduler_class->error = GST_DEBUG_FUNCPTR (gst_basic_scheduler_error);
gstscheduler_class->pad_connect = GST_DEBUG_FUNCPTR (gst_basic_scheduler_pad_connect);
gstscheduler_class->pad_disconnect = GST_DEBUG_FUNCPTR (gst_basic_scheduler_pad_disconnect);
gstscheduler_class->pad_select = GST_DEBUG_FUNCPTR (gst_basic_scheduler_pad_select);
gstscheduler_class->clock_wait = GST_DEBUG_FUNCPTR (gst_basic_scheduler_clock_wait);
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
gstscheduler_class->iterate = GST_DEBUG_FUNCPTR (gst_basic_scheduler_iterate);
gstscheduler_class->show = GST_DEBUG_FUNCPTR (gst_basic_scheduler_show);
do_cothreads_init(NULL);
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
}
static void
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
gst_basic_scheduler_init (GstBasicScheduler *scheduler)
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
{
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
scheduler->elements = NULL;
scheduler->num_elements = 0;
scheduler->chains = NULL;
scheduler->num_chains = 0;
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
}
static void
gst_basic_scheduler_dispose (GObject *object)
{
G_OBJECT_CLASS (parent_class)->dispose (object);
}
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
static gboolean
plugin_init (GModule *module, GstPlugin *plugin)
{
GstSchedulerFactory *factory;
gst_plugin_set_longname (plugin, "A basic scheduler");
factory = gst_scheduler_factory_new ("basic"COTHREADS_NAME,
"A basic scheduler using "COTHREADS_NAME" cothreads",
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
gst_basic_scheduler_get_type());
if (factory != NULL) {
gst_plugin_add_feature (plugin, GST_PLUGIN_FEATURE (factory));
}
else {
g_warning ("could not register scheduler: "COTHREADS_NAME);
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
}
return TRUE;
}
GstPluginDesc plugin_desc = {
GST_VERSION_MAJOR,
GST_VERSION_MINOR,
"gstbasic"COTHREADS_NAME"scheduler",
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
plugin_init
};
static int
gst_basic_scheduler_loopfunc_wrapper (int argc, char *argv[])
{
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
GstElement *element = GST_ELEMENT_CAST (argv);
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
G_GNUC_UNUSED const gchar *name = GST_ELEMENT_NAME (element);
GST_DEBUG_ENTER ("(%d,'%s')", argc, name);
gst_object_ref (GST_OBJECT (element));
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
do {
GST_DEBUG (GST_CAT_DATAFLOW, "calling loopfunc %s for element %s",
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
GST_DEBUG_FUNCPTR_NAME (element->loopfunc), name);
(element->loopfunc) (element);
GST_DEBUG (GST_CAT_DATAFLOW, "element %s ended loop function", name);
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
} while (!GST_ELEMENT_IS_COTHREAD_STOPPING (element));
GST_FLAG_UNSET (element, GST_ELEMENT_COTHREAD_STOPPING);
/* due to oddities in the cothreads code, when this function returns it will
* switch to the main cothread. thus, we need to unlock the current element. */
if (SCHED (element)) {
if (SCHED (element)->current && SCHED (element)->current->post_run_func) {
SCHED (element)->current->post_run_func (SCHED (element)->current);
}
SCHED (element)->current = NULL;
}
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
GST_DEBUG_LEAVE ("(%d,'%s')", argc, name);
gst_object_unref (GST_OBJECT (element));
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
return 0;
}
static int
gst_basic_scheduler_chain_wrapper (int argc, char *argv[])
{
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
GstElement *element = GST_ELEMENT_CAST (argv);
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
G_GNUC_UNUSED const gchar *name = GST_ELEMENT_NAME (element);
GST_DEBUG_ENTER ("(\"%s\")", name);
GST_DEBUG (GST_CAT_DATAFLOW, "stepping through pads");
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
gst_object_ref (GST_OBJECT (element));
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
do {
GList *pads = element->pads;
while (pads) {
GstPad *pad = GST_PAD (pads->data);
GstRealPad *realpad;
pads = g_list_next (pads);
if (!GST_IS_REAL_PAD (pad))
continue;
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
realpad = GST_REAL_PAD_CAST (pad);
if (GST_RPAD_DIRECTION (realpad) == GST_PAD_SINK &&
GST_PAD_IS_USABLE (realpad)) {
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
GstBuffer *buf;
GST_DEBUG (GST_CAT_DATAFLOW, "pulling data from %s:%s", name,
GST_PAD_NAME (pad));
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
buf = gst_pad_pull (pad);
if (buf) {
if (GST_IS_EVENT (buf) && !GST_ELEMENT_IS_EVENT_AWARE (element)) {
gst_pad_send_event (pad, GST_EVENT (buf));
}
else {
GST_DEBUG (GST_CAT_DATAFLOW, "calling chain function of %s:%s %p",
name, GST_PAD_NAME (pad), buf);
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
GST_RPAD_CHAINFUNC (realpad) (pad, buf);
GST_DEBUG (GST_CAT_DATAFLOW,
"calling chain function of element %s done", name);
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
}
}
}
}
} while (!GST_ELEMENT_IS_COTHREAD_STOPPING (element));
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
GST_FLAG_UNSET (element, GST_ELEMENT_COTHREAD_STOPPING);
/* due to oddities in the cothreads code, when this function returns it will
* switch to the main cothread. thus, we need to unlock the current element. */
if (SCHED (element)) {
if (SCHED (element)->current && SCHED (element)->current->post_run_func) {
SCHED (element)->current->post_run_func (SCHED (element)->current);
}
SCHED (element)->current = NULL;
}
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
GST_DEBUG_LEAVE ("(%d,'%s')", argc, name);
gst_object_unref (GST_OBJECT (element));
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
return 0;
}
static int
gst_basic_scheduler_src_wrapper (int argc, char *argv[])
{
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
GstElement *element = GST_ELEMENT_CAST (argv);
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
GList *pads;
GstRealPad *realpad;
GstBuffer *buf = NULL;
G_GNUC_UNUSED const gchar *name = GST_ELEMENT_NAME (element);
GST_DEBUG_ENTER ("(%d,\"%s\")", argc, name);
do {
pads = element->pads;
while (pads) {
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
if (!GST_IS_REAL_PAD (pads->data))
continue;
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
realpad = GST_REAL_PAD_CAST (pads->data);
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
pads = g_list_next (pads);
if (GST_RPAD_DIRECTION (realpad) == GST_PAD_SRC && GST_PAD_IS_USABLE (realpad)) {
GST_DEBUG (GST_CAT_DATAFLOW, "calling _getfunc for %s:%s", GST_DEBUG_PAD_NAME (realpad));
g_return_val_if_fail (GST_RPAD_GETFUNC (realpad) != NULL, 0);
buf = GST_RPAD_GETFUNC (realpad) (GST_PAD_CAST (realpad));
if (buf) {
GST_DEBUG (GST_CAT_DATAFLOW, "calling gst_pad_push on pad %s:%s %p",
GST_DEBUG_PAD_NAME (realpad), buf);
gst_pad_push (GST_PAD_CAST (realpad), buf);
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
}
}
}
} while (!GST_ELEMENT_IS_COTHREAD_STOPPING (element));
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
GST_FLAG_UNSET (element, GST_ELEMENT_COTHREAD_STOPPING);
/* due to oddities in the cothreads code, when this function returns it will
* switch to the main cothread. thus, we need to unlock the current element. */
if (SCHED (element)->current->post_run_func)
SCHED (element)->current->post_run_func (SCHED (element)->current);
SCHED (element)->current = NULL;
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
GST_DEBUG_LEAVE ("");
return 0;
}
static void
gst_basic_scheduler_chainhandler_proxy (GstPad * pad, GstBuffer * buf)
{
gint loop_count = 100;
GstElement *parent;
GstRealPad *peer;
parent = GST_PAD_PARENT (pad);
peer = GST_RPAD_PEER (pad);
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
GST_DEBUG_ENTER ("(%s:%s)", GST_DEBUG_PAD_NAME (pad));
GST_DEBUG (GST_CAT_DATAFLOW, "putting buffer %p in peer \"%s:%s\"'s pen", buf,
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
GST_DEBUG_PAD_NAME (peer));
/*
* loop until the bufferpen is empty so we can fill it up again
*/
while (GST_RPAD_BUFPEN (GST_RPAD_PEER (pad)) != NULL && --loop_count) {
GST_DEBUG (GST_CAT_DATAFLOW, "switching to %p to empty bufpen %d",
GST_ELEMENT_THREADSTATE (parent), loop_count);
do_element_switch (parent);
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
/* we may no longer be the same pad, check. */
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
if (GST_RPAD_PEER (peer) != (GstRealPad *) pad) {
GST_DEBUG (GST_CAT_DATAFLOW, "new pad in mid-switch!");
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
pad = (GstPad *) GST_RPAD_PEER (peer);
}
parent = GST_PAD_PARENT (pad);
peer = GST_RPAD_PEER (pad);
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
}
if (loop_count == 0) {
gst_element_error (parent,
"(internal error) basic: maximum number of switches exceeded");
return;
}
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
g_assert (GST_RPAD_BUFPEN (GST_RPAD_PEER (pad)) == NULL);
/* now fill the bufferpen and switch so it can be consumed */
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
GST_RPAD_BUFPEN (GST_RPAD_PEER (pad)) = buf;
GST_DEBUG (GST_CAT_DATAFLOW, "switching to %p to consume buffer %p",
GST_ELEMENT_THREADSTATE (GST_PAD_PARENT (pad)), buf);
do_element_switch (parent);
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
GST_DEBUG (GST_CAT_DATAFLOW, "done switching");
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
}
static void
gst_basic_scheduler_select_proxy (GstPad * pad, GstBuffer * buf)
{
GstElement *parent;
parent = GST_PAD_PARENT (pad);
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
GST_DEBUG_ENTER ("(%s:%s)", GST_DEBUG_PAD_NAME (pad));
GST_DEBUG (GST_CAT_DATAFLOW, "putting buffer %p in peer's pen", buf);
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
g_assert (GST_RPAD_BUFPEN (GST_RPAD_PEER (pad)) == NULL);
/* now fill the bufferpen and switch so it can be consumed */
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
GST_RPAD_BUFPEN (GST_RPAD_PEER (pad)) = buf;
GST_DEBUG (GST_CAT_DATAFLOW, "switching to %p",
GST_ELEMENT_THREADSTATE (parent));
parent->select_pad = pad;
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
do_element_switch (parent);
GST_DEBUG (GST_CAT_DATAFLOW, "done switching");
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
}
static GstBuffer *
gst_basic_scheduler_gethandler_proxy (GstPad * pad)
{
GstBuffer *buf;
GstElement *parent;
GstRealPad *peer;
GST_DEBUG_ENTER ("(%s:%s)", GST_DEBUG_PAD_NAME (pad));
parent = GST_PAD_PARENT (pad);
peer = GST_RPAD_PEER (pad);
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
/* FIXME this should be bounded */
/* we will loop switching to the peer until it's filled up the bufferpen */
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
while (GST_RPAD_BUFPEN (pad) == NULL) {
GST_DEBUG (GST_CAT_DATAFLOW, "switching to \"%s\": %p to fill bufpen",
GST_ELEMENT_NAME (parent),
GST_ELEMENT_THREADSTATE (parent));
do_element_switch (parent);
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
/* we may no longer be the same pad, check. */
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
if (GST_RPAD_PEER (peer) != (GstRealPad *) pad) {
GST_DEBUG (GST_CAT_DATAFLOW, "new pad in mid-switch!");
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
pad = (GstPad *) GST_RPAD_PEER (peer);
if (!pad) {
gst_element_error (parent, "pad unconnected");
}
parent = GST_PAD_PARENT (pad);
peer = GST_RPAD_PEER (pad);
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
}
}
GST_DEBUG (GST_CAT_DATAFLOW, "done switching");
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
/* now grab the buffer from the pen, clear the pen, and return the buffer */
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
buf = GST_RPAD_BUFPEN (pad);
GST_RPAD_BUFPEN (pad) = NULL;
return buf;
}
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
static gboolean
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
gst_basic_scheduler_cothreaded_chain (GstBin * bin, GstSchedulerChain * chain)
{
GList *elements;
GstElement *element;
cothread_func wrapper_function;
const GList *pads;
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
GstPad *pad;
GST_DEBUG (GST_CAT_SCHEDULING, "chain is using COTHREADS");
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
g_assert (chain->sched->context != NULL);
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
/* walk through all the chain's elements */
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
elements = chain->elements;
while (elements) {
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
gboolean decoupled;
element = GST_ELEMENT_CAST (elements->data);
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
elements = g_list_next (elements);
decoupled = GST_FLAG_IS_SET (element, GST_ELEMENT_DECOUPLED);
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
/* start out without a wrapper function, we select it later */
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
wrapper_function = NULL;
/* if the element has a loopfunc... */
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
if (element->loopfunc != NULL) {
wrapper_function = GST_DEBUG_FUNCPTR (gst_basic_scheduler_loopfunc_wrapper);
GST_DEBUG (GST_CAT_SCHEDULING, "element '%s' is a loop-based",
GST_ELEMENT_NAME (element));
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
}
else {
/* otherwise we need to decide what kind of cothread
* if it's not DECOUPLED, we decide based on
* whether it's a source or not */
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
if (!decoupled) {
/* if it doesn't have any sinks, it must be a source (duh) */
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
if (element->numsinkpads == 0) {
wrapper_function = GST_DEBUG_FUNCPTR (gst_basic_scheduler_src_wrapper);
GST_DEBUG (GST_CAT_SCHEDULING,
"element '%s' is a source, using _src_wrapper",
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
GST_ELEMENT_NAME (element));
}
else {
wrapper_function = GST_DEBUG_FUNCPTR (gst_basic_scheduler_chain_wrapper);
GST_DEBUG (GST_CAT_SCHEDULING,
"element '%s' is a filter, using _chain_wrapper",
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
GST_ELEMENT_NAME (element));
}
}
}
/* now we have to walk through the pads to set up their state */
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
pads = gst_element_get_pad_list (element);
while (pads) {
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
GstPad *peerpad;
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
pad = GST_PAD (pads->data);
pads = g_list_next (pads);
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
if (!GST_IS_REAL_PAD (pad))
continue;
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
peerpad = GST_PAD_PEER (pad);
if (peerpad) {
GstElement *peerelement = GST_ELEMENT_CAST (GST_PAD_PARENT (peerpad));
gboolean different_sched = (peerelement->sched != GST_SCHEDULER (chain->sched));
gboolean peer_decoupled = GST_FLAG_IS_SET (peerelement, GST_ELEMENT_DECOUPLED);
GST_DEBUG (GST_CAT_SCHEDULING,
"inspecting pad %s:%s", GST_DEBUG_PAD_NAME (peerpad));
/* we don't need to check this for decoupled elements */
if (!decoupled) {
/* if the peer element is in another schedule,
* it's not decoupled and we are not decoupled
* either, we have an error */
if (different_sched && !peer_decoupled)
{
gst_element_error (element,
"element \"%s\" is not decoupled but has pads "
"in different schedulers",
GST_ELEMENT_NAME (element), NULL);
return FALSE;
}
/* ok, the peer is in a different scheduler and is decoupled,
* we need to set the
* handlers so we can talk with it */
else if (different_sched) {
if (GST_RPAD_DIRECTION (peerpad) == GST_PAD_SINK) {
GST_DEBUG (GST_CAT_SCHEDULING,
"copying chain func into push proxy for peer %s:%s",
GST_DEBUG_PAD_NAME (peerpad));
GST_RPAD_CHAINHANDLER (peerpad) = GST_RPAD_CHAINFUNC (peerpad);
}
else {
GST_DEBUG (GST_CAT_SCHEDULING,
"copying get func into pull proxy for peer %s:%s",
GST_DEBUG_PAD_NAME (peerpad));
GST_RPAD_GETHANDLER (peerpad) = GST_RPAD_GETFUNC (peerpad);
}
}
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
}
}
/* if the element is DECOUPLED or outside the manager, we have to chain */
if (decoupled) {
/* set the chain proxies */
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
if (GST_RPAD_DIRECTION (pad) == GST_PAD_SINK) {
GST_DEBUG (GST_CAT_SCHEDULING,
"copying chain function into push proxy for %s:%s",
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
GST_DEBUG_PAD_NAME (pad));
GST_RPAD_CHAINHANDLER (pad) = GST_RPAD_CHAINFUNC (pad);
}
else {
GST_DEBUG (GST_CAT_SCHEDULING,
"copying get function into pull proxy for %s:%s",
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
GST_DEBUG_PAD_NAME (pad));
GST_RPAD_GETHANDLER (pad) = GST_RPAD_GETFUNC (pad);
}
}
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
/* otherwise we really are a cothread */
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
else {
if (GST_RPAD_DIRECTION (pad) == GST_PAD_SINK) {
GST_DEBUG (GST_CAT_SCHEDULING,
"setting cothreaded push proxy for sinkpad %s:%s",
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
GST_DEBUG_PAD_NAME (pad));
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
GST_RPAD_CHAINHANDLER (pad) = GST_DEBUG_FUNCPTR (gst_basic_scheduler_chainhandler_proxy);
}
else {
GST_DEBUG (GST_CAT_SCHEDULING,
"setting cothreaded pull proxy for srcpad %s:%s",
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
GST_DEBUG_PAD_NAME (pad));
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
GST_RPAD_GETHANDLER (pad) = GST_DEBUG_FUNCPTR (gst_basic_scheduler_gethandler_proxy);
}
}
}
/* need to set up the cothread now */
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
if (wrapper_function != NULL) {
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
if (GST_ELEMENT_THREADSTATE (element) == NULL) {
GST_DEBUG (GST_CAT_SCHEDULING,
"about to create a cothread, wrapper for '%s' is &%s",
GST_ELEMENT_NAME (element),
GST_DEBUG_FUNCPTR_NAME (wrapper_function));
do_cothread_create (GST_ELEMENT_THREADSTATE (element),
chain->sched->context,
wrapper_function, 0, (char **) element);
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
if (GST_ELEMENT_THREADSTATE (element) == NULL) {
gst_element_error (element, "could not create cothread for \"%s\"",
GST_ELEMENT_NAME (element), NULL);
return FALSE;
}
GST_DEBUG (GST_CAT_SCHEDULING, "created cothread %p for '%s'",
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
GST_ELEMENT_THREADSTATE (element),
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
GST_ELEMENT_NAME (element));
} else {
/* set the cothread wrapper function */
GST_DEBUG (GST_CAT_SCHEDULING,
"about to set the wrapper function for '%s' to &%s",
GST_ELEMENT_NAME (element),
GST_DEBUG_FUNCPTR_NAME (wrapper_function));
do_cothread_setfunc (GST_ELEMENT_THREADSTATE (element),
chain->sched->context,
wrapper_function, 0, (char **) element);
GST_DEBUG (GST_CAT_SCHEDULING, "set wrapper function for '%s' to &%s",
GST_ELEMENT_NAME (element),
GST_DEBUG_FUNCPTR_NAME (wrapper_function));
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
}
}
}
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
return TRUE;
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
}
static GstSchedulerChain *
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
gst_basic_scheduler_chain_new (GstBasicScheduler * sched)
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
{
GstSchedulerChain *chain = g_new (GstSchedulerChain, 1);
/* initialize the chain with sane values */
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
chain->sched = sched;
chain->disabled = NULL;
chain->elements = NULL;
chain->num_elements = 0;
chain->entry = NULL;
chain->cothreaded_elements = 0;
chain->schedule = FALSE;
/* add the chain to the schedulers' list of chains */
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
sched->chains = g_list_prepend (sched->chains, chain);
sched->num_chains++;
/* notify the scheduler that something changed */
GST_FLAG_SET(sched, GST_BASIC_SCHEDULER_CHANGE);
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
GST_INFO (GST_CAT_SCHEDULING, "created new chain %p, now are %d chains in sched %p",
chain, sched->num_chains, sched);
return chain;
}
static void
gst_basic_scheduler_chain_destroy (GstSchedulerChain * chain)
{
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
GstBasicScheduler *sched = chain->sched;
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
/* remove the chain from the schedulers' list of chains */
sched->chains = g_list_remove (sched->chains, chain);
sched->num_chains--;
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
/* destroy the chain */
g_list_free (chain->disabled); /* should be empty... */
g_list_free (chain->elements); /* ditto */
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
GST_INFO (GST_CAT_SCHEDULING, "destroyed chain %p, now are %d chains in sched %p", chain,
sched->num_chains, sched);
g_free (chain);
/* notify the scheduler that something changed */
GST_FLAG_SET(sched, GST_BASIC_SCHEDULER_CHANGE);
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
}
static void
gst_basic_scheduler_chain_add_element (GstSchedulerChain * chain, GstElement * element)
{
GST_INFO (GST_CAT_SCHEDULING, "adding element \"%s\" to chain %p", GST_ELEMENT_NAME (element),
chain);
/* set the sched pointer for the element */
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
element->sched = GST_SCHEDULER (chain->sched);
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
/* add the element to the list of 'disabled' elements */
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
chain->disabled = g_list_prepend (chain->disabled, element);
chain->num_elements++;
/* notify the scheduler that something changed */
GST_FLAG_SET(chain->sched, GST_BASIC_SCHEDULER_CHANGE);
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
}
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
static gboolean
gst_basic_scheduler_chain_enable_element (GstSchedulerChain * chain,
GstElement * element)
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
{
GST_INFO (GST_CAT_SCHEDULING, "enabling element \"%s\" in chain %p",
GST_ELEMENT_NAME (element), chain);
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
/* remove from disabled list */
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
chain->disabled = g_list_remove (chain->disabled, element);
/* add to elements list */
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
chain->elements = g_list_prepend (chain->elements, element);
/* notify the scheduler that something changed */
GST_FLAG_SET(chain->sched, GST_BASIC_SCHEDULER_CHANGE);
/* GST_FLAG_UNSET(element, GST_ELEMENT_COTHREAD_STOPPING); */
/* reschedule the chain */
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
return gst_basic_scheduler_cothreaded_chain (GST_BIN (GST_SCHEDULER (chain->sched)->parent), chain);
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
}
static void
gst_basic_scheduler_chain_disable_element (GstSchedulerChain * chain,
GstElement * element)
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
{
GST_INFO (GST_CAT_SCHEDULING, "disabling element \"%s\" in chain %p",
GST_ELEMENT_NAME (element), chain);
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
/* remove from elements list */
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
chain->elements = g_list_remove (chain->elements, element);
/* add to disabled list */
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
chain->disabled = g_list_prepend (chain->disabled, element);
/* notify the scheduler that something changed */
GST_FLAG_SET(chain->sched, GST_BASIC_SCHEDULER_CHANGE);
GST_FLAG_SET(element, GST_ELEMENT_COTHREAD_STOPPING);
/* reschedule the chain */
/* FIXME this should be done only if manager state != NULL */
/* gst_basic_scheduler_cothreaded_chain(GST_BIN(chain->sched->parent),chain); */
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
}
static void
gst_basic_scheduler_chain_remove_element (GstSchedulerChain * chain, GstElement * element)
{
GST_INFO (GST_CAT_SCHEDULING, "removing element \"%s\" from chain %p", GST_ELEMENT_NAME (element),
chain);
/* if it's active, deactivate it */
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
if (g_list_find (chain->elements, element)) {
gst_basic_scheduler_chain_disable_element (chain, element);
}
/* we have to check for a threadstate here because a queue doesn't have one */
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
if (GST_ELEMENT_THREADSTATE (element)) {
do_cothread_destroy (GST_ELEMENT_THREADSTATE (element));
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
GST_ELEMENT_THREADSTATE (element) = NULL;
}
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
/* remove the element from the list of elements */
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
chain->disabled = g_list_remove (chain->disabled, element);
chain->num_elements--;
/* notify the scheduler that something changed */
GST_FLAG_SET(chain->sched, GST_BASIC_SCHEDULER_CHANGE);
/* if there are no more elements in the chain, destroy the chain */
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
if (chain->num_elements == 0)
gst_basic_scheduler_chain_destroy (chain);
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
}
static void
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
gst_basic_scheduler_chain_elements (GstBasicScheduler * sched, GstElement * element1, GstElement * element2)
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
{
GList *chains;
GstSchedulerChain *chain;
GstSchedulerChain *chain1 = NULL, *chain2 = NULL;
GstElement *element;
/* first find the chains that hold the two */
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
chains = sched->chains;
while (chains) {
chain = (GstSchedulerChain *) (chains->data);
chains = g_list_next (chains);
if (g_list_find (chain->disabled, element1))
chain1 = chain;
else if (g_list_find (chain->elements, element1))
chain1 = chain;
if (g_list_find (chain->disabled, element2))
chain2 = chain;
else if (g_list_find (chain->elements, element2))
chain2 = chain;
}
/* first check to see if they're in the same chain, we're done if that's the case */
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
if ((chain1 != NULL) && (chain1 == chain2)) {
GST_INFO (GST_CAT_SCHEDULING, "elements are already in the same chain");
return;
}
/* now, if neither element has a chain, create one */
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
if ((chain1 == NULL) && (chain2 == NULL)) {
GST_INFO (GST_CAT_SCHEDULING, "creating new chain to hold two new elements");
chain = gst_basic_scheduler_chain_new (sched);
gst_basic_scheduler_chain_add_element (chain, element1);
gst_basic_scheduler_chain_add_element (chain, element2);
/* FIXME chain changed here */
/* gst_basic_scheduler_cothreaded_chain(chain->sched->parent,chain); */
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
/* otherwise if both have chains already, join them */
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
}
else if ((chain1 != NULL) && (chain2 != NULL)) {
GST_INFO (GST_CAT_SCHEDULING, "merging chain %p into chain %p", chain2, chain1);
/* take the contents of chain2 and merge them into chain1 */
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
chain1->disabled = g_list_concat (chain1->disabled, g_list_copy (chain2->disabled));
chain1->elements = g_list_concat (chain1->elements, g_list_copy (chain2->elements));
chain1->num_elements += chain2->num_elements;
/* FIXME chain changed here */
/* gst_basic_scheduler_cothreaded_chain(chain->sched->parent,chain); */
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
gst_basic_scheduler_chain_destroy (chain2);
/* otherwise one has a chain already, the other doesn't */
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
}
else {
/* pick out which one has the chain, and which doesn't */
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
if (chain1 != NULL)
chain = chain1, element = element2;
else
chain = chain2, element = element1;
GST_INFO (GST_CAT_SCHEDULING, "adding element to existing chain");
gst_basic_scheduler_chain_add_element (chain, element);
/* FIXME chain changed here */
/* gst_basic_scheduler_cothreaded_chain(chain->sched->parent,chain); */
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
}
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
}
/* find the chain within the scheduler that holds the element, if any */
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
static GstSchedulerChain *
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
gst_basic_scheduler_find_chain (GstBasicScheduler * sched, GstElement * element)
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
{
GList *chains;
GstSchedulerChain *chain;
GST_INFO (GST_CAT_SCHEDULING, "searching for element \"%s\" in chains",
GST_ELEMENT_NAME (element));
chains = sched->chains;
while (chains) {
chain = (GstSchedulerChain *) (chains->data);
chains = g_list_next (chains);
if (g_list_find (chain->elements, element))
return chain;
if (g_list_find (chain->disabled, element))
return chain;
}
return NULL;
}
static void
gst_basic_scheduler_chain_recursive_add (GstSchedulerChain * chain, GstElement * element)
{
GList *pads;
GstPad *pad;
GstElement *peerelement;
/* add the element to the chain */
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
gst_basic_scheduler_chain_add_element (chain, element);
GST_DEBUG (GST_CAT_SCHEDULING, "recursing on element \"%s\"", GST_ELEMENT_NAME (element));
/* now go through all the pads and see which peers can be added */
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
pads = element->pads;
while (pads) {
pad = GST_PAD (pads->data);
pads = g_list_next (pads);
GST_DEBUG (GST_CAT_SCHEDULING, "have pad %s:%s, checking for valid peer",
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
GST_DEBUG_PAD_NAME (pad));
/* if the peer exists and could be in the same chain */
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
if (GST_PAD_PEER (pad)) {
GST_DEBUG (GST_CAT_SCHEDULING, "has peer %s:%s", GST_DEBUG_PAD_NAME (GST_PAD_PEER (pad)));
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
peerelement = GST_PAD_PARENT (GST_PAD_PEER (pad));
if (GST_ELEMENT_SCHED (GST_PAD_PARENT (pad)) == GST_ELEMENT_SCHED (peerelement)) {
GST_DEBUG (GST_CAT_SCHEDULING, "peer \"%s\" is valid for same chain",
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
GST_ELEMENT_NAME (peerelement));
/* if it's not already in a chain, add it to this one */
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
if (gst_basic_scheduler_find_chain (chain->sched, peerelement) == NULL) {
gst_basic_scheduler_chain_recursive_add (chain, peerelement);
}
}
}
}
}
/*
* Entry points for this scheduler.
*/
static void
gst_basic_scheduler_setup (GstScheduler *sched)
{
/* first create thread context */
if (GST_BASIC_SCHEDULER_CAST (sched)->context == NULL) {
GST_DEBUG (GST_CAT_SCHEDULING, "initializing cothread context");
GST_BASIC_SCHEDULER_CAST (sched)->context = do_cothread_context_init ();
}
}
static void
gst_basic_scheduler_reset (GstScheduler *sched)
{
cothread_context *ctx;
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
GList *elements = GST_BASIC_SCHEDULER_CAST (sched)->elements;
while (elements) {
GstElement *element = GST_ELEMENT (elements->data);
if (GST_ELEMENT_THREADSTATE (element)) {
do_cothread_destroy (GST_ELEMENT_THREADSTATE (element));
GST_ELEMENT_THREADSTATE (element) = NULL;
}
elements = g_list_next (elements);
}
ctx = GST_BASIC_SCHEDULER_CAST (sched)->context;
do_cothread_context_destroy (ctx);
GST_BASIC_SCHEDULER_CAST (sched)->context = NULL;
}
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
static void
gst_basic_scheduler_add_element (GstScheduler * sched, GstElement * element)
{
GstSchedulerChain *chain;
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
GstBasicScheduler *bsched = GST_BASIC_SCHEDULER (sched);
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
GST_INFO (GST_CAT_SCHEDULING, "adding element \"%s\" to scheduler", GST_ELEMENT_NAME (element));
/* only deal with elements after this point, not bins */
/* exception is made for Bin's that are schedulable, like the autoplugger */
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
if (GST_IS_BIN (element) && !GST_FLAG_IS_SET (element, GST_BIN_SELF_SCHEDULABLE))
return;
/* first add it to the list of elements that are to be scheduled */
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
bsched->elements = g_list_prepend (bsched->elements, element);
bsched->num_elements++;
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
/* create a chain to hold it, and add */
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
chain = gst_basic_scheduler_chain_new (bsched);
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
gst_basic_scheduler_chain_add_element (chain, element);
}
static void
gst_basic_scheduler_remove_element (GstScheduler * sched, GstElement * element)
{
GstSchedulerChain *chain;
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
GstBasicScheduler *bsched = GST_BASIC_SCHEDULER (sched);
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
if (g_list_find (bsched->elements, element)) {
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
GST_INFO (GST_CAT_SCHEDULING, "removing element \"%s\" from scheduler",
GST_ELEMENT_NAME (element));
/* if we are removing the currently scheduled element */
if (bsched->current == element) {
GST_FLAG_SET(element, GST_ELEMENT_COTHREAD_STOPPING);
if (element->post_run_func)
element->post_run_func (element);
bsched->current = NULL;
}
/* find what chain the element is in */
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
chain = gst_basic_scheduler_find_chain (bsched, element);
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
/* remove it from its chain */
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
gst_basic_scheduler_chain_remove_element (chain, element);
/* remove it from the list of elements */
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
bsched->elements = g_list_remove (bsched->elements, element);
bsched->num_elements--;
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
/* unset the scheduler pointer in the element */
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
}
}
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
static GstElementStateReturn
gst_basic_scheduler_state_transition (GstScheduler *sched, GstElement *element, gint transition)
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
{
GstSchedulerChain *chain;
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
GstBasicScheduler *bsched = GST_BASIC_SCHEDULER (sched);
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
/* check if our parent changed state */
if (GST_SCHEDULER_PARENT (sched) == element) {
GST_INFO (GST_CAT_SCHEDULING, "parent \"%s\" changed state", GST_ELEMENT_NAME (element));
if (transition == GST_STATE_PLAYING_TO_PAUSED) {
GST_INFO (GST_CAT_SCHEDULING, "setting scheduler state to stopped");
GST_SCHEDULER_STATE (sched) = GST_SCHEDULER_STATE_STOPPED;
}
else if (transition == GST_STATE_PAUSED_TO_PLAYING) {
GST_INFO (GST_CAT_SCHEDULING, "setting scheduler state to running");
GST_SCHEDULER_STATE (sched) = GST_SCHEDULER_STATE_RUNNING;
}
else {
GST_INFO (GST_CAT_SCHEDULING, "no interesting state change, doing nothing");
}
}
else if (transition == GST_STATE_PLAYING_TO_PAUSED ||
transition == GST_STATE_PAUSED_TO_PLAYING) {
/* find the chain the element is in */
chain = gst_basic_scheduler_find_chain (bsched, element);
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
/* remove it from the chain */
if (chain) {
if (transition == GST_STATE_PLAYING_TO_PAUSED) {
gst_basic_scheduler_chain_disable_element (chain, element);
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
}
else if (transition == GST_STATE_PAUSED_TO_PLAYING) {
if (!gst_basic_scheduler_chain_enable_element (chain, element)) {
GST_INFO (GST_CAT_SCHEDULING,
"could not enable element \"%s\"",
GST_ELEMENT_NAME (element));
return GST_STATE_FAILURE;
}
}
}
else {
GST_INFO (GST_CAT_SCHEDULING,
"element \"%s\" not found in any chain, no state change",
GST_ELEMENT_NAME (element));
}
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
}
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
return GST_STATE_SUCCESS;
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
}
static void
gst_basic_scheduler_lock_element (GstScheduler * sched, GstElement * element)
{
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
if (GST_ELEMENT_THREADSTATE (element))
do_cothread_lock (GST_ELEMENT_THREADSTATE (element));
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
}
static void
gst_basic_scheduler_unlock_element (GstScheduler * sched, GstElement * element)
{
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
if (GST_ELEMENT_THREADSTATE (element))
do_cothread_unlock (GST_ELEMENT_THREADSTATE (element));
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
}
static void
gst_basic_scheduler_yield (GstScheduler *sched, GstElement *element)
{
if (GST_ELEMENT_IS_COTHREAD_STOPPING (element)) {
do_switch_to_main (sched);
/* no need to do a pre_run, the cothread is stopping */
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
}
}
static gboolean
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
gst_basic_scheduler_interrupt (GstScheduler *sched, GstElement *element)
{
GST_FLAG_SET (element, GST_ELEMENT_COTHREAD_STOPPING);
do_switch_to_main (sched);
return FALSE;
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
}
static void
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
gst_basic_scheduler_error (GstScheduler *sched, GstElement *element)
{
GstBasicScheduler *bsched = GST_BASIC_SCHEDULER (sched);
if (GST_ELEMENT_THREADSTATE (element)) {
GstSchedulerChain *chain;
chain = gst_basic_scheduler_find_chain (bsched, element);
if (chain)
gst_basic_scheduler_chain_disable_element (chain, element);
GST_SCHEDULER_STATE (sched) = GST_SCHEDULER_STATE_ERROR;
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
do_switch_to_main (sched);
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
}
}
static void
gst_basic_scheduler_pad_connect (GstScheduler * sched, GstPad *srcpad, GstPad *sinkpad)
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
{
GstElement *srcelement, *sinkelement;
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
GstBasicScheduler *bsched = GST_BASIC_SCHEDULER (sched);
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
srcelement = GST_PAD_PARENT (srcpad);
g_return_if_fail (srcelement != NULL);
sinkelement = GST_PAD_PARENT (sinkpad);
g_return_if_fail (sinkelement != NULL);
GST_INFO (GST_CAT_SCHEDULING, "have pad connected callback on %s:%s to %s:%s",
GST_DEBUG_PAD_NAME (srcpad), GST_DEBUG_PAD_NAME (sinkpad));
GST_DEBUG (GST_CAT_SCHEDULING, "srcpad sched is %p, sinkpad sched is %p",
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
GST_ELEMENT_SCHED (srcelement), GST_ELEMENT_SCHED (sinkelement));
if (GST_ELEMENT_SCHED (srcelement) == GST_ELEMENT_SCHED (sinkelement)) {
GST_INFO (GST_CAT_SCHEDULING, "peer %s:%s is in same scheduler, chaining together",
GST_DEBUG_PAD_NAME (sinkpad));
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
gst_basic_scheduler_chain_elements (bsched, srcelement, sinkelement);
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
}
}
static void
gst_basic_scheduler_pad_disconnect (GstScheduler * sched, GstPad * srcpad, GstPad * sinkpad)
{
GstElement *element1, *element2;
GstSchedulerChain *chain1, *chain2;
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
GstBasicScheduler *bsched = GST_BASIC_SCHEDULER (sched);
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
GST_INFO (GST_CAT_SCHEDULING, "disconnecting pads %s:%s and %s:%s",
GST_DEBUG_PAD_NAME (srcpad), GST_DEBUG_PAD_NAME (sinkpad));
/* we need to have the parent elements of each pad */
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
element1 = GST_ELEMENT_CAST (GST_PAD_PARENT (srcpad));
element2 = GST_ELEMENT_CAST (GST_PAD_PARENT (sinkpad));
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
/* first task is to remove the old chain they belonged to.
* this can be accomplished by taking either of the elements,
* since they are guaranteed to be in the same chain
* FIXME is it potentially better to make an attempt at splitting cleaner??
*/
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
chain1 = gst_basic_scheduler_find_chain (bsched, element1);
chain2 = gst_basic_scheduler_find_chain (bsched, element2);
if (chain1 != chain2) {
/* elements not in the same chain don't need to be separated */
GST_INFO (GST_CAT_SCHEDULING, "elements not in the same chain");
return;
}
if (chain1) {
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
GST_INFO (GST_CAT_SCHEDULING, "destroying chain");
gst_basic_scheduler_chain_destroy (chain1);
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
/* now create a new chain to hold element1 and build it from scratch */
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
chain1 = gst_basic_scheduler_chain_new (bsched);
gst_basic_scheduler_chain_recursive_add (chain1, element1);
}
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
/* check the other element to see if it landed in the newly created chain */
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
if (gst_basic_scheduler_find_chain (bsched, element2) == NULL) {
/* if not in chain, create chain and build from scratch */
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
chain2 = gst_basic_scheduler_chain_new (bsched);
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
gst_basic_scheduler_chain_recursive_add (chain2, element2);
}
}
static GstPad *
gst_basic_scheduler_pad_select (GstScheduler * sched, GList * padlist)
{
GstPad *pad = NULL;
GList *padlist2 = padlist;
GST_INFO (GST_CAT_SCHEDULING, "performing select");
while (padlist2) {
pad = GST_PAD (padlist2->data);
padlist2 = g_list_next (padlist2);
}
/* else there is nothing ready to consume, set up the select functions */
while (padlist) {
pad = GST_PAD (padlist->data);
GST_RPAD_CHAINHANDLER (pad) = GST_DEBUG_FUNCPTR (gst_basic_scheduler_select_proxy);
padlist = g_list_next (padlist);
}
if (pad != NULL) {
GstRealPad *peer = GST_RPAD_PEER (pad);
do_element_switch (GST_PAD_PARENT (peer));
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
pad = GST_ELEMENT (GST_PAD_PARENT (pad))->select_pad;
g_assert (pad != NULL);
}
return pad;
}
static GstClockReturn
gst_basic_scheduler_clock_wait (GstScheduler *sched, GstElement *element,
GstClock *clock, GstClockTime time, GstClockTimeDiff *jitter)
{
GstClockID id;
id = gst_clock_new_single_shot_id (clock, time);
return gst_clock_id_wait (id, jitter);
}
static GstSchedulerState
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
gst_basic_scheduler_iterate (GstScheduler * sched)
{
GstBin *bin = GST_BIN (sched->parent);
GList *chains;
GstSchedulerChain *chain;
GstElement *entry;
GList *elements;
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
gint scheduled = 0;
GstBasicScheduler *bsched = GST_BASIC_SCHEDULER (sched);
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
GST_DEBUG_ENTER ("(\"%s\")", GST_ELEMENT_NAME (bin));
/* clear the changes flag */
GST_FLAG_UNSET(bsched, GST_BASIC_SCHEDULER_CHANGE);
/* step through all the chains */
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
chains = bsched->chains;
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
if (chains == NULL)
return GST_SCHEDULER_STATE_STOPPED;
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
while (chains) {
chain = (GstSchedulerChain *) (chains->data);
chains = g_list_next (chains);
/* all we really have to do is switch to the first child */
/* FIXME this should be lots more intelligent about where to start */
GST_DEBUG (GST_CAT_DATAFLOW, "starting iteration via cothreads using %s scheduler",
_SCHEDULER_NAME);
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
if (chain->elements) {
entry = NULL; /*MattH ADDED?*/
GST_DEBUG (GST_CAT_SCHEDULING, "there are %d elements in this chain", chain->num_elements);
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
elements = chain->elements;
while (elements) {
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
entry = GST_ELEMENT_CAST (elements->data);
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
elements = g_list_next (elements);
if (GST_FLAG_IS_SET (entry, GST_ELEMENT_DECOUPLED)) {
GST_DEBUG (GST_CAT_SCHEDULING, "entry \"%s\" is DECOUPLED, skipping",
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
GST_ELEMENT_NAME (entry));
entry = NULL;
}
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
else if (GST_FLAG_IS_SET (entry, GST_ELEMENT_INFINITE_LOOP)) {
GST_DEBUG (GST_CAT_SCHEDULING, "entry \"%s\" is not valid, skipping",
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
GST_ELEMENT_NAME (entry));
entry = NULL;
}
else
break;
}
if (entry) {
GstSchedulerState state;
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
GST_FLAG_SET (entry, GST_ELEMENT_COTHREAD_STOPPING);
GST_DEBUG (GST_CAT_DATAFLOW, "set COTHREAD_STOPPING flag on \"%s\"(@%p)",
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
GST_ELEMENT_NAME (entry), entry);
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
if (GST_ELEMENT_THREADSTATE (entry)) {
do_switch_from_main (entry);
state = GST_SCHEDULER_STATE (sched);
/* if something changed, return - go on else */
if (GST_FLAG_IS_SET(bsched, GST_BASIC_SCHEDULER_CHANGE) &&
state != GST_SCHEDULER_STATE_ERROR)
return GST_SCHEDULER_STATE_RUNNING;
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
}
else {
GST_DEBUG (GST_CAT_DATAFLOW, "cothread switch not possible, element has no threadstate");
return GST_SCHEDULER_STATE_ERROR;
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
}
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
/* following is a check to see if the chain was interrupted due to a
* top-half state_change(). (i.e., if there's a pending state.)
*
* if it was, return to gstthread.c::gst_thread_main_loop() to
* execute the state change.
*/
GST_DEBUG (GST_CAT_DATAFLOW, "cothread switch ended or interrupted");
if (state != GST_SCHEDULER_STATE_RUNNING) {
GST_INFO (GST_CAT_DATAFLOW, "scheduler is not running, in state %d", state);
return state;
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
}
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
scheduled++;
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
}
else {
GST_INFO (GST_CAT_DATAFLOW, "no entry in this chain, trying the next one");
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
}
}
else {
GST_INFO (GST_CAT_DATAFLOW, "no enabled elements in this chain, trying the next one");
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
}
}
GST_DEBUG (GST_CAT_DATAFLOW, "leaving (%s)", GST_ELEMENT_NAME (bin));
if (scheduled == 0) {
GST_INFO (GST_CAT_DATAFLOW, "nothing was scheduled, return STOPPED");
return GST_SCHEDULER_STATE_STOPPED;
}
else {
GST_INFO (GST_CAT_DATAFLOW, "scheduler still running, return RUNNING");
return GST_SCHEDULER_STATE_RUNNING;
}
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
}
static void
gst_basic_scheduler_show (GstScheduler * sched)
{
GList *chains, *elements;
GstElement *element;
GstSchedulerChain *chain;
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
GstBasicScheduler *bsched = GST_BASIC_SCHEDULER (sched);
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
if (sched == NULL) {
g_print ("scheduler doesn't exist for this element\n");
return;
}
g_return_if_fail (GST_IS_SCHEDULER (sched));
g_print ("SCHEDULER DUMP FOR MANAGING BIN \"%s\"\n", GST_ELEMENT_NAME (sched->parent));
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
g_print ("scheduler has %d elements in it: ", bsched->num_elements);
elements = bsched->elements;
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
while (elements) {
element = GST_ELEMENT (elements->data);
elements = g_list_next (elements);
g_print ("%s, ", GST_ELEMENT_NAME (element));
}
g_print ("\n");
This is an attempt at not segfaulting on errors but reporting some usefull info instead. Original commit message from CVS: This is an attempt at not segfaulting on errors but reporting some usefull info instead. - bin changes so errors can propagate. - changed the _FAST macros to _CAST because that is what they do. - removed all references to cothreads out of the core, they are really a scheduler issue, handler with a sched_private gpointer. - added a live buffer count, for debugging buffer leaks. - added error checking in gst_scheduler_state_transition this solves the "out of cothreads" problem. - GST_ELEMENT_NO_ENTRY == GST_ELEMENT_INFINITE_LOOP - added 2 private element flasg for use by the scheduler (_COTHREAD_STOPPING) is now - added scheduler entry points: - _yield : to create possible scheduling points. - _interrupt: to stop execution of an element. - _error: to signal en error condition to the scheduler. - improved error messages for pads. - signal gst_element_error where appropriate. - added the a new bin to the parent before entering it so one can reference its children. - queue memleak fixes on dispose. - added possible deadlock detection in queue (turned off be default) - GstBasicScheduler is a real class of its own now, hiding its internal variables. - GST_ELEMENT_IS_COTHREAD_STOPPING is gone. either call explicit _yield operations, or make a sane loop. - Better state change handling in filesrc. Better error reporting/recovery too. - updated core plugins. - detect non decoupled elements on scheduler boundries and error.
2001-12-22 21:18:17 +00:00
g_print ("scheduler has %d chains in it\n", bsched->num_chains);
chains = bsched->chains;
- some fixes to int2float making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it bui... Original commit message from CVS: - added playondemand plugin by Leif Morgan Johnson <lmjohns3@eos.ncsu.edu> - some fixes to int2float - aplied a patch from wrobell <wrobell@ite.pl> that is a first attempt at making automake 1.5 happy (gst now requires automake1.5). It's still not perfect but it builds. - Made the schedulers plugable. The default scheduler now lives inside a plugin. - Added a new mpeg1/2 parser/demuxer. - Fixed some compiler warnings in the core libs. - substantial work to GstThread (hopefully less race conditions). simplified the code in GstThread a bit. A state change can now also happen in the thread context. - reworked the state semantics of a bin. it'll now automatically get the highest state of its children. - the autoplugger now nests the threads so that a state change failure of one thread doesn't make its upstream thread lock. - GstQueue refuses to go to PLAYING if the sinkpad is not connected. This way the queue will not wedge in the _get lock. - GstQueue unlocks its mutexes when going to PAUSED. - make sure that when all elements in a bin/thread go to PAUSED, the bin is set to PAUSED too. - make a parent bin wait for its children to PAUSE before ending the iteration with FALSE (EOS) - Some changes to GstPlay to deal with EOS. - aplied the latest patch from Zeenix to gstrtp. end result: GstPlay doesn't crash on EOS and the pipeline is now shut down properly.
2001-12-04 22:12:50 +00:00
while (chains) {
chain = (GstSchedulerChain *) (chains->data);
chains = g_list_next (chains);
g_print ("%p: ", chain);
elements = chain->disabled;
while (elements) {
element = GST_ELEMENT (elements->data);
elements = g_list_next (elements);
g_print ("!%s, ", GST_ELEMENT_NAME (element));
}
elements = chain->elements;
while (elements) {
element = GST_ELEMENT (elements->data);
elements = g_list_next (elements);
g_print ("%s, ", GST_ELEMENT_NAME (element));
}
g_print ("\n");
}
}