| 1 |
OUTDATED |
| 2 |
-------- |
| 3 |
|
| 4 |
|
| 5 |
case 1) |
| 6 |
|
| 7 |
(--------------------------------------------------) |
| 8 |
! bin ! |
| 9 |
! (--------) (--------) (--------) ! |
| 10 |
! ! fakesrc! !identity! !fakesink! ! |
| 11 |
! ! src ----- sink src ---- sink ! ! |
| 12 |
! (--------) (--------) (--------) ! |
| 13 |
(--------------------------------------------------) |
| 14 |
|
| 15 |
.scheduling. |
| 16 |
|
| 17 |
case1 has just one scheduled entity (chain) no problem here. |
| 18 |
|
| 19 |
.eos. |
| 20 |
|
| 21 |
fakesrc detects the end of stream. It just returned the last buffer. |
| 22 |
The next _pull will cause the srcpad to trigger gst_pad_set_eos (). |
| 23 |
After that it will return a NULL buffer. |
| 24 |
|
| 25 |
gst_pad_set_eos() will notify the parent about the plugins attempt to |
| 26 |
signal eos. the parent disables scheduling for the chain containing |
| 27 |
fakesrc. |
| 28 |
|
| 29 |
no more chains are being scheduled so the bin fires EOS. |
| 30 |
|
| 31 |
|
| 32 |
case 2) |
| 33 |
|
| 34 |
|
| 35 |
(-----------------------------------------------------------------------) |
| 36 |
! bin ! |
| 37 |
! (---------------) ! |
| 38 |
! !thread ! ! |
| 39 |
! (--------) (--------) (--------) ! (--------)! ! |
| 40 |
! ! fakesrc! !identity! ! queue ! ! !fakesink!! ! |
| 41 |
! ! src ----- sink src ---- sink src ---- sink !! ! |
| 42 |
! (--------) (--------) (--------) ! (--------)! ! |
| 43 |
! (---------------) ! |
| 44 |
(-----------------------------------------------------------------------) |
| 45 |
|
| 46 |
.scheduling. |
| 47 |
|
| 48 |
case2 has two scheduled entities: fsr-i-q, q-fsk. |
| 49 |
The first one is scheduled by the bin, the second one by the thread. |
| 50 |
|
| 51 |
.eos. |
| 52 |
|
| 53 |
fakesrc detects the end of stream. It just returned the last buffer. |
| 54 |
The next _pull will cause the srcpad to trigger gst_pad_set_eos (). |
| 55 |
After that it will return a NULL buffer. |
| 56 |
|
| 57 |
gst_pad_set_eos() will notify the parent about the plugins attempt to |
| 58 |
signal eos. the parent will take chain1 out if its scheduled chains. |
| 59 |
|
| 60 |
the bin does not fire EOS yet because the thread has not yet signaled |
| 61 |
EOS. |
| 62 |
|
| 63 |
after the queue has sent out the last buffer, its calls eos on its |
| 64 |
src pad. the chain with the queue is removed from the threads scheduled |
| 65 |
chains. there are no more chains to schedule, the thread fires EOS. |
| 66 |
|
| 67 |
The bin now has no more scheduled chains and no more pending subbins, |
| 68 |
it fires EOS. |
| 69 |
|
| 70 |
|
| 71 |
case 3) |
| 72 |
|
| 73 |
(-----------------------------------------------------------------------) |
| 74 |
! bin ! |
| 75 |
! (---------------) ! |
| 76 |
! !thread ! ! |
| 77 |
! (--------) (--------) (--------) ! (--------)! ! |
| 78 |
! ! fakesrc! ! tee ! ! queue1 ! ! !fakesink!! ! |
| 79 |
! ! src ----- sink src ---- sink src ---- sink !! ! |
| 80 |
! (--------) ! ! (--------) ! (--------)! ! |
| 81 |
! ! ! (---------------) ! |
| 82 |
! ! ! ! |
| 83 |
! ! ! (---------------) ! |
| 84 |
! ! ! !thread ! ! |
| 85 |
! ! ! (--------) ! (--------)! ! |
| 86 |
! ! ! ! queue2 ! ! !fakesink!! ! |
| 87 |
! ! src ---- sink src ---- sink !! ! |
| 88 |
! ! ! (--------) ! (--------)! ! |
| 89 |
! (--------) (---------------) ! |
| 90 |
(-----------------------------------------------------------------------) |
| 91 |
|
| 92 |
.scheduling. |
| 93 |
|
| 94 |
case3 has tree scheduled entities: fsr-t(-q1,-q2), q1-fsk. q2-fsk |
| 95 |
the first one is managed by the bin, the second two are managed by the |
| 96 |
threads. |
| 97 |
|
| 98 |
.eos. |
| 99 |
|
| 100 |
fakesrc detects the end of stream. It just returned the last buffer. |
| 101 |
The next _pull will cause the srcpad to trigger gst_pad_set_eos (). |
| 102 |
After that it will return a NULL buffer. |
| 103 |
|
| 104 |
chain1 is taken out of the scheduling algorithm of the bin. |
| 105 |
|
| 106 |
after a while, the threads finish their scheduling and fire EOS too. |
| 107 |
|
| 108 |
The bin now has no more scheduled chains and no more pending subbins, |
| 109 |
it fires EOS. |
| 110 |
|
| 111 |
|
| 112 |
case 4) |
| 113 |
|
| 114 |
(-----------------------------------------------------------------------) |
| 115 |
! bin ! |
| 116 |
! (---------------) ! |
| 117 |
! !thread ! ! |
| 118 |
! (--------) (----------) (--------) ! (--------)! ! |
| 119 |
! ! fakesrc! !mpeg1parse! ! queue1 ! ! !fakesink!! ! |
| 120 |
! ! src -- sink src ---- sink src ---- sink !! ! |
| 121 |
! (--------) ! ! (--------) ! (--------)! ! |
| 122 |
! ! ! (---------------) ! |
| 123 |
! ! ! ! |
| 124 |
! ! ! (---------------) ! |
| 125 |
! ! ! !thread ! ! |
| 126 |
! ! ! (--------) ! (--------)! ! |
| 127 |
! ! ! ! queue2 ! ! !fakesink!! ! |
| 128 |
! ! src ---- sink src ---- sink !! ! |
| 129 |
! ! ! (--------) ! (--------)! ! |
| 130 |
! (----------) (---------------) ! |
| 131 |
(-----------------------------------------------------------------------) |
| 132 |
|
| 133 |
this case differs from case3 in that one of the queues can be empty |
| 134 |
while the other isn't. we assume queue1 is empty while queue2 isn't yet. |
| 135 |
|
| 136 |
.scheduling. |
| 137 |
|
| 138 |
case3 has tree scheduled entities: fsr-t(-q1,-q2), q1-fsk. q2-fsk |
| 139 |
the first one is managed by the bin, the second two are managed by the |
| 140 |
threads. |
| 141 |
|
| 142 |
.eos. |
| 143 |
|
| 144 |
fakesrc detects the end of stream. It just returned the last buffer. |
| 145 |
The next _pull will cause the srcpad to trigger gst_pad_set_eos (). |
| 146 |
After that it will return a NULL buffer. |
| 147 |
|
| 148 |
chain1 is taken out of the scheduling algorithm of the bin. |
| 149 |
|
| 150 |
after a while, the threads finish their scheduling and fire EOS too. |
| 151 |
chain2 (with queue1) is taken out of the scheduling first because it |
| 152 |
is empty. |
| 153 |
|
| 154 |
chain3 (with queue2) finally removes its last scheduled chain and |
| 155 |
fires EOS. |
| 156 |
|
| 157 |
The bin now has no more scheduled chains and no more pending subbins, |
| 158 |
it fires EOS. |
| 159 |
|
| 160 |
|
| 161 |
case 5) |
| 162 |
|
| 163 |
(-----------------------------------------------------------------------) |
| 164 |
! bin ! |
| 165 |
! (---------------) ! |
| 166 |
! !audiosink ! ! |
| 167 |
! (--------) (----------) (--------) ! ! ! |
| 168 |
! !disksrc1! ! effect ! ! mixer ! ! ! ! |
| 169 |
! ! src -- sink src ---- sink src -- sink ! ! |
| 170 |
! (--------) (----------) ! ! ! ! ! |
| 171 |
! ! ! (---------------) ! |
| 172 |
! (--------) (----------) ! ! ! |
| 173 |
! !disksrc2! ! effect ! ! ! ! |
| 174 |
! ! src -- sink src ---- sink ! ! |
| 175 |
! (--------) (----------) (--------) ! |
| 176 |
(-----------------------------------------------------------------------) |
| 177 |
|
| 178 |
This is a simple audio mixer (mp3audio mixer). |
| 179 |
We assume disksrc1 is going to be in EOS while disksrc2 is still in !EOS |
| 180 |
|
| 181 |
|
| 182 |
.scheduling. |
| 183 |
|
| 184 |
two chains are detected: disksrc1-effect-mixer-audiosink, |
| 185 |
disksrc2-effect-mixer-audiosink |
| 186 |
|
| 187 |
.eos. |
| 188 |
|
| 189 |
disksrc1 has passed the last buffer in the _get function. with its next |
| 190 |
call to _get, it will return a NULL buffer and set the src pad to EOS. |
| 191 |
|
| 192 |
the EOS call travels up till the mixer element because it has overridden |
| 193 |
the handler. The first chain is disabled from the scheduler. |
| 194 |
|
| 195 |
finally disksrc2 run into EOS, same procedure as with disksrc1, except |
| 196 |
now the mixer also sets EOS to its src pad, chain2 becomes inactive too |
| 197 |
and the bin fires EOS because there are no more chains to schedule. |
| 198 |
|
| 199 |
somebody sets the location of disksrc1 to a new mp3 file, the srcpad |
| 200 |
is set to !EOS, chain1 is scheduled again, mixer is activated again too |
| 201 |
and audio is playing again... |
| 202 |
|
| 203 |
|
| 204 |
case 6) |
| 205 |
|
| 206 |
(-------------------------------------------------------------------------------------) |
| 207 |
! bin ! |
| 208 |
! (---------------) ! |
| 209 |
! (---------------------------------------------) !audiosink ! ! |
| 210 |
! ! thread1 ! ! ! ! |
| 211 |
! ! (--------) (----------) (-----) ! (--------) ! ! ! |
| 212 |
! ! !disksrc1! ! effect ! !queue! ! ! mixer ! ! ! ! |
| 213 |
! ! ! src -- sink src -- sink src -- sink src -- sink ! ! |
| 214 |
! ! (--------) (----------) (-----) ! ! ! ! ! ! |
| 215 |
! (---------------------------------------------) ! ! ! ! ! |
| 216 |
! ! ! (---------------) ! |
| 217 |
! (---------------------------------------------) ! ! ! |
| 218 |
! ! thread2 ! ! ! ! |
| 219 |
! ! (--------) (----------) (-----) ! ! ! ! |
| 220 |
! ! !disksrc2! ! effect ! !queue! ! ! ! ! |
| 221 |
! ! ! src -- sink src -- sink src -- sink ! ! |
| 222 |
! ! (--------) (----------) (-----) ! (--------) ! |
| 223 |
! (---------------------------------------------) ! |
| 224 |
(-------------------------------------------------------------------------------------) |
| 225 |
|
| 226 |
This is a simple audio mixer (mp3audio mixer). |
| 227 |
We assume disksrc1 is going to be in EOS while disksrc2 is still in !EOS |
| 228 |
|
| 229 |
.scheduling. |
| 230 |
|
| 231 |
Scheduling is a bit tricky here: we have two threads with each one chain ds1-e-q and |
| 232 |
ds2-e-q. the mixer element is loop based and forms a third chain, mixer-audiosink |
| 233 |
|
| 234 |
.eos. |
| 235 |
|
| 236 |
disksrc1 has passed the last buffer in the _get function. with its next |
| 237 |
call to _get, it will return a NULL buffer and set the src pad to EOS. |
| 238 |
|
| 239 |
the EOS call travels up till the queue element because it has overridden |
| 240 |
the handler. The first chain is disabled from the scheduler. thread1 fires EOS |
| 241 |
because it does not have any chains to schedule. |
| 242 |
|
| 243 |
the mixer element pulls buffers from queue1 until queue1 is empty. At that time, |
| 244 |
queue1 sets the pad to EOS, mixer now knows that it does not have to request |
| 245 |
another buffer from that pad again. |
| 246 |
|
| 247 |
disksrc2 finally sets its pad to EOS, rendering chain2 unscheduled which makes |
| 248 |
the thread2 fire an EOS. |
| 249 |
|
| 250 |
the mixer element pulls buffers from queue2 until queue2 is empty. At that time, |
| 251 |
queue2 sets the pad to EOS, mixer now knows that it does not have to request |
| 252 |
another buffer from that pad again. all pads for the mixer are in EOS now so |
| 253 |
it sets the src pad to EOS. the last chain now also becomes unscheduled and |
| 254 |
the bin fires EOS. |
| 255 |
|
| 256 |
|